When I was starting my last company, I somehow came across “Bessemer Ventures Top 10 E-commerce Rules.” Though only a few pages, it was like a bible to me. It had one quip that has driven immense value for me, whether in running and growing a company or investing. It went like this:
If your Customer Lifetime Contribution is at least two times the cost of acquiring a new customer … you’re in great shape. If you’re doing any better than that, it’s irresponsible for you not to be spending more aggressively on marketing.
They called it Customer Lifetime Contribution (CLTC); in my company we called it Lifetime Value (LTV). I’ve seen it under various names, but the metric we’re talking about is what a customer is worth to your company (it’s a bit more involved as I’ll explain, but essentially, take the average number of orders per customer, multiplied by your average order net value). Once you know how much one is worth, it is IRRESPONSIBLE to be spending less than half of that number on acquiring new customers.
That caught my attention: this is the alchemy of business. If I know that a customer is worth $10 after 1 month, and I can acquire that customer for anything less than $5, I am doubling my money.
I should be taking out million dollar loans and doubling my money. Mortgage the house, bet the farm, then double your money. But not so fast! Doing this with any degree of confidence means you have to know your numbers, and in the case of LTV, you’d better be calculating it right; you’re betting the farm after all.
Most people have the wrong number
I love e-commerce; I work in e-commerce, I invest in e-commerce companies, I mentor e-commerce companies. I see a lot of them, and I’m blown away at how bad people are at building this number.
One of the most common mistakes I see is people determining their LTV based on revenue rather than net income.
If I spend $1 on advertising to get a $1 sale, it is not a wash. I paid for fulfillment, storing inventory, hosting, accounting, marketing salaries, et cetera to get that $1 item to the customer. I’m losing money all day. This is a dangerously common pitfall. If you bet the farm with this as your model, you will lose the farm.
Another is factoring LTV only on buying customers; I calculate based on every email that goes into my database. It’s easier to build models around customer acquisition if you’re not trying to guess who will buy and who won’t.
It’s also attractive to look at LTV as a single data point regardless of the timeline of that number. Don’t do it! You should never give an LTV number without its accompanying timeframe. An LTV point at year three is different than at month three.
How do I want to see LTV calculated
Let me walk you through how we built our model. This isn’t the only way to do it, but it is a great foundation and has served the companies I work on and me very well.
We separated our customers into cohorts based on the month in which we got their email (they placed an order, registered for a newsletter, etc), and looked at the performance of that cohort specifically.
We figured their LTV (cohort’s average # of orders * average order size * net profit percentage) at different points in their growth: 1 month, 6 month, 12 month, and 3 year.
This is incredibly valuable for young startups with little data or fast growing companies whose LTV from 3 years ago isn’t representative of the company performance and retention today.
So rather than build an overall average, I plot points and look for trends. If I need a number to spend against, I typically take the LTV from last month, or an average of the last few months, if it looked trustworthy.
When starting out, you have to pick a profitability number to work off of, so we used an adjusted EBITDA average (adjusted just means you subtract any major expenses that won’t be there the next year, to keep the number representative of the true nature of your business).
We looked at how much profit we were making annually (we ranged from 5% — 25% depending on what was going on that year). If we were investing heavily in something new, EBITDA was low. If we were only making sales and everything was returning on the year, EBITDA was high. So we put a stake in the ground at 12% and said, “even if we are below 12% we know it should be around 12%, so 12% is the number we base our calculation on.”
This matters because you want to be able to compare apples to apples year over year.
So here’s how we looked at the business: if your month 1 LTV was $10, and you could get a customer for $5, then you could go ahead and spend ’til you die because you can double your money in month 1.
When you cross that threshold, you make a business decision knowing that you aren’t going to double that money until month 6, then year 1, and then year 3. It gives you a lot of power when you have that kind of confidence in your marketing spend — you understand how far your investment into acquiring new customers is taking your cash reserves.
One interesting thing about our business is that our customer behavior was to shop a lot in the first few months, and then level off into more steady spending. From this we saw a majority of our LTV realized in the first 6 months, which is great because there’s a big top-end opportunity to increase the LTV after 1–2 years that we aren’t taking full advantage of yet, and a majority of our investment in customer growth will be returned quickly.
Next, we need to consider what the finance world calls ‘discounting’: effectively that a dollar today is worth more than a dollar three years from now.
If you are spending a lot to get a customer, and aren’t realizing the full LTV value until three years out, you should consider the declining value of every dollar you spend that you have to wait to get back. Think about inflation, other projects you could invest in, or what your money would have made if you’d just dropped it in an index fund. If you are taking years to recoup an investment in new customer growth, make sure you are doing it with your eyes wide open.
To do this calculation, first consider inflation, say, 3%. You would say this dollar is worth 1/ 1.03² — again you’re saying, don’t count the return on this dollar in a year as the same dollar you see today.
Also, consider the other ventures you could be spending that money on. For simplicity, at our company we decided if the business is making 12% EBITDA, then that’s the other potential (putting money into other efforts inside the company and earning 12% on it at the end of the year), so a dollar not invested inside company projects is worth 1/1.12²
Keeping with our example above, let’s say your LTV over 3 years is $500, you get $350 of that back in year 1, and then $75 in year two, and $75 in year three. The model would say the LTV in year 1 is (350/1) — so effectively nothing changes because there is no opportunity cost. In year 2 it would be (75/1.15²) and year 3 would be (75/1.15³). Therefore, the actual customer LTV in year 1 is $350, in year 2 is $406, in year 3 is $456.
You can do this math with months instead of years if you’re being a bit more precise; simply adjust to the needs and cadence of your business. Avoid the analysis paralysis--you could rabbit hole on this pretty easily.
Spend the time you need to build the model once, and then go build a company.
So remember: once you figure this thing out, it is then irresponsible to not maximize your spending in customer acquisition knowing you are getting sound returns on it in timelines your business is prepared for.
Hope this helps.
Al
*edit — had to correct some notation stuff, thanks for the comments!
