Transparency and Customer-Based Corporate Valuation
Welcome to Marketing BS, where I share a weekly article dismantling a little piece of the Marketing-Industrial Complex — and sometimes I offer simple ideas that actually work.
If you enjoy this article, I invite you to subscribe to Marketing BS — the weekly newsletters feature bonus content, including follow-ups from the previous week, commentary on topical marketing news, and information about unlisted career opportunities.
Thanks for reading and keep it simple,
The Devil’s in the Details
Earlier this month, Google shared YouTube revenue numbers for the very first time. As always, Ben Thompson offered an insightful perspective on Stratechery (subscription required):
The reason, by the way, why I am pretty sure that Google made this move on their own is that this is the least possible amount of disclosure possible. We got YouTube’s top-line revenue number, and that’s it. No word on cost-of-revenue (a particularly interesting number in the case of YouTube given its surely massive bandwidth costs), how much it pays out to creators, how much it spends on content moderation, basically nothing that allows outsiders to understand the business other than that top line item. I strongly suspect that Google is trying to foreclose the SEC revisiting the issue.
Every analyst on the planet would love to see enough of Google’s data to understand how their “other” businesses are really doing. But let’s imagine that Alphabet actually shared detailed information about cost-of-revenue, creator payments, and cost of content moderation. I expect that we would still struggle to develop a model that could (accurately) predict what YouTube’s revenue and costs might be a year from now. Plus, we would have an even harder time making a confident estimation of YouTube’s “true” value.
During WeWork’s last private round of funding, the company's valuation hovered around $47 billion. Before the August 2019 announcement of WeWork’s intention to go public, many investment banks believed that the company would fetch an even high price. By September, WeWork executives downgraded their expectation to $20 billion. Projected valuations quickly dropped below $8 billion, forcing WeWork to withdraw the IPO and nudging Softbank to bail the company out.
Casper — one of the major online mattress companies — is the latest unicorn to face sharp criticisms for their IPO filing. Over the last year, we have seen similar concerns about companies like Uber, Lyft, Peloton, Slack, and many more. Startup culture can lead to high valuations from the private markets, but what happens when the IPO process requires companies to provide detailed financial information? In many cases, companies’ current valuations — as well as their future prospects — are damaged.
Historically, companies that launched an IPO would experience a “bump” in their valuation, because previously locked-out investors were excited about the opportunity to purchase shares. In recent years, the situation has changed, due to a few factors:
Many investors (at least sophisticated ones with deep pockets) have had opportunities to invest in late-stage private companies. Thus, there is less pent up demand.
At the same time, illiquidity still exists in private markets for the existing shareholders. When a company goes public, there are not only more opportunities for people to buy stocks, but to SELL them.
The IPO process requires specific disclosures of business metrics.
You would think that private investors possess access to all the information they might need to value a business, but that is often not true. As such, the process of going public shines a bright light on many companies, providing more liquidity to both buyers and sellers. The result? We approach a “consensus” valuation based on how the business is really doing.
Although the transparency of public companies is better than private firms, it is still a long way from “fully transparent.” Forecasting the distant future is never easy. Where will Uber be in 20 years — extremely profitable or completely irrelevant? The answer is impossible for us to predict with any accuracy. But we should be able to review a company’s financial documents and project an outlook — with relative confidence — for the next few years. Unfortunately, most companies don’t even share the information required for that level of evaluation.
In the absence of comprehensive financial documents, what other options can we use to evaluate a company? Here’s one method: “customer-based valuation.” When I attended the Wharton School 15+ years ago, I studied under Professor Peter Fader (who I quoted two weeks ago during a post on advertising adjacency). Fader, along with Professor Bruce Hardie from the London School of Economics, has developed a model for predicting future customer behaviour — with far more accuracy than traditional methods. Here is a simplified explanation of their ideas about customer-based valuation:
Assume that customers remain with a firm for some time, until they permanently end their relationships with the firm (i.e., they “die” — and that likelihood to “die” is governed by a statistical distribution, e.g., exponential).
While customers are “alive,” assume that they will make purchases with some frequency governed by another statistical distribution (e.g., Poisson).
Assume that when customers make purchases, the spend associated with those purchases fluctuates around some average spend level, governed by a third statistical distribution (e.g., gamma).
Assume that (1) the propensity of customers to remain with the firm, (2) the propensity of customers to make purchases while they are alive, and (3) the average spend associated with each customer’s purchases can vary, possibly significantly, from customer to customer (e.g., some customers may be highly prone to churning but are frequent buyers while they are “alive,” spending small amounts each time they buy; others may be highly loyal light buyers who tend to spend a lot whenever they buy, etc.).
Assume that these propensities, while varying between customers, stay constant for any given customer over time. Note this does not mean that customer behavior does not change — but it means how that behavior changes can be modeled assuming a constant propensity to churn, purchase and spend.
Multiply all three distributions together to get one “master” equation which predicts future purchases for every customer in the database (it will not be accurate for any given customer, but the key is to be accurate in aggregate).
Each of the equations have “constants” or “parameters.” You can run simulations in Excel to find the best fit for those parameters, and you can assume the parameters will be unchanged going forward.
The model includes a lot of assumptions, but you should be nodding your head in agreement with — at the very least — points one through four. The only “leap” in logic you might need to make is believing that an individual’s parameters don’t change over time. The rest is just math.
That math can, in theory, use any number of different distribution curves to see what fits best. In practice, Fader and Hardie have found that the same curves end up being the winners time and again, even across radically different industries.
One big question: does this model work? Is the assumption that “people don’t change” true enough to predict future sales?
When the model is tested against hold-out data, it turns out to be shockingly accurate — far more accurate than any competing models it has been run against. I have not encountered any better way to calculate “customer lifetime value.”
If you know what it costs to serve your existing customers AND you know how much they are going to purchase in the future, then you are a long way toward understanding the value of your existing business. To close the loop, you just need to identify (1) how much it costs you to acquire new customers, and (2) how many new customers you are likely to acquire in the future. You can use Fader and Hardie’s model, plus assumptions about customer acquisition, to more accurately forecast your “baseline” future revenues. In turn, that information can be used to drive the baseline future P&L of your entire business (more on that below). By employing this model, you can also consider possible scenarios like, “will competitive pressures change?” or “how are my labor costs going to change?” or “what if I try radically new marketing channels?”
Show me the Data
Last year, I wrote a post about Starbucks Rewards. The conclusion was clear: although retail loyalty programs tend to destroy value, they help companies collect data about individual customers. Before the internet, loyalty programs offered one of the few ways to track a customer's purchase history over time.
Today’s corporate landscape, though, is obviously different. With the widespread availability of data collection and management systems, tracking your customers over time should be child’s play — especially for the many direct-to-consumer businesses in the marketplace. Any company could plug their data into Fader and Hardie’s model and gain a clear understanding of the future value of all the customers they have already acquired.
Of course, if you have ever worked for a company, you know that strategies like these are not where managers apply their energy — never mind the idea of managers being transparent enough to share information like this with outside experts. Instead, analysts are left to fill in the blanks with guesses, evaluating companies on metrics like “revenue/EBITDA multiple.” For good measure, we make eye-balled adjustments based on historical growth rate, future management projections, or “same store sale growth.”
There is a better way.
Fader and Professor Daniel McCarthy from Emory University have created a framework they call “customer-based corporate valuation.” CBCV uses standard discounted-cash-flow models with a notable difference: instead of relying on future financial predictions built off of trend lines, CBCV uses projections that are built ground up from individual customer purchase and spending behavior.
Customer-level retention, purchasing, and spending forecasts are developed by using the types of behavioral models mentioned earlier, drawing on whatever customer data may be available — anything from aggregated disclosures in SEC filings to detailed transaction logs and CRM databases. The end result? A MUCH more accurate prediction of future revenue and expenses, which can then be “valued” using traditional methods.
In a LinkedIn post, McCarthy shared his vision for CBCV’s possibilities (if you are even remotely interested in marketing effectiveness, I recommend reading the entire thing):
CBCV is a framework that augments (instead of replaces) any traditional corporate valuation method by forecasting key financial line items (most notably revenue) within that method off period-by-period customer-base activity projections. The underlying model that we use to characterize customer behavior may actually be the same as the one driving the CE calculation, but instead of using it to get CLVs that are then summed up, we insert it into your favorite traditional corporate valuation model to make that valuation model better (i.e., to make the financial projections more accurate).
For an elegant and accessible expansion of these ideas, check out McCarthy and Fader’s article in Harvard Business Review. For an even more detailed look at the methodology, you can consult this academic article co-written by McCarthy and Fernando Pereda.
According to corporate mythology, marketers used to spend their time telling stories and recommending colors for a new company logo. I’m not sure those descriptions were ever really true, but there’s no question that today’s marketers have been forced to expand their skill set far beyond the art of words and pictures. Companies expect the marketing team to include data-driven analysts focused on short-feedback customer acquisition channels. We know that marketing is more than just paid acquisition on Google and Facebook, but the lack of data in the “other” part of the function has lowered traditional marketing’s status to the point that experience with the softer side of marketing has become a “nice to have” in most CMO roles.
McCarthy believes that widespread adoption of CBCV — by investors, executives, and everyone else involved with valuation — will change the current expectations: “CMOs will get a lot more powerful but also a lot more accountable. CMOs will start sounding like CFOs and CFOs will [start] sounding more like CMOs.”
And McCarthy has put his money where his mouth is. Along with Fader, he founded Theta Equity — a research firm that provides CBCV “as a service” to financial firms (e.g., private equity, hedge funds, and late-stage VC firms), helping them better understand the valuation of particular companies. Theta also works with private companies, which could share some of the analysis publicly — especially before big IPOs. Check out Theta’s pre-IPO take on Slack(valuation), Lyft (valuation), and the retention and unit economics analysis of Blue Apron.
If you are a marketer who has already moved from the soft side to the analytical side, this type of analysis is the next step in your journey. Deriving the math is not easy, but applying the math is actually not that hard. Grasping the concepts gets easier when you broaden your perspectives about the way to think about company valuation and customer lifetime value. And as more professionals become comfortable with this type of analysis, we may find more companies sharing these metrics publicly. In that case, all investors will have a better idea what is happening under the hood.
Keep it simple,
If you enjoyed this article, I invite you to subscribe to Marketing BS — the weekly newsletters feature bonus content, including follow-ups from the previous week, commentary on topical marketing news, and information about unlisted career opportunities.
Edward Nevraumont is a Senior Advisor with Warburg Pincus. The former CMO of General Assembly and A Place for Mom, Edward previously worked at Expedia and McKinsey & Company. For more information, including details about his latest book, check out Marketing BS.