Ode #4 — The Shape of the Bet

For my first major research assignment at Sanford Bernstein, my boss asked me to determine the household penetration of personal computers in the United States. All of the stocks we covered --- Best Buy, Circuit City, CompUSA --- were flying high on PC sales. The big question: how long could it last?

Our firm had relationships with the two top technology consulting firms of the era, and I called them each up. One thought PC penetration was 10-15%. The other thought it was over 50%. Neither had any real data to support their position --- they were just guessing. So I did what any young nerdy kid would do: I conducted my own research. I hired a professional survey company to call thousands of households across the country, properly stratified for demographic diversity and statistical significance, and in the end I had my number. If memory serves it was 27%. We called the resulting report A PC in Every Pot. A few weeks later, I was quoted in The Economist.

This is how I fell in love with data.

From that point on, data became my competitive edge --- and my obsession. When I ran Bumble and bumble, I invented a new KPI to measure salon sales performance and convinced the owner of the most popular salon management software to build it into his platform. When I got to DE Shaw, I watched first-party data explode with the rise of ecommerce, and saw that almost no consumer companies knew what to do with any of it. When I left to start Hairstory, I decided data would be our competitive advantage from day one. Ten years later, I'm confident we know more about our business than any other beauty brand I've encountered.

There is only one problem with being a data obsessive who wants to start a new business: there is no data yet to analyze.

So what do you do?

You make it up. We nerds call it modeling.

My grandmother once tasted an angel food cake I'd made her from a box mix and said, deadpan in her heavy German accent: "I can understand why someone would buy this once. But who would ever buy it again?"

She didn't know she was describing the central question of every business model ever built. But she was.

When I first started building the Hairstory model, I was sitting on the rubberized floor of my local community center gym, my back against the wooden bleachers, laptop open, watching my son attempt one-armed free throws for the rec department basketball program. Not easy for a kid who had a stroke in utero --- but he figured it out, which probably should have told me something about how the business would go.

Building a model forces you to think through every element of a business. If you do it right, it pressure-tests your key assumptions and tells you what you need to believe for the business to succeed. Most importantly, it tells you how much money you will need to invest before you reach cash flow breakeven --- the point at which the business can sustain itself --- and how long that will take.

The output is financial, but this is fundamentally an operating model: a granular, working representation of how the business will function before it exists. The trick is to build it with as few assumptions as possible. Every assumption is a place where you can be wrong and the more assumptions you have the greater the permutations you need to test. Fewer assumptions, better insights.

Let me start where my grandmother did: with the question of who comes back.

In the world of ecommerce, new customer acquisition has become relatively measurable. You spend money on marketing, you get customers. The two metrics that define that exchange are CAC (customer acquisition cost --- how much you spent to earn each new customer) and AOV (average order value --- how much they spent when they arrived). The relationship between them is called ROAS --- return on ad spend --- and it tells you the efficiency of your marketing. And these metrics apply to pretty much all businesses, not just ecommerce.

But ROAS only tells half the story. The more important question is what happens next.

At Hairstory, our core product works so well that once people try it, they come back. Not quite forever, but almost. Of first-time buyers, 40-50% return for a second purchase. Of those, \~75% come back a third time. And once someone has bought three times, they stay with only minor attrition. Every new customer we acquire will generate, on average, hundreds of dollars into the future. That changes everything about how you think about the economics.

The metric that captures this is LTR --- lifetime revenue. It measures the total revenue generated by a cohort of new customers over time, divided by the number of customers in that cohort. It lets you compare sales productivity across time periods and tells you whether your retention is getting better or worse. Not whether you're growing --- whether the customers you worked so hard to acquire are actually coming back.

You can also track what I call iLTR --- incremental lifetime revenue --- the revenue added in each subsequent month after the first. This tells you how long it takes to recoup the money you spent acquiring each customer. That window is called your payback period, and it matters enormously for cash flow. If your CAC rises, your payback period lengthens, which means you need more cash just to sustain the same rate of growth. If your iLTR declines, same problem from the other direction.

My first Hairstory model assumed the hairdresser side of the business would be the primary revenue driver. I thought DTC would be small --- icing on the cake. I was completely wrong. Hairdressers were cautious about something so new and different; it took years for that channel to develop. Meanwhile, the DTC business took off. Fueled by exactly the retention dynamics above, we reached cash flow breakeven within 18 months of launch.

Predicting new customer acquisition is hard. Retention is where the truth lives.

Now for the other side of the equation --- the one my grandmother would have understood intuitively, because she grew up in Weimar Germany and never wasted anything.

When I was selling used cars at iMotors, we weren't thinking about LTR or retention. We were thinking about whether we made money on each car we sold. This is what's called unit economics --- the math behind each individual transaction that answers the most basic question in business: after I earn this revenue, how much is left?

Start with revenue per transaction --- your AOV. To earn that, you had to spend something: ingredients, manufacturing, raw materials, the cost of buying and reconditioning a used car. These are your COGS --- cost of goods sold. What remains after COGS is your gross profit.

Then there are direct expenses you only incur when you actually have a transaction: shipping, payment processing fees, commissions. Subtract those and you arrive at your contribution profit --- the amount left to cover everything else required to run the business. As a percentage of revenue, this is your contribution margin, and it is one of the most telling metrics in any business.

Here is one of my golden rules: every expense above your contribution margin should be variable --- directly tied to generating revenue. Every expense below it should be fixed, or at a minimum independent of incremental revenue. This discipline is what lets you compare businesses across completely different industries in a common framework.

Software businesses are attractive to investors because their marginal cost of serving one more customer is nearly zero --- each new customer is almost pure contribution. Heavy manufacturing is the opposite: enormous fixed overhead requires very high volume to reach profitability and contribution margins are thin. Hotels and airlines are capacity businesses --- they lose money below a certain occupancy rate and hit a ceiling at full utilization. Understanding where your business sits on this spectrum is essential to knowing what scale you actually need.

With those two lenses in place --- the revenue model and the unit economics --- you can build a model worth trusting.

I start with the revenue drivers: new customer acquisition and repeat purchases, structured as a cohort waterfall so each month's new customers are tracked individually as they generate revenue over time. Then I pressure-test the unit economics and layer in fixed expenses. I work hard to minimize assumptions and build a sensitivity dashboard so I can see exactly how much each variable moves the output.

Which output matters most? Not sales. Not profit. Cash required to reach breakeven.

Every model I build produces a simple P&L and a bare-bones balance sheet. I set opening cash to zero and look for two numbers: how long until the business stops losing money, and how much cash does it burn before that happens? Then I stress-test the inputs. What if CAC rises 30%? If retention drops? If your margin assumptions turn out to be optimistic? If you need twice as much capital to launch as you planned?

Getting to cash flow breakeven is the most important milestone any entrepreneur can achieve. It is the inflection point where the power shifts --- from investors back to the founder, from the market back to you. It is the moment you stop surviving and start deciding: how big, how fast, which direction?

That is what the model is really asking. Not whether the idea is good. Whether it can become self-sustaining --- and what it will cost, in time and money and nerve, to find out.

I built my first Hairstory model on a gym floor, half-watching my son shoot free throws. I built the Sans Savon model in a spreadsheet while Erica and I debated whether to bet our savings on ourselves rather than the stock market. In both cases the model didn't tell me the idea was good. It told me what I had to believe for it to be good --- and gave me the chance to decide whether I believed it.

That is what good models do. They don't predict the future. They show you the shape of the bet you're making.

The rest is up to you.

Grandma Hella at our wedding in my parents' house.