Amazon HQ Award: Huge Positive for New York City and State

My December 2016 post analyzed the Trump deal to retain Carrier workers in the United States and concluded it was positive for the country and for the state of Indiana. It saved 800 jobs and had a payback to the government of more than 14X their investment. I was clear in the post that I hadn’t voted for Trump and consider myself an independent. While I remain an independent, the opportunity to analyze the recently announced deal to get Amazon to commit 25,000 – 40,000 jobs for New York City is irresistible to me as my conclusions will be in support of politicians on the opposite side of the spectrum from Trump: New York Mayor Bill de Blasio and Governor Andrew Cuomo. I believe that:

Analysis of benefits and drawbacks of any major negotiation should be politically independent.

Unfortunately, this has become less and less the case given the divisive politics that we have in our country. What was shocking to me in this case was that some members of their own party (Democrats), heavily criticized de Blasio and Cuomo.

Major Assumption: Jobs are good for a City/State if the cost to government is reasonable

One of the major responsibilities of a political leader is improving the economy in their State/City. The crux of the discussion is really the question: ‘what is a reasonable cost’ for doing so? On one hand, it can be measured in pure cash flow of moneys paid to Amazon (or any other entity a government wants to attract) versus the cash the government will receive from additional tax dollars. On the other hand, there are other factors that benefit or degrade life in the community. Since the former is more measurable, I’ll start with that.

What are the Actual Out-of-Pocket dollars New York City (NYC) and New York State (NYS) will give to Amazon?

I can’t tell if its rhetoric or a lack of clear communication, but many detractors, like state Senator Michael Gianaris, are saying “We’ve got $3 billion dollars to spend, how would you spend it? Amazon would be very low on the list of where that money would go.” To be blunt, this is a ridiculous comment. NYC is spending zero dollars in cash and while the state is providing $505 million of actual cash as a capital grant, far more money will flow back to it. The Capital grant is based on $2.5 billion that Amazon has promised to invest in New York City (to build their HQ, a 600-seat public school, affordable space for manufacturers and to develop a 3.5-acre waterfront esplanade and park).

The rest of the $3 billion falls into 3 incentive programs that have existed for many years to help woo companies to NYS. They are:

  1. The Excelsior Jobs Program was created in 2010 to replace the expiring Empire Zone Program. Like the prior program, it’s objectives are to provide job creation incentives to firms in targeted industries, like high-tech, for relocating in NYS. The credits are based on the wages added and several other factors. This program, which is available to all companies in the targeted sectors, will generate $1.2 billion in state business income tax credits for Amazon if it meets its commitments.
  2. The Relocation and Employment Assistance Program (REAP), first established in 2003, targets creating jobs in parts of the city more in need of them, rather than adding to the heavy cluster in downtown and midtown Manhattan, namely the outer boroughs or north of 96th street in Manhattan. The tax credits generated from this program total $897 million and can be used towards reducing Amazon’s New York City corporate taxes over 12 years. This credit is based on the rules of the existing law.
  3. The Industrial and Commercial Abatement Program (ICAP), which replaced a prior program, created in 2008, provides tax incentives for commercial and industrial buildings that are built, modernized, or expanded. The credits are based on the taxable value created if the city believes it is beneficial based on its location and other factors. This program is generally available and the $386 million in credits are directly tied to the rules of the law.

Table 1: Benefits to Amazon from NYS and NYC

It is important to note most of the benefits to Amazon are “as of right”, so any company can get them. Since these programs scale based on the number of employees or the amount of capital investment, the sheer size of the Amazon commitment creates a “sticker shock” given the associated benefits. The 3 programs were not created for Amazon but have been in existence for years to encourage job creation and industrial development in targeted areas. The credits under REAP and ICAP appear to be as mandated by those programs and not discretionary. It’s harder for me to tie the state tax abatement amount granted under the Excelsior program (by the state) to the law, but the calculation appears to follow it with some judgement in the cap of what is awarded. The capital grant seems to be the only discretionary part of the package and is the only portion that involves out of pocket dollars from the state (the city will not provide any cash incentives).

Could New York Have won the HQ with lesser incentives?

Given the large return on investment to NYC and NYS, the only question in my mind is whether they could have succeeded with even less incentives and generated an even greater return! A whitepaper by Reis, an analytic company for real estate evaluation, judged New York City as a top candidate without considering incentives offered to Amazon. It’s difficult to judge whether New York City would have been chosen with reduced incentives. On the one hand it has the best public transportation, strong cultural advantages, and several great Universities (as a source of employees), especially the new Cornell-Technion campus located directly next to Amazon’s HQ2 location. On the other hand, it is a very expensive place to do business which is why these incentive programs were created to begin with. As a basis of comparison, consider the bundle of incentives Wisconsin offered to get the Taiwanese technology company Foxconn to build a U.S. plant there. For the 13,000 jobs (at an average annual wage of $53,000) that Foxconn has committed to, Wisconsin plus the County and local village have provided about $3.8 billion in tax credits and breaks. The taxable wages in NY will be 6-9 times as much and the incentives are lower. Therefore, I suspect other locations offered Amazon incentives at the same or a greater level as those from New York.

How Does Revenue to the City and State Compare to the out of pocket cost?

I’m going to make the following assumptions:

  1. New York State Corporate Tax is 6.5% and NYC is 8.85% but I’m assuming the business tax incentives from Excelsior and REAP will be sufficient to preclude Amazon paying any incremental taxes to NYC or NYS (above what they currently pay) for the 12 years they apply. Subsequently, there should be substantial incremental taxes for the additional 8 years of the time horizon I’m using. Since I’m not including this income flow to the city and state, there is considerable upside to my calculations.
  2. The PILOT program payments, estimated by Deputy Mayor Alicia Glen, at $600 to $650 million are the only real estate taxes Amazon will pay. I’m not sure what it would have been without the ICAP credit but the range for the PILOT program amount appears to be known.
  3. NYS and/or NYC will benefit from income, sales and property taxes on employees hired by Amazon and taxes on any additional jobs that get created because of Amazon. I’ll assume taxes are on full wages but that employees have no other income (like interest, capital gains, etc.) and all individuals are single. This puts my model for some who are married without a working spouse at higher taxes then they will pay, but my estimates will be too low for those with a working spouse or any with other sources of income. For this purpose, I’ll use the initial 25,000 jobs plus half of the additional 15,000 (32,500) as the average number over a 20 year period. Since the incremental employment should average longer, that seemed a conservative average to use. I’ll also use an average starting salary of $150,000 for the future Amazon employees as that has been in the announcement. As another assumption, to keep my calculations below what should occur, I haven’t assumed any increases in salary. Even a 4% increase per year would cause salaries to more than double by the end of the 20 years (and NYS and NYC income taxes grow by even more). Since those involved in the project would likely have wage increases over time their income and other taxes would be considerably higher than those based on my assumptions. This coupled with the fact that the negotiators for NYC and NYS used 25 years as the horizon, means their tax calculations will be considerably higher (and more accurate) than mine for the direct employees.

I used the website Smart Asset calculator to generate estimates of NYS and NYC Income tax, sales tax, and property tax per year for each income level. As stated before, the actual numbers will be higher because many of these individuals will have other sources of income, a working spouse and will have salaries escalating over time. Table 2 shows the totals for these estimated taxes to be nearly $15B.

Table 2: NYS & NYC Tax Impact from Amazon HQ

 

  1. Scholars have found strong evidence of the presence of a local multiplier effect. These come from the direct employees hired, indirect jobs created from suppliers and partners and induced jobs that are a result of the spending of the direct and indirect jobs as well as each layer of induced jobs. For example, a noted scholar on the subject, Enrico Moretti, determined that when Apple Computer was employing 12,000 workers locally, an additional 60,000 jobs were created. These included 36,000 unskilled positions like restaurant or retail workers, and 24,000 skilled jobs like lawyers or doctors. If I assume this 5 to 1 ratio would hold for the highly paid Amazon workers, then 32,500 technology jobs would generate 162,500 more jobs in NYC! Based on the Apple example, 60% of these would be unskilled and 40% highly skilled. Assuming an average salary of $35,000 for the unskilled, an average of $100,00 for half of the skilled and $150,000 for the other half, taxes generated from the multiplier effect over the 20 years would be over $28 billion.

Table 3: NYS & NYC Tax Impact from Amazon HQ Multiplier Effect

  1. The $2.5 billion Amazon has committed to spend on capital projects would in turn generate further jobs in construction and an associated multiplier impact, but since this is a temporary benefit over 2-5 years, I have omitted it from the analysis.
  2. The $43 billion estimated total of these income streams to the city and state assume the tax abatements cause no incremental corporate taxes from Amazon. While Amazon will be paying rent on the land leased from the city, I also left out this benefit as I couldn’t estimate the amount. While I believe the actual benefit could be higher, consider that even if I’m off by 75% on the multiplier effect, the total would still be over $22 billion and the payback about 44X the $505 million cash outlay!

Other Benefits and Negatives of Attracting Amazon

There are a variety of more difficult factors to analyze than the straight forward financial windfall the city and state will get from this agreement. Living in the San Francisco Bay Area has taught me that what I may view as obvious might not be so to others. Becoming the Florence of the Tech World has meant that the Bay Area is incredibly wealthy, in turn generating a huge tax base for government to use to fund helping the homeless, stem research and many other perceived public good initiatives. Attracting 25,000 – 40,000 technology jobs will vault New York City into a clear contender for tech community leadership. It will lead to others following and to the creation of more startups, one or two who could become the next Amazon, Apple, Google, Facebook, or Microsoft (generating more jobs and more tax income to NYC and NYS). This is not universally celebrated in the Bay Area as it also has led to traffic congestion, higher housing prices, and increased cost of entertainment. But It has meant increased employment opportunities across the full spectrum of jobs. However, an average worker, while making more than elsewhere, can find it a difficult place to afford. In New York City these issues are partly offset for those renting apartments due to rent control and rent stabilization as over 50% of all rental units are under some form of regulation.

New York City is large enough to be able to absorb 25,000 to 40,000 workers relatively easily, but it could add to the problems for the already strained subway system. I believe it’s no accident that Amazon chose a location near the water so that its employees could take advantage of the new, highly praised, NYC Ferry system. While many of the workers may choose to live near the Amazon facilities, some may decide to buy houses in locations that require utilizing mass transit. If I were to guess, I would say a portion of the increased cash flow, to the city will be used to improve the subway system, Long Island Railroad and to add more Ferries each of which will benefit all New Yorkers.

Conclusion: The Amazon Agreement for HQ2 to be in NYC is a Huge Positive for NYC and NYS

While detractors may nitpick at the deal, it has a great ROI for the City and State, will increase employment, provide revenue to improve mass transit and follows incentives mandated by existing laws. Clearly a coup for Mayor de Blasio and Governor Cuomo.

Soundbytes

  • The SF Chronicle published an article on November 28, 2018 touting Stephan Curry’s strong credentials as a possible MVP this year. In it they used several of the statistics we discussed a year ago. Namely, how much better his teammates shoot when they play with Curry and his amazing plus/minus.
  • Sticking with sports, I can’t help ruminating on how the NFL keeps shooting itself in the foot. I won’t comment on the latest unsavory incidents among players towards women or the Kaepernick fiasco. Instead, I keep thinking about what to call the team about to leave Oakland:
    • Oakland Raiders, their current name
    • Oakland Traders, given their propensity to exchange top players for draft choices
    • Oakland Traitors, trading away their best current players, which has insured a terrible season – thus completing the betrayal of the City of Oakland and the most loyal and colorful fan base in the league

How to Improve Contribution Margin

This post is the third in my series on Key Performance Indicators (KPIs), with a heavy emphasis on contribution margin (CM). Previously, I analyzed why CM is such a strong predictor of success. Given that, companies should consistently look at ways of improving it while still maintaining sufficient growth in their business.

In Azure’s recent full day marketing seminar for our consumer (B2C) focused companies, my session highlighted 6 methods of improving CM:

  1. Increase follow-on sales from existing customers
  2. Raise the average invoice value of the initial and subsequent sales to a customer
  3. Increase GM (Gross Margin) through price increases
  4. Increase GM by reducing cost of goods sold (COGs)
  5. Reduce Blended CAC (cost of customer acquisition) by increasing free or very low cost traffic
  6. Decrease marketing spend as a % of revenue

Before drilling down on each of these I want to define several key terms that will be used throughout the discussion:

  • Contribution Margin = GM – Marketing/Sales Costs – other cost that vary with sales
  • Paid CAC = Market Spend/New Customers acquired through this spend
  • Blended CAC = Market Spend/All new customers
  • CAC Recovery Time (CAC RT) = the number of months until variable profit on a customer equals CAC
  • LTV/CAC = Life Time Value (LTV) of a customer/CAC

I will now review each of these strategies and provide some thoughts on how to activate these in consumer-facing businesses:

1. Increase Follow-On sales from existing customers 

Since existing customers have little or no cost associated with getting them to buy, this will decrease blended CAC, increasing CM.

  • Increasing customer retention through improvements in customer care, more interesting and more targeted emails to a customer, or launching a subscription of one kind or another can all help.

On the first point here is an email I received shortly after subscribing to Harry’s, that I thought did an excellent job at engaging me with their customer support, increasing my likelihood to keep my subscription active:

Hi there,
My name is Katie, and I’m a member of the Harry’s team. I wanted to reach out and say thanks for supporting Harry’s.
You are important to us, and I am here to personally help you however I can to make your Harry’s experience as smooth as possible – both literally and figuratively. Please don’t hesitate to reach out with any thoughts or questions about your Harry’s products or Shave Plan, or just life in general. (And just a reminder that your next box is scheduled to ship on October 27th.) Thanks again for your support, and I hope to speak soon!
All the best,
Katie

On the subscription concept, think about Amazon Prime. How many of you buy more frequently from Amazon because of being a prime member?

  • Add to product portfolio. By giving your customers more options of what to buy (all within the concept of your brand) customers are given the opportunity to spend more often.
  • Make sure your emails are interesting. This will increase the open rate and drive more follow on sales. If all your emails are about discounting your product, then customers will have less interest in opening them and your brand will be devalued. I’ve received emails from numerous sites that say an X% discount is available until a certain date, and then when that date passes, I receive a new offer that is the same or sometimes better.  The most frequently opened emails have headers and content that creates interest beyond whatever products you sell. A/B test different headers and different content. It doesn’t matter how small or large you are or how many emails you send, it always pays to try different variations to increase open rates and conversion. Experiment with different messaging to different customer segments like those who purchased recently, those who “liked” an item, those that have never purchased, etc.
  • Build a Community of your customers. The more you can get customers engaged with you and with each other, the more committed to you they become and the longer they are retained. Think through how you can build an active community among your users through shared photos, videos, chatting, podcasts or events. Most of this should not involve trying to push new purchases but engaging your community to interact with you and each other.

2. Raise the average invoice value of the initial and subsequent sales to a customer

Since shipping costs will not increase proportionately, this will raise GM dollars and therefore CM.

  • Increase pricing. Most startups underprice their product thinking that will increase market adoption. Even some of the largest companies in the world have found there was ample room to increase prices. Thinking differently, Apple upped prices to over $1,000 for an iPhone. And then increased it again to $1,349 for the top of the line product. Five years ago, how many of you thought people would pay over $1,000 for a cell phone? This shows that unless you A/B test different price points you have no idea whether a price increase is the right strategy.
  • Upsell logical add-on products. While trying to get a customer to add to their shopping cart may seem obvious, many companies do not do this on a consistent basis. Some examples of ones that have: a flower company added vases to the offer, a mattress company added pillows and sheets; a subscription razor company added shaving gel; a cell phone company added a case. All of these led to reasonable attach rates of the add-on product and higher average invoice value. Testing what you could add to generate upsell should be a constant process.
  • “Selling” value added services is another form of upsell. This could include things like concierge customer service, service contacts, premier membership with benefits like: invites to special events, early access to new products, reduced shipping cost, preferred discounts on products, etc. If you get your customers to engage in one or more services, you will significantly increase their connection to your product and likely increase retention.

3. Increase Gross Margin through price Increases

Surprisingly, sometimes higher prices position a product as premium (having more value) and generate increased unit sales. Often higher prices generate more revenue even when fewer unit sales result. What may be counter intuitive is that GM$ can increase even if revenue declines. For example, suppose a company has COGs of $50 for a product and is currently pricing it at $100. If a price increase of 20% causes 20% lower unit sales, revenue would decline by 4% while GM$ would increase 12%. Higher gross margin dollars provide more ability to spend on marketing.

 

4. Improving GM by reducing COGs

  • Better Pricing: When your volume increases, ask for better pricing from suppliers. Just as its important to price test regularly, its also important to talk to multiple potential suppliers of your parts/product. An existing supplier may not be eager to voluntarily offer a price discount that goes with increased volume but is more likely to do so if it knows you are checking with others.
  • Changing Packaging: Packaging should be re-examined regularly as improvements may help customer retention. But it also may be possible to lower the cost of the packaging or to change it in a way that lowers shipping costs since that may be based on the size of the box rather than weight.
  • Shipping Costs: Lower shipping cost per $ of revenue (increasing GM and CM) by generating larger orders. In addition to upsell, this can be done by offering better discounts if the order size is larger. One site I have purchased from offers 10% discount if your net spend (after discount) is over $100, 15% if over $150 and 20% if over $200. Getting to the highest discount lowers the price of the product by enough to motivate buyers (including me) to try to buy over $200 in merchandise. The extra revenue creates incremental product margin dollars and decreases shipping cost as a percentage of revenue. This in turn increases GM$.

For a subscription company this can be done by scheduling less frequent (larger) deliveries. The shipping cost of the larger order will be a much smaller percent of revenue, raising GM.

  • Opening a Second distribution center to reduce shipping cost. Orders shipped from a west coast distribution center to an east coast customer will have 5 zone pricing. By having a second distribution center in a place like Columbus, Ohio (a frequently used location) those same orders will usually be 1 zone, sometimes 2 zone pricing, resulting in substantial savings per order. The caveat here is that a company needs enough volume for the total savings on orders to exceed the fixed cost of a second distribution center.

5. Improving CM by driving “free” or “nearly free” traffic

The higher the proportion of free or inexpensive traffic to total traffic, the lower the blended CAC.

  • Improving SEO (search engine optimization). I’ve learned from SEO experts that optimizing SEO is not free, but rather very low cost compared to paid traffic. Our previous post walks through some of the science involved in making improvements. I would suggest using an SEO consultant as it is likely to lead to far better results.
  • Convert a visitor not ready to buy to an email recipient. If you do that than you will have subsequent opportunities to market to her or him. A slightly costlier version of this is to use remarketing to woo visitors who came to your site but didn’t buy. While using remarketing (advertising) has a cost, it is usually much lower CAC than other methods.
  • Produce emails that get forwarded and go viral. Such emails need to motivate recipients to forward them due to being very funny, of human interest, etc. While there is typically a product offering embedded in them, the header emphasizes the reason to read it. One Azure portfolio company, Shinesty, recently had an email that was opened by about 7X the number of people it was initially sent to.  That generated a lot of potential customers without spending extra marketing dollars. Engaging emails has enabled Shinesty to maintain high CM and high growth.
  • Use social networking to generate incremental customers. Having the right posts on a social network like Instagram can lead to new potential customers finding out about you and lead to additional sales.
  • Optimize Customer Retention. Or as my good friend Chris Bruzzo (CMO of EA) spoke about at the Azure Marketing conference: “Love the ones you’re with.” Existing customers are usually the largest source of “free” buyers in a period. The longer you retain a customer, the more repeat buyers you have, increasing contribution margin. So, it’s imperative to take great care of your existing customers.
  • Drive PR. Like SEO, there is some cost involved in this but if you are judicious in any agency spend and thoughtful in creating news worthy press releases this can be a great source of traffic at a modest cost. However, I recommend you try to understand what you are getting from PR because I have seen situations where the spend did not produce meaningful results.

6. Decrease Marketing Spend as a % of Revenue.

The CAC Recovery Time plays a major role in how to manage your market spend to balance growth and burn. For example, if CAC Recovery Time is one month, spending more will not drive up burn appreciably. If it takes more than a year to recover your CAC, moderating market spend is critical to achieving a reasonable CM. If you recoup CAC faster, you can invest more quickly in the next round of customers. In the consumer space, I won’t invest in a company that has a long (a year or more) CAC Recovery Time as customers are likely to churn in an average of 2-3 years, making it difficult to achieve a reasonable business model. For B2B company’s customer longevity tends to be much longer, and the LTV/CAC can be 5X or more even if CAC Recovery Time is a year.

When a company decreases its market spend as a % of revenue it may experience lower growth but better CM. However, many companies have waste in their marketing spend so it’s important to measure the efficacy of each area of spend separately and to eliminate programs with a low return. This will allow you to reduce the spend with minimal impact on growth rates. There is a balance needed to try to optimize the relationship between CM and revenue growth as higher burn requires raising money more frequently and can put your company at risk. On the other hand, a company generating $1M in revenue needs to be growing at 100% or more to warrant most VCs to consider investing. Since CM should improve with scale, spending more on marketing may be a viable strategy for early stage companies. Once a company reaches $10M in revenue, annual growth of 50% will get it to $76M in revenue in 5 years so such a company should consider better CM rather than driving much higher growth rates and continuing to burn excessive cash.

In summary, Contribution Margin is the lifeblood of a company. If it is weak, the company is likely to fail over time. If it is strong and revenue growth is high, success seems likely. Improving CM is an ongoing process. I realize many of you probably feel much of what I’ve said is obvious, but my question is:“How many of these suggestions are you already doing on a regular basis?”

While you may be using several of the suggestions in this post, I encourage you to try more and to also double down where you can on the ones you already are trying. The results will make your company more valuable!

 

SoundBytes

  • I just want to remind readers that my collaborator on my blog posts, Andrea Drager, doesn’t typically take a bow for her significant contributions. Also, in this post, Chris Bruzzo added several improvements that have been incorporated. So many thanks to Andrea and Chris.
  • Can’t help but comment on the start to the NBA season. Not surprisingly, the Warriors are off to a great start with Curry and Durant leading the way. Greene and Thompson now have moved close to their usual contribution so I’m hopeful that the team can keep up its current pace.
  • What surprised me early on was the lack of recognition that both Toronto and San Antonio would be greatly improved. Remember, while San Antonio lost Kawhi, he only played a few games last year so with the addition of DeRozan should improve and once again reach the playoffs. For Toronto the change to Kawhi is a marked improvement placing them very competitive with the Celtics for eastern leadership.
  • I also feel it necessary to comment on the “Las Vegas” Raiders. I call them that already as they have shown zero regard for Oakland fans. While commentators have criticized their trading of all-star level players for draft choices, this is precisely on-strategy. When they get to Vegas they want a brand-new set of rising stars that the new fan base can identify with (using the numerous first round draft choices they traded for), and they don’t mind having the worst record in the league while still in Oakland. I believe Oakland fans should stop attending games as a response. I also think the NFL continues to shoot itself in the foot, allowing one of the most loyal and visible fan bases in the league to once again be abandoned

Why Contribution Margin is a Strong Predictor of Success for Companies

In the last post I concluded with a brief discussion of Contribution Margin as a key KPI. Recall:

Contribution Margin = Variable Profits – Sales and Marketing Cost

The higher the contribution margin, the more dollars available towards covering G&A. Once contribution margin exceeds G&A, a company reaches operating profits. For simplicity in this post, I’ll use gross margin (GM) as the definition of variable profits even though there may be other costs that vary directly with revenue.

The Drivers of Contribution Margin (CM)

There is an absolute correlation between GM percent and CM. Very high gross margin companies will, in general, get to strong contribution margins and low gross margin companies will struggle to get there. But the sales and marketing needed to drive growth is just as important. There are several underlying factors in how much needs to be spent on sales and marketing to drive growth:

  1. The profits on a new customer relative to the cost of acquiring her (or him). That is, the CAC (customer acquisition cost) for customers derived from paid advertising compared to the profits on those customers’ first purchase
  2. The portion of new traffic that is “free” from SEO (search engine optimization), PR, existing customers recommending your products, etc.
  3. The portion of revenue that comes from repeat customers

The Relationship Between CAC and First Purchase Profits Has a Dramatic Impact on CM

Suppose Company A spends $60 to acquire a customer and has GM of $90 on the initial purchase by that customer. The contribution margin will already be positive $30 without accounting for customers that are organic or those that are repeat customers; in other words, this tends to be extremely positive! Of course, the startups I see in eCommerce are rarely in this situation but those that are can get to profitability fairly quickly if this relationship holds as they scale.

It would be more typical for companies to find that the initial purchase GM only covers a portion of CAC but that subsequent purchases lead to a positive relationship between the LTV (life time value) of the customer and CAC. If I assume the spend to acquire a customer is $60 and the GM is $30 then the CM on the first purchase would be negative (-$30), and it would take a second purchase with the same GM dollars to cover that initial cost. Most startups require several purchases before recovering CAC which in turn means requiring investment dollars to cover the outlay.

Free Traffic and Contribution Margin

If a company can generate a high proportion of free/organic traffic, there is a benefit to contribution margin. CAC is defined as the marketing spend divided by the number of new customers derived from this spend. Blended CAC is defined as the marketing spend divided by all customers who purchased in the period. The more organically generated and return customers, the lower the “blended CAC”. Using the above example, suppose 50% of the new customers for Company A come from organic (free) traffic. Then the “blended CAC“ would be 50% of the paid CAC. In the above example that would be $30 instead of $60 and if the GM was only $30 the initial purchase would cover blended CAC.

Of course, in addition to obtaining customers for free from organic traffic, companies, as they build their customer base, have an increasing opportunity to obtain free traffic by getting existing customers to buy again. So, a company should never forget that maintaining a persistent relationship with customers leads to improved Contribution Margin.

Spending to Drive Higher Growth Can Mean Lower Contribution Margin

Unless the GM on the first purchase a new customer makes exceeds their CAC, there is an inverse relationship between expanding growth and achieving high contribution margin. Think of it this way: suppose that going into a month the likely organic traffic and repeat buyers are somewhat set. Boosting that month’s growth means increasing the number of new paid customers, which in turn makes paid customers a higher proportion of blended CAC and therefore increases CAC. For an example consider the following assumptions for Company B:

  • The GM is $60 on an average order of $100
  • Paid CAC is $150
  • The company will have 1,000 new customers through organic means and 2,000 repeat buyers or $300,000 in revenue with 60% GM ($180,000) from these customers before spending on paid customers
  • G&A besides marketing for the month will be $150,000
  • Last year Company B had $400,000 in revenue in the same month
  • The company is considering the ramifications of targeting 25%, 50% or 100% year-over-year growth

Table 1: The Relationship Between Contribution Margin & Growth

Since the paid CAC is $150 while Gross Margin is only $60 per new customer, each acquired customer generates negative $90 in contribution margin in the period. As can be seen in Table 1, the company would shrink 25% if there is no acquisition spend but would have $180,000 in contribution margin and positive operating profit. On the other end of the spectrum, driving 100% growth requires spending $750,000 to acquire 5,000 new customers and results in a negative $270,000 in contribution margin and an Operating Loss of $420,000 in the period. Of course, if new customers are expected to make multiple future purchases than the number of repeat customers would rise in future periods.

Subscription Models Create More Consistency but are not a Panacea

When a company’s customers are monthly subscribers, each month starts with the prior month’s base less churn. To put it another way, if churn from the prior month is modest (for example 5%) then that month already has 95% of the prior months revenue from repeat customers. Additionally, if the company increases the average invoice value from these customers, it might even have a starting point where return customers account for as much revenue as the prior month. For B-to-B companies, high revenue retention is the norm, where an average customer will pay them for 10 years or more.

Consumer ecommerce subscriptions typically have much more substantial churn, with an average life of two years being closer to the norm. Additionally, the highest level of churn (which can be as much as 30% or more) occurs in the second month, and the next highest, the third month before tapering off. What this means is that companies trying to drive high sequential growth will have a higher % churn rate than those that target more modest growth. Part of a company’s acquisition spend is needed just to stay even. For example, if we assume all new customers come from paid acquisition, the CAC is $200, and that 15% of 10,000 customers churn then the first $300,000 in marketing spend would just serve to replace the churned customers and additional spend would be needed to drive sequential growth.

Investing in Companies with High Contribution Margin

As a VC, I tend to appreciate strong business models and like to invest after some baseline proof points are in place.  In my last post I outlined a number of metrics that were important ways to track a company’s health with the ratio of LTV (life time value) to CAC being one of the most important. When a company has a high contribution margin they have the time to build that ratio by adding more products or establishing subscriptions without burning through a lot of capital. Further, companies that have a high LTV/CAC ratio should have a high contribution margin as they mature since this usually means customers buy many times – leading to an expansion in repeat business as part of each month’s total revenue.

This thought process also applies to public companies. One of the most extreme is Facebook, which I’ve owned and recommended for five years. Even after the recent pullback its stock price is about 7x what it was five years ago (or has appreciated at a compound rate of nearly 50% per year since I’ve been recommending it). Not a surprise as Facebook’s contribution margin runs over 70% and revenue was up year/year 42% in Q2. These are extraordinary numbers for a company its size.

To give the reader some idea of how this method can be used as one screen for public companies, Table 2 shows gross margin, contribution margin, revenue growth and this year’s stock market performance for seven public companies.

Table 2: Public Company Contribution Margin Analysis

Two of the seven companies shown stand out as having both high Contribution Margin and strong revenue growth: Etsy and Stitch Fix. Each had year/year revenue growth of around 30% in Q2 coupled with 44% and 29% contribution margins, respectively. This likely has been a factor in Stitch Fix stock appreciating 53% and Etsy 135% since the beginning of the year.

Three of the seven have weak models and are struggling to balance revenue growth and contribution margin: Blue Apron, Overstock, and Groupon. Both Blue Apron and Groupon have been attempting to reduce their losses by dropping their marketing spend. While this increased their CM by 10% and 20% respectively, it also meant that they both have negative growth while still losing money. The losses for Blue Apron were over 16% of revenue. This coupled with shrinking revenue feels like a lethal combination. Blue Apron stock is only down a marginal amount year-to-date but is 59% lower than one year ago. Groupon, because of much higher gross margins than Blue Apron (52% vs 35%), still seems to have a chance to turn things around, but does have a lot of work to do. Overstock went in the other direction, increasing marketing spend to drive modest revenue growth of 12%. But this led to a negative CM and substantially increased losses. That strategy did not seem to benefit shareholders as the stock has declined 53% since the beginning of the year.

eBay is a healthy company from a contribution margin point of view but has sub 10% revenue growth. I can’t tell if increasing their market spend by a substantial amount (at the cost of lower CM) would be a better balance for them.

For me, Spotify is the one anomaly in the table as its stock has appreciated 46% since the IPO despite weak contribution margins which was one reason for my negative view expressed in a prior post. I think that is driven by three reasons: its product is an iconic brand; there is not a lot of float in the stock creating some scarcity; and contribution margin has been improving giving bulls on the stock a belief that it can get to profitability eventually. I say it is an anomaly, as comparing it to Facebook, it is hard to justify the relative valuations. Facebook grew 42% in Q2, Spotify 26%; Facebook is trading at a P/E of 24 whereas even if we assume Spotify can eventually get to generating 6% net profit (it currently is at a 7% loss before finance charges and 31% loss after finance charges, so this feels optimistic) Spotify would be trading at 112 times this theoretic future earnings.

 

SoundBytes

I found the recent controversy over Elon Musk’s sharing his thoughts on taking Tesla private interesting. On the one hand, people want transparency from companies and Elon certainly provides that! On the other hand, it clearly impacted the stock price for a few days and the SEC abhors anything that can be construed as stock manipulation. Of course, Elon may not have been as careful as he should have been when he sent out his tweet regarding whether financing was lined up…but like most entrepreneurs he was optimistic.

Interesting KPIs (Key Performance Indicators) for a Subscription Company

what-are-key-performance-indicators-kpis

In working with early stage businesses, I often get the question as to what metrics should management and the board use to help understand a company’s progress. It is important for every company to establish a set of consistent KPIs that are used to objectively track progress. While these need to be a part of each board package, it is even more important for the executive team to utilize this for managing their company. While this post focuses on SaaS/Subscription companies, the majority of it applies to most other types of businesses.

Areas KPIs Should Cover

  1. P&L Trends
  2. MRR (Monthly Recurring Revenue) and LTR (Lifetime Revenue)
  3. CAC (Cost of Customer Acquisition)
    1. Marketing to create leads
    2. Customers acquired electronically
    3. Customers acquired using sales professionals
  4. Gross Margin and LTV (Life Time Value of a customer)
  5. Marketing Efficiency

Many companies will also need KPIs regarding inventory in addition to the ones above.

While there may be very complex analysis behind some of these numbers, it’s important to try to keep KPIs to 2-5 pages of a board package. Use of the right KPIs will give a solid, objective, consistent top-down view of the company’s progress. The P&L portion of the package is obviously critical, but I have a possibly unique view on how this should be included in the body of a board package.

P&L Trends: Less is More

One mistake many companies make is confusing detail with better analysis. I often see models that have 50-100 line items for expenses and show this by month for 3 or more years out… but show one or no years of history. What this does is waste a great deal of time on predicting things that are inconsequential and controllable (by month), while eliminating all perspective. Things like seasonality are lost if one is unable to view 3 years of revenue at a time without scrolling from page to page. Of course, for the current year’s budget it is appropriate for management to establish monthly expectations in detail, but for any long-term planning, success revolves around revenue, gross margins, marketing/sales spend and the number of employees. For some companies that are deep technology players there may be significant costs in R&D other than payroll, but this is the exception. By using a simple formula for G&A based on the number of employees, the board can apply a sanity check on whether cost estimates in the long-term model will be on target assuming revenue is on target. So why spend excessive time on nits? Aggregating cost frees up time for better understanding how and why revenue will ramp, the relationship between revenue types and gross margin, the cost of acquiring a customer, the lifetime value of a customer and the average spend per employee.

In a similar way, the board is well served by viewing a simple P&L by quarter for 2 prior years plus the current one (with a forecast of remaining quarters). The lines could be:

Table1: P&L by Quarter

A second version of the P&L should be produced for budget comparison purposes. It should have the same rows but have the columns be current period actual, current period budget, year to date (YTD) actual, year to date budget, current full year forecast, budget for the full year.

Table 2: P&L Actual / Budget Comparison

Tracking MRR and LTR

For any SaaS/Subscription company (I’ll simply refer to this as SaaS going forward) MRR growth is the lifeblood of the company with two caveats: excessive churn makes MRR less valuable and excessive cost in growing MRR also leads to deceptive prosperity. More about that further on. MRR should be viewed on a rolling basis. It can be done by quarter for the board but by month for the management team. Doing it by quarter for the board enables seeing a 3-year trend on one page and gives the board sufficient perspective for oversight. Management needs to track this monthly to better manage the business. A relatively simple set of KPIs for each of 12 quarterly periods would be:

Table 3: MRR and Retention

Calculating Life Time Revenue through Cohort Analysis

The detailed method of calculating LTR does not need to be shown in every board package but should be included at least once per year, but calculated monthly for management.

The LTR calculation uses a grid where the columns would be the various Quarterly cohorts, that is all customers that first purchased that quarter (management might also do this using monthly instead of quarterly). This analysis can be applied to non-SaaS companies as well as SaaS entities. The first row would be the number of customers in the cohort. The next row would be the first month’s revenue for the cohort, the next the second months revenue, and so on until reaching 36 months (or whatever number the board prefers for B2B…I prefer 60 months). The next row would be the total for the full period and the final row would be the average Lifetime Revenue, LTR, per member of the cohort.

Table 4: Customer Lifetime Revenue

A second table would replicate the grid but show average per member of the cohort for each month (row). That table allows comparisons of cohorts to see if the average revenue of a newer cohort is getting better or worse than older ones for month 2 or month 6 or month 36, etc.

Table 5: Average Revenue per Cohort

Cohorts that have a full 36 months of data need to be at least 36 months old. What this means is that more recent cohorts will not have a full set of information but still can be used to see what trends have occurred. For example, is the second months average revenue for a current cohort much less than it was for a cohort one year ago? While newer cohorts do not have full sets of monthly revenue data, they still are very relevant in calculating more recent LTR. This can be done by using average monthly declines in sequential months and applying them to cohorts with fewer months of data.

Customer Acquisition Cost (CAC)

Calculating CAC is done in a variety of ways and is quite different for customers acquired electronically versus those obtained by a sales force.  Many companies I’ve seen have a combination of the two.

Marketing used to generate leads should always be considered part of CAC. The marketing cost in a month first is divided by the number of leads to generate a cost/lead. The next step is to estimate the conversion rate of leads to customers. A simple table would be as follows:

Table 6: Customer Acquisition Costs

table 6.1

For an eCommerce company, the additional cost to convert might be one free month of product or a heavily subsidized price for the first month. If the customer is getting the item before becoming a regular paying customer than the CAC would be:

CAC = MCTC / the percent that converts from the promotional trial to a paying customer.

CAC when a Sales Force is Involved

For many eCommerce companies and B2B companies that sell electronically, marketing is the primary cost involved in acquiring a paying customer. For those utilizing a sales force, the marketing expense plus the sales expense must be accumulated to determine CAC.

Typically, what this means is steps 1 through 3 above would still be used to determine CPL, but step 1 above might include marketing personnel used to generate leads plus external marketing spend:

  1. CPL (cost per lead) as above
  2. Sales Cost = current month’s cost of the sales force including T&E
  3. New Customers in the month = NC
  4. Conversion Rate to Customer = NC/number of leads= Y%
  5. CAC = CPL/Y% + (Sales Cost)/NC

There are many nuances ignored in the simple method shown. For example, some leads may take many months to close. Some may go through a pilot before closing. Therefore, there are more sophisticated methods of calculating CAC but using this method would begin the process of understanding an important indicator of efficiency of customer acquisition.

Gross Margin (GM) is a Critical Part of the Equation

While revenue is obviously an important measure of success, not all revenue is the same. Revenue that generates 90% gross margin is a lot more valuable per dollar than revenue that generates 15% gross margin. When measuring a company’s potential for future success it’s important to understand what level of revenue is required to reach profitability. A first step is understanding how gross margin may evolve. When a business scales there are many opportunities to improve margins:

  • Larger volumes may lead to larger discounts from suppliers
  • Larger volumes for products that are software/content may lower the hosting cost as a percent of revenue
  • Shipping to a larger number of customers may allow opening additional distribution centers (DCs) to facilitate serving customers from a DC closer to their location lowering shipping cost
  • Larger volumes may mean improved efficiency in the warehouse. For example, it may make more automation cost effective

When forecasting gross margin, it is important to be cautious in predicting some of these savings. The board should question radical changes in GM in the forecast. Certain efficiencies should be seen in a quarterly trend, and a marked improvement from the trend needs to be justified. The more significant jump in GM from a second DC can be calculated by looking at the change in shipping rates for customers that will be serviced from the new DC vs what rates are for these customers from the existing one.

Calculating LTV (Lifetime Value)

Gross Margin, by itself may be off as a measure of variable profits of a customer. If payment is by credit card, then the credit card cost per customer is part of variable costs. Some companies do not include shipping charges as part of cost of goods, but they should always be part of variable cost. Customer service cost is typically another cost that rises in proportion to the number of customers. So:

Variable cost = Cost of Goods sold plus any cost that varies directly with sales

Variable Profit = Revenue – Variable Cost

Variable Profit% (VP%) = (Variable Profit)/Revenue

LTV = LTR x VP%

The calculation of VP% should be based on current numbers as they will apply going forward. Determining a company’s marketing efficiency requires comparing LTV to the cost of customer acquisition. As mentioned earlier in the post, if the CAC is too large a proportion of LTV, a company may be showing deceptive (profitless) growth. So, the next set of KPIs address marketing efficiency.

Marketing Efficiency

It does not make sense to invest in an inefficient company as they will burn through capital at a rapid rate and will find it difficult to become profitable. A key measure of efficiency is the relationship between LTV and CAC or LTV/CAC. Essentially this is how many dollars of variable profit the company will make for every dollar it spends on marketing and sales. A ratio of 5 or more usually means the company is efficient. The period used for calculating LTR will influence this number. Since churn tends to be much lower for B2B companies, 5 years is often used to calculate LTR and LTV. But, using 5 years means waiting longer to receive resulting profits and can obscure cash flow implications of slower recovery of CAC. So, a second metric important to understand burn is how long it takes to recover CAC:

CAC Recovery Time = number of months until variable profit equals the CAC

The longer the CAC recovery time, the more capital required to finance growth. Of course, existing customers are also contributing to the month’s revenue alongside new customers. So, another interesting KPI is contribution margin which measures the current state of balance between marketing/sales and Variable Profits:

Contribution Margin = Variable Profits – Sales and Marketing Cost

Early on this number will be negative as there aren’t enough older customers to cover the investment in new ones. But eventually the contribution margin in a month needs to turn positive. To reach profitability it needs to exceed all other costs of the business (G&A, R&D, etc.). By reducing a month’s marketing cost, a company can improve contribution margin that month at the expense of sequential growth… which is why this is a balancing act.

I realize this post is long but wanted to include a substantial portion of KPIs in one post. However, I’ll leave more detailed measurement of sales force productivity and deeper analysis of several of the KPIs discussed here for one or more future posts.

Soundbytes

I’ll begin by apologizing for a midyear brag, but I always tell others to enjoy success and therefore am about to do that myself. In my top ten predictions for 2018 I included a market prediction and 4 stock predictions. I was feeling pretty good that they were all working well when I started to create this post. However, the stock prices for high growth stocks can experience serious shifts in very short periods. Facebook and Tesla both had (what I consider) minor shortfalls against expectations in the 10 days since and have subsequently declined quite a bit in that period. But given the strength of my other two recommendations, Amazon and Stitchfix, the four still have an average gain of 15% as of July 27. Since I’ve only felt comfortable predicting the market when it was easy (after 9/11 and after the 2008 mortgage blowup), I was nervous about predicting the S&P would be up this year as it was a closer call and was somewhat controversial given the length of the bull market prior to this year. But it seemed obvious that the new tax law would be very positive for corporate earnings. So, I thought the S&P would be up despite the likelihood of rising interest rates. So far, it is ahead 4.4% year to date driven by stronger earnings. Since I always fear that my record of annual wins can’t continue I wanted to take a midyear victory lap just in case everything collapses in the second half of the year (which I don’t expect but always fear). So I continue to hold all 4 stocks and in fact bought a bit more Facebook today.

Highlights From the 2018 Azure CEO Summit

It’s All About the Network

On June 13th, 2018, Azure held our 12th Annual CEO Summit, hosted at the Citrix Templeton Conference Center. Success for our companies is typically predicated on the breadth and depth of their networks in Silicon Valley and beyond. This event is a cornerstone of how we support this, providing a highly curated, facilitated opportunity to expand connections for business development, fund-raising, and strategic partner dialogue. It is also an opportunity for our portfolio companies to develop strong relationships with our investors, networks, and among each other, which provides business partnership opportunities, potential future investors and is a first step towards engaging with future acquirers. An incidental benefit to Azure is that the appeal of the event also leads to expansion of our own network.

Throughout the day, we had participation of nearly 70 corporate entities, venture funds and financial institutions, including Amazon, Google, Apple, P&G, Citrix, Ericsson, Intel, Microsoft, Oracle, Trinet, Arcserv, Citibank, SVB, and UBS, in addition to 28 of Azure’s portfolio companies, and six Canadian startups which were invited as part of Azure’s Canada-Bridge initiative. The Canadian companies were selected from a group of about 100 nominated by Canadian VCs. At the event, the six winners gained access to Azure’s Silicon Valley network not only through participation along with our portfolio CEOs in the approximately 370 one-on-one meetings we arranged but also through networking opportunities throughout the rest of the day and into the evening.

Nearly all the Azure portfolio companies participating gave demo-day style presentations to the full audience, which expanded the reach of their message beyond the more intimate one-on-one meetings.

Visionary Keynote Speakers

Azure was quite fortunate in once again having several visionary keynote speakers who provided inspiration and thought-provoking inputs from their experiences as highly successful entrepreneurs and investors.

The first was David Ko, currently President and COO, Rally Health, and formerly SVP, Yahoo and COO, Zynga (famous for Farmville which peaked at 34.5 million daily active users). David provided his vision for the consumer-focused future for managing health and shared lessons learned from his journeys both in taking Zynga public and in leading Rally Health as it has grown in eight years from a company with low single-digit millions in revenue to more than a billion in revenue. Rally works with more than 200,000 employers to help drive employee engagement in their health. Accessible to more than 35 million people, Rally’s digital platform and solutions help people adopt healthier lifestyles, select health benefits, and choose the best doctor at the right price for their needs. The company’s wellness solution focuses on four key areas to improve health: nutrition, exercise, stress reduction and preventive health. Given the astronomical increase in the portion of U.S. GDP spent on healthcare, David pointed out how critical it is to help individuals improve their “wellness” tactics. He believes this is one of the waves of the future to curb further acceleration of healthcare cost.

Shai Agassi, Former President, Product and Technology Group, SAP, and former CEO, Better Place responded to questions posed by me and the audience during a fireside chat.  Shai first shared his experience of building a business that successfully became integrated into SAP, but the heart of his session revolved around his perspectives on the evolution of the electric car and the future emergence of (safe) automated vehicles. He painted a vivid picture of what the oncoming transition to a new generation of vehicles means for the future, where automated, electric cars will become the norm (in 5-10 years). As a result, he believes people will reduce their use of their own cars and instead, use an “automated Uber-like service” for much of their transportation. In such a world, many people won’t own a car and for those that do, their autos will have much longer useful lives thereby reducing the need to replace cars with the same frequency. If he proves correct, this would clearly have major ramifications for auto manufacturers and the oil industry.

Our final keynote speaker was Ron Suber, President Emeritus, Prosper Marketplace, who is referred to as “The Godfather of Fintech”.  Ron shared with us his perspective that we’re at the beginning stages of the ‘Golden Age of Fintech’ which he believes will be a 20-year cycle. He expects to continue to see a migration to digital, accessible platforms driven by innovation by existing players and new entrants to the market that will disrupt the incumbents. What must be scary to incumbents is that the new entrants in fintech include tech behemoths like Paypal, Google, Amazon, Tencent (owner of WeChat), Facebook and Apple.  While traditional banks may have access to several hundred million customers, these players can leverage their existing reach into relationships with billions of potential customers. For example,  WeChat and Instagram have both recently surpassed one billion users. With digital/mobile purchasing continuing to gain market share, a player like Apple can nearly force its users to include Apple Pay as one of their apps giving Apple some unique competitive advantages. Amazon and WeChat (in China) are in a strong position to leverage their user bases.

All That Plus a Great Dinner

After an action packed daytime agenda, the Summit concluded with a casual cocktail hour and outdoor dinner in Atherton. Most attendees joined, and additional members of the Azure network were invited as well. The dinner enabled significant networking to continue and provided an additional forum for some who were not able to be at the daytime event to meet some of our portfolio executives.

The Bottom Line – It’s About Results

How do we measure the success of the Summit? We consider it successful if several of our companies garner potential investors, strike business development deals, etc.  As I write this, only nine days after the event, we already know of a number of investment follow-ups, more than ten business-development deals being discussed, and multiple debt financing conversations. Investment banks and corporate players have increased awareness of the quality of numerous companies who presented. Needless to say, Azure is pleased with the bottom line.

Company Valuations Implied by my Valuations Bible: Are Snap, Netflix, Square and Twitter Grossly Overvalued?

Applying the Gross Margin Multiple Method to Public Company Valuation

In my last two posts I’ve laid out a method to value companies not yet at their mature business models. The method provides a way to value unprofitable growth companies and those that are profitable but not yet at what could be their mature business model. This often occurs when a company is heavily investing in growth at the expense of near-term profits. In the last post, I showed how I would estimate what I believed the long-term model would be for Tesla, calling the result “Potential Earnings” or “PE”. Since this method requires multiple assumptions, some of which might not find agreement among investors, I provided a second, simplified method that only involved gross margin and revenue growth.

The first step was taking about 20 public companies and calculating how they were valued as a multiple of gross margin (GM) dollars. The second step was to determine a “least square line” and formula based on revenue growth and the gross margin multiple for these companies. The coefficient of 0.62 shows that there is a good correlation between Gross Margin and Revenue Growth, and one significantly better than the one between Revenue Growth and a company’s Revenue Multiple (that had a coefficient of 0.36 which is considered very modest).

Where’s the Beef?

The least square formula derived in my post for relating revenue growth to an implied multiple of Gross Margin dollars is:

GM Multiple = (24.773 x Revenue growth percent) + 4.1083

Implied Company Market Value = GM Multiple x GM Dollars

Now comes the controversial part. I am going to apply this formula to 10 companies using their data (with small adjustments) and compare the Implied Market Value (Implied MKT Cap) to their existing market Cap as of several days ago. I’ll than calculate the Implied Over (under) Valuation based on the comparison. If the two values are within 20% I view it as normal statistical variation.

Table 1: Valuation Analysis of 10 Tech Companies

  • * Includes net cash included in expected market cap
  • ** Uses adjusted GM%
  • *** Uses 1/31/18 year end
  • **** Growth rate used in the model is q4 2017 vs q4 2016.  See text

This method suggests that 5 companies are over-valued by 100% or more and a fifth, Workday, by 25%. Since Workday is close to a normal variation, I won’t discuss it further. I have added net cash for Facebook, Snap, Workday and Twitter to the implied market cap as it was material in each case but did not do so for the six others as the impact was not as material.

I decided to include the four companies I recommended, in this year’s top ten list, Amazon, Facebook, Tesla and Stitchfix, in the analysis. To my relief, they all show as under-valued with Stitchfix, (the only one below the Jan 2 price) having an implied valuation more than 100% above where it currently trades. The other three are up year to date, and while trading below what is suggested by this method, are within a normal range. For additional discussion of these four see our 2018 top Ten List.

 

Digging into the “Overvalued” Five

Why is there such a large discrepancy between actual market cap and that implied by this method for 5 companies? There are three possibilities:

  1. The method is inaccurate
  2. The method is a valid screen but I’m missing some adjustment for these companies
  3. The companies are over-valued and at some point, will adjust, making them risky investments

While the method is a good screen on valuation, it can be off for any given company for three reasons:  the revenue growth rate I’m using will radically change; a particular company has an ability to dramatically increase gross margins, and/or a particular company can generate much higher profit margins than their gross margin suggests. Each of these may be reflected in the company’s actual valuation but isn’t captured by this method.

To help understand what might make the stock attractive to an advocate, I’ll go into a lot of detail in analyzing Snap. Since similar arguments apply to the other 4, I’ll go into less detail for each but still point out what is implicit in their valuations.

Snap

Snap’s gross margin (GM) is well below its peers and hurts its potential profitability and implied valuation. Last year, GM was about 15%, excluding depreciation and amortization, but it was much higher in the seasonally strong Q4. It’s most direct competitor, Facebook, has a gross margin of 87%.  The difference is that Facebook monetizes its users at a much higher level and has invested billions of dollars and executed quite well in creating its own low-cost infrastructure, while Snap has outsourced its backend to cloud providers Google and Amazon. Snap has recently signed 5-year contracts with each of them to extend the relationships. Committing to lengthy contracts will likely lower the cost of goods sold.  Additionally, increasing revenue per user should also improve GM.  But, continuing to outsource puts a cap on how high margins can reach. Using our model, Snap would need 79% gross margin to justify its current valuation. If I assume that scale and the longer-term contracts will enable Snap to double its gross margins to 30%, the model still shows it as being over-valued by 128% (as opposed to the 276% shown in our table). The other reason bulls on Snap may justify its high valuation is that they expect it to continue to grow revenue at 100% or more in 2018 and beyond. What is built into most forecasts is an assumed decline in revenue growth rates over time… as that is what typically occurs. The model shows that growing revenue 100% a year for two more years without burning cash would leave it only 32% over-valued in 2 years. But as a company scales, keeping revenue growth at that high a level is a daunting task. In fact, Snap already saw revenue growth decline to 75% in Q4 of 2017.

Twitter

Twitter is not profitable.  Revenue declined in 2017 after growing a modest 15% in 2016, and yet it trades at a valuation that implies that it is a growth company of about 50%. While it has achieved such levels in the past, it may be difficult to even get back to 15% growth in the future given increased competition for advertising.

Netflix

I recommended Netflix in January 2015 as one of my stock picks for the year, and it proved a strong recommendation as the stock went up about 140% that year. However, between January 2015 and January 2018, the stock was up over 550% while trailing revenue only increased 112%.  I continue to like the fundamentals of Netflix, but my GM model indicates that the stock may have gotten ahead of itself by a fair amount, and it is unlikely to dramatically increase revenue growth rates from last year’s 32%.

Square

Square has followed what I believe to be the average pattern of revenue growth rate decline as it went from 49% growth in 2015, down to 35% growth in 2016, to under 30% growth in 2017. There is no reason to think this will radically change, but the stock is trading as if its revenue is expected to grow at a nearly 90% rate. On the GM side, Square has been improving GM each year and advocates will point out that it could go higher than the 38% it was in 2017. But, even if I use 45% for GM, assuming it can reach that, the model still implies it is 90% over-valued.

Blue Apron

I don’t want to beat up on a struggling Blue Apron and thought it might have reached its nadir, but the model still implies it is considerably over-valued. One problem that the company is facing is that investors are negative when a company has slow growth and keeps losing money. Such companies find it difficult to raise additional capital. So, before running out of cash, Blue Apron began cutting expenses to try to reach profitability. Unfortunately, given their customer churn, cutting marketing spend resulted in shrinking revenue in each sequential quarter of 2017. In Q4 the burn was down to $30 million but the company was now at a 13% decline in revenue versus Q4 of 2016 (which is what we used in our model). I assume the solution probably needs to be a sale of the company. There could be buyers who would like to acquire the customer base, supplier relationships and Blue Apron’s understanding of process. But given that it has very thin technology, considerable churn and strong competition, I’m not sure if a buyer would be willing to pay a substantial premium to its market cap.

 

An Alternative Theory on the Over Valued Five

I have to emphasize that I am no longer a Wall Street analyst and don’t have detailed knowledge of the companies discussed in this post, so I easily could be missing some important factors that drive their valuation.  However, if the GM multiple model is an accurate way of determining valuation, then why are they trading at such lofty premiums to implied value? One very noticeable common characteristic of all 5 companies in question is that they are well known brands used by millions (or even tens of millions) of people. Years ago, one of the most successful fund managers ever wrote a book where he told readers to rely on their judgement of what products they thought were great in deciding what stocks to own. I believe there is some large subset of personal and professional investors who do exactly that. So, the stories go:

  • “The younger generation is using Snap instead of Facebook and my son or daughter loves it”
  • “I use Twitter every day and really depend on it”
  • “Netflix is my go-to provider for video content and I’m even thinking of getting rid of my cable subscription”

Once investors substitute such inclinations for hard analysis, valuations can vary widely from those suggested by analytics. I’m not saying that such thoughts can’t prove correct, but I believe that investors need to be very wary of relying on such intuition in the face of evidence that contradicts it.

The Valuation Bible – Part 2: Applying the Rules to Tesla and Creating an Adjusted Valuation Method for Startups

This post is part 2 of our valuation discussion (see this post for part 1).  As I write this post Tesla’s market cap is about $56 billion. I thought it would be interesting to show how the rules discussed in the first post apply to Tesla, and then to take it a step further for startups.

Revenue and Revenue Growth

Revenue for Tesla in 2017 was $11.8 billion, about 68% higher than 2016, and it is likely to grow faster this year given the over $20 billion in pre-orders (and growing) for the model 3 coupled with continued strong demand for the model S and model X. Since it is unclear when the new sports car or truck will ship, I assume no revenue in those categories. As long as Tesla can increase production at the pace they expect, I estimate 2018 revenue will be up 80% – 120% over 2017, with Q4 year over year growth at or above 120%.

If I’m correct on Tesla revenue growth, its 2018 revenue will exceed $20 billion. So, Rule Number 1 from the prior post indicates that Tesla’s high growth rates should merit a higher “theoretical PE” than the S&P (by at least 4X if one believes that growth will continue at elevated rates).

Calculating TPE

Tesla gross margins have varied a bit while ramping production for each new model, but in the 16 quarters from Q1, 2014 to Q4, 2017 gross margin averaged 23% and was above 25%, 6 of the 16 quarters. Given that Tesla is still a relatively young company it appears likely margins will increase with scale, leading me to believe that long term gross margins are very likely to be above 25%. While it will dip during the early production ramp of the model 3, 25% seems like the lowest percent to use for long term modeling and I expect it to rise to between 27% and 30% with higher production volumes and newer factory technology.

Tesla recognizes substantial cost based on stock-based compensation (which partly occurs due to the steep rise in the stock). Most professional investors ignore artificial expenses like stock-based compensation, as I will for modeling purposes, and refer to the actual cost as net SG&A and net R&D. Given that Tesla does not pay commissions and has increased its sales footprint substantially in advance of the roll-out of the model 3, I believe Net SG&A and Net R&D will each increase at a much slower pace than revenue. If they each rise 20% by Q4 of this year and revenue is at or exceeds $20 billion, this would put their total at below 20% of revenue by Q4. Since they should decline further as a percent of revenue as the company matures, I am assuming 27% gross margin and 18% operating cost as the base case for long term operating profit. While this gross margin level is well above traditional auto manufacturers, it seems in line as Tesla does not have independent dealerships (who buy vehicles at a discount) and does not discount its cars at the end of each model year.

Estimated TPE

Table 1 provides the above as the base case for long term operating profit. To provide perspective on the Tesla opportunity, Table 1 also shows a low-end case (25% GM and 20% operating cost) and a high-end profit case (30% GM and 16% operating cost).  Recall, theoretic earnings are derived from applying the mature operating profit level to trailing and to forward revenue. For calculating theoretic earnings, I will ignore interest payments and net tax loss carry forwards as they appear to be a wash over the next 5 years. Finally, to derive the Theoretic Net Earnings Percent a potential mature tax rate needs to be applied. I am using 20% for each model case which gives little credit for tax optimization techniques that could be deployed. That would make theoretic earnings for 2017 and 2018 $0.85 billion and $1.51 billion, respectively and leads to:

  • 2017 TPE=$ 56.1 billion/$0.85 = 66.0
  • 2018 TPE= $ 56.1 billion/1.51 = 37.1

The S&P trailing P/E is 25.5 and forward P/E is about 19X. Based on our analysis of the correlation between growth and P/E provided in the prior post, Tesla should be trading at a minimum of 4X the trailing S&P ratio (or 102 TP/E) and at least 3.5X S&P forward P/E (or 66.5 TP/E). To me that shows that the current valuation of Tesla does not appear out of market.  If the market stays at current P/E levels and Tesla reaches $21B in revenue in 2018 this indicates that there is strong upside for the stock.

Table 1: Tesla TPE 2017 & 2018

The question is whether Tesla can continue to grow revenue at high rates for several years. Currently Tesla has about 2.4% share of the luxury car market giving it ample room to grow that share. At the same time, it is entering the much larger medium-priced market with the launch of the Model 3 and expects to produce vehicles in other categories over the next few years. Worldwide sales of new cars for the auto market is about 90 million in 2017 and growing about 5% a year. Tesla is the leader in several forward trends: electric vehicles, automated vehicles and technology within a car. Plus, it has a superior business model as well. If it reaches $21 billion in revenue in 2018, its share of the worldwide market would be about 0.3%. It appears poised to continue to gain share over the next 3-5 years, especially as it fills out its line of product.  Given that it has achieved a 2.4% share of the market it currently plays in, one could speculate that it could get to a similar share in other categories. Even achieving a 1% share of the worldwide market in 5 years would mean about 40% compound growth between 2018 and 2022 and imply a 75X-90X TP/E at the end of this year.

The Bear Case

I would be remiss if I omitted the risks that those negative on the stock point out. Tesla is a very controversial stock for a variety of reasons:

  1. Gross Margin has been volatile as it adds new production facilities so ‘Bears’ argue that even my 25% low case is optimistic, especially as tax rebate subsidies go away
  2. It has consistently lost money so some say it will never reach the mature case I have outlined
  3. As others produce better electric cars Tesla’s market share of electric vehicles will decline so high revenue growth is not sustainable
  4. Companies like Google have better automated technology that they will license to other manufacturers leading to a leap frog of Tesla

As they say, “beauty is in the eyes of the beholder” and I believe my base case is realistic…but not without risk. In response to the bear case that Tesla revenue growth can’t continue, it is important to recognize that Tesla already has the backlog and order momentum to drive very high growth for the next two years. Past that, growing market share over the 4 subsequent years to 1% (a fraction of their current share of the luxury market) would generate compound annual growth of 40% for that 4-year period. In my opinion, the biggest risk is Tesla’s own execution in ramping production. Bears will also argue that Tesla will never reach the operating margins of my base case for a variety of reasons. This is the weakness of the TPE approach: it depends on assumptions that have yet to be proven. I’m comfortable when my assumptions depend on momentum that is already there, gross margin proof points and likelihood that scale will drive operating margin improvements without any radical change to the business model.

Applying the rules to Startups

As a VC I am often in the position of helping advise companies regarding valuation. This occurs when they are negotiating a round of financing or in an M&A situation.  Because the companies are even earlier than Tesla, theoretic earnings are a bit more difficult to establish. Some investors ignore the growth rates of companies and look for comps in the same business. The problem with the comparable approach is that by selecting companies in the same business, the comps are often very slow growth companies that do not merit a high multiple. For example, comparing Tesla to GM or Ford to me seems a bit ludicrous when Tesla’s revenue grew 68% last year and is expected to grow even faster this year while Ford and GM are growing their revenue at rates below 5%. It would be similar if investors compared Apple (in the early days of the iPhone) to Nokia, a company it was obsoleting.

Investors look for proxies to use that best correlate to what future earnings will be and often settle on a multiple of revenue. As Table 2 shows, there is a correlation between valuation as a multiple of revenue and revenue growth regardless of what industry the companies are in. This correlation is closer than one would find by comparing high growth companies to their older industry peers.

Table 2: Multiple of Revenue and Revenue Growth

However, using revenue as the proxy for future earnings suffers from a wide variety of issues. Some companies have 90% or greater gross margins like our portfolio company Education.com, while others have very low gross margins of 10% – 20%, like Spotify. It is very likely that the former will generate much higher earnings as a percent of revenue than the latter. In fact, Education.com is already cash flow positive at a relatively modest revenue level (in the low double-digit millions) while Spotify continues to lose a considerable amount of money at billions of dollars in revenue. Notice, this method also implies that Tesla should be valued about 60% higher than its current market price.

This leads me to believe a better proxy for earnings is gross margin as it is more closely correlated with earnings levels. It also removes the issue of how revenue is recognized and is much easier to analyze than TPE. For example, Uber recognizing gross revenue or net revenue has no impact on gross margin dollars but would radically change its price to revenue. Table 3 uses the same companies as Table 2 but shows their multiple of gross margin dollars relative to revenue growth. Looking at the two graphs, one can see how much more closely this correlates to the valuation of public companies. The correlation coefficient improves from 0.36 for the revenue multiple to 0.62 for the gross margin multiple.

Table 3: Multiple of Gross Margin vs. Revenue Growth

So, when evaluating a round of financing for a pre-profit company the gross margin multiple as it relates to growth should be considered. For example, while there are many other factors to consider, the formula implies that a 40% revenue growth company should have a valuation of about 14X trailing gross margin dollars.  Typically, I would expect that an earlier stage company’s mature gross margin percent would likely increase. But they also should receive some discount from this analysis as its risk profile is higher than the public companies shown here.

Notice that the price to sales graph indicates Tesla should be selling at 60% more than its multiple of 5X revenue. On the other hand, our low-end case for Tesla Gross Margin, 25%, puts Tesla at 20X Gross Margin dollars, just slightly undervalued based on where the least square line in Table 3 indicates it should be valued.