There are two types of companies: single-purchase and “customer lifetime value” (LTV) businesses. Crudely, mattress sellers and gyms. It’s crucial that your marketing investment strategy matches the kind of business you’re in: if you run your LTV businesses as if you’re selling mattresses, you’ll be in serious trouble.
When you’re in the mattress business, you sell a product that customers buy every 7 years. There’s no such thing as “lifetime value:” whatever you make on the sale is what you’ll make from this customer, ever. One might argue that fancy mattress stores (ex. Macy’s Home) have found a way around it - they upsell you on pillows, mattress covers, sheets; for this example, let’s assume you’re running a pure-play mattress store that doesn’t do that.
If you’re selling mattresses, you must attain first-purchase profitability - and as much of it as you can, ideally, hundreds or thousands of dollars. That is, your customer acquisition cost (CAC) cannot possibly be more than your total profit from the first sale1. You have pretty weak incentives to develop a relationship with the customer; you don’t really need to get to know them - it’s a waste of time. They’ll move to another city before they need another mattress anyway.
An opposite kind of business is a gym. There, your monthly profit is humble, but most customers stick around for years, and that’s how you rack up hundreds of dollars in profit. Netflix and Hulu are examples of these “gyms.” Monthly subscription is barely $10. And yet, when someone becomes a customer, they are very likely to stick around indefinitely if they have a good experience, generating $10/month for years to come. Their LTV is in the hundreds of dollars. Suddenly, only $10 in revenue per transaction doesn’t look so bad.
In an LTV business, you MUST focus on the relationship. You must keep this customer around and happy in order to realize the potential profit. You also have to understand how cumulative profits add up over time from each customer that you acquire.
Once you have a solid sense of how customer cohorts behave, you have an IMPERATIVE to spend beyond first-transaction profitability: if you don’t, you’re leaving money on the table. How exactly? Let’s say a customer makes you $100 in profit in the first year, but you only spend up to $20 to acquire a new customer. This is as if you come to a bank, they say “put in $20, we’ll return $100 in a year” and you choose not to. What?? Why?!? Do you know of someplace to invest money that provides a 4x return within a year4?.. I’ve written more about this investment framework for marketing, read up on it if you’re interested.
That is, in an LTV business, you should allow for some period of loss for each new customer - where you spend, say, $20 to acquire a customer, and make that up after a few months of their lifetime. How long you’re willing to wait to get your money back - enough profit to make up the initial CAC - is called a breakeven period; typical values here are 6, 12, 18, or 24 months. Breakeven period is most often set by the CEO or the board. The longer the period, the faster your company’s growth will be, as you’ll be able to pour more money in to acquire more customers. Thus, if you run your LTV business to first-order profitability, your growth will be really hampered. This is a classical “long-term growth vs short-term profitability” tradeoff in action.
The breakeven period should be one of the two primary goals that the CEO sets for the CMO around growth; the other one should be “the number of customers acquired.” Notice the complete absence of CAC in these goals. Companies that pay too much attention to optimizing CAC often find that the CAC drops but revenue does not improve: it’s easy to reduce your CAC if you’re OK with LTV also dropping. One obvious technique to do so is couponing / discounts, but there are many others3.
Beware of CAC goals; use breakeven periods instead. What you’re buying with marketing is future cash flows: from an investment standpoint, you shouldn’t care whether you bought 20 customers each worth $5 or 10 customers each worth $10. Good marketing teams should have flexibility around LTV vs CAC tradeoffs, as long as they’re able to break-even within the time horizon that the CEO wants.
Allow me to get into one important detail of this “breakeven point” framework. How do you actually know when the breakeven point is going to be for customers that you acquired in the last couple of weeks? If your target breakeven is 6 months, you’ve got to be able to tell if these newly acquired customers are “on track” to get to 6 month profitability or not, without waiting for 6 months.
This is where LTV proxy metrics, or LTV prediction models, come in. Your marketing analytics, or sometimes finance function needs to develop a way to confidently predict LTV of newly acquired customers after a fairly short period of time - ideally, a week after they signed up. If that isn’t possible, 2-4 weeks after signup2.. Once you have that, every launch of a new marketing program will be easy to evaluate: acquire a few weeks worth of customers, look at how much you’ve spent on them, compare that to their predicted LTV. See where the breakeven point is, evaluate that against the goal.
How can you develop such a proxy metric? Start with simple correlation analysis. Do an Excel dump of every single customer as a row. Calculate 7, 14, 28 day profit from each customer as a column; add in X-month LTV (X will be dependant on your breakeven period goal). Then, determine the correlation coefficient between 7d profit and LTV; 14d profit and LTV; etc. Excel even has a built-in function for that. Found a correlation coefficient greater than 0.7? Good, use the shortest time window that has that.
Once you have this basic model, recognize its limitations: correlation and causation are not the same things; also, a better predictive model can be built using more sophisticated machine learning techniques. For example, if you’re Netflix, your “real” LTV model would likely include features like “did this customer watch a Netflix Original.” If you’re Instacart, it’ll probably include “is the customer a good tipper.” This more complete model would likely take your data scientist several weeks to build; it will be a bit - not drastically - better than your single-feature model; in the meantime, you can have a solid feedback loop with just a simple approach.
In closing, let me leave you with one thought: the simpler the framework, the higher the chance that it’ll be adopted. Breakeven period concept is so intuitive. Your LTV proxy metric can be really simple too. Don’t over-analyze, don’t go for perfection out of the gate. Get 80% of the way there with 20% of the effort.
If you don’t make money selling mattresses, you’re Casper Inc: Scott Galloway rightfully says that “the economics work better if Casper sent you a mattress for free, stuffed with $300.”
Don’t use 30 days… 28 days is a better rule of thumb - most businesses have significant day-of-week variation.
For example, reallocating investment to channels where CAC is lower by a buck - even if LTV is lower by 5 bucks. Measure just CAC at your own peril: you’ll sooner or later discover how good humans are at gaming incentives.
There are some very good reasons why you wouldn’t want to increase your CAC targets in this case. For example, if you don’t have enough capital to last you these 12 months: in that case, if you were to increase the CAC target, you’d be long-term rich and short-term bankrupt. Or, if you don’t have the inventory to sell to all these new customers, there’s no point in acquiring them - instead, invest into inventory first. These good reasons are usually not why entrepreneurs don’t invest up to the LTV; more often than not, it’s because they don’t trust or know their CAC and LTV numbers. Check out my recipes on what to do if that’s the case if you’d like.