A few months ago, I wrote a blog post warning entrepreneurs that their LTV math was wrong. In other words, their methodology for calculating the lifetime value of a customer was incorrect and, typically, a grossly optimistic figure. It got a lot of attention and even inspired a new module in InsightSquared to help companies do the correct calculation.

In talking about getting these kinds of calculations correct, one of my portfolio company CEOs made an interesting observation to me. We were talking about our customer acquisition costs (CAC), which represents the other side of the unit economics coin. The company is converging on profitability and is thus getting more and more focused on its unit economics and CAC.

“Our historical CAC is obviously irrelevant,” he declared in the board meeting. “All that matters is our marginal CAC.” This struck me as such an important insight that I would try to unpack it a bit because, as with LTV, most entrepreneurs think about their CAC incorrectly.

Many entrepreneurs calculate CAC by simply adding the amount of money they spend on sales and marketing and then dividing it by the number of customers that they have. If last quarter you spent $100,000 in sales and marketing and you acquired 10 customers, your CAC is $10k. A hypothetical consumer company may spend $100,000 in sales and marketing to acquired 50,000 customers to achieve a CAC of $2.

That simple math may be historically accurate but it’s not very helpful as a planning framework. The historical customer count in a consumer company, for example, typically contains customers that came in through a range of sources – paid (e.g., SEM, SEO, Facebook) and unpaid (e.g., organic). Some of these tactics are highly scalable and others are not. Some may even have reverse economies of scale and, thus, actually get worse with scale.

Customer acquisition is a bit like drilling for oil. Sometimes you hit a gusher and it flows beautifully, but eventually that well runs dry and you need to find another one. Similarly, you may come upon a particular customer acquisition tactic and it might be working beautifully and with favorable economics, but eventually it runs its course and you need to deploy another tactic.

That’s where the concept of “marginal CAC” really matters, as my portfolio company CEO wisely observed. When looking ahead, you want to know what the cost is of the next set of customers that you’re trying to acquire. That is, what is the acquisition cost of the marginal customer. If the only tactic you can scale is Facebook ads, then the marginal cost of acquisition of Facebook ads matters more when forecasting your future CAC, not the blended average that you’ve been able to achieve historically.

An e-commerce entrepreneur we were speaking with the other day cited a historical customer acquisition cost of $150. The CEO then projected that those costs would decline with time – a perfectly logical assertion.  As part of our due diligence, we asked for how the numbers broke down by acquisition channel and discovered that the $150 was a blend of organic and paid channels and that the only paid channel that would scale going forward represented an expensive $300 CAC. Thus, their marginal CAC was most likely to be going up over time with scale, not down.

The lesson? As we head into an era of savvier investors who scrutinize unit economics more thoroughly, don’t get caught looking naive when presenting your CAC projections – make sure you’re thinking about marginal costs, not just historical ones.