Acquisition Payback Audit · System 03 · deciding where the next dollar goes
Compare Channels by What Their Customers Earn
You’ll often hear that a customer should be worth at least three times what it cost to win them, or that they should pay that cost back within a year. Those targets come from investors and research firms who mostly study software companies, and they shift from one year’s report to the next. The most useful guide for your own marketing money is already in your records: what customers from each channel have paid you so far.
The idea in one line: compare your channels with each other, using what their customers have paid you, and treat any published target as a rough check at most.
01
Where the two common targets come from
Two numbers get quoted more than any others when people talk about the cost of winning customers.
Payback time
How many months of profit from a customer it takes to cover what you spent winning them. Published targets run from under 12 months to under 24, mostly from venture investors and surveys of software companies, and the “good” range moves between their own reports.
Customer value vs. cost to win
What a customer is worth over their whole time with you, divided by what it cost to win them. “Three times or more” is the version you hear most; the same sources have used four to six times for other kinds of business. Research supports comparing value with cost. It doesn’t support any single ratio as the line between healthy and broken.
So a channel whose customers are worth 2.4 times what they cost to win can be doing fine, and one at 3.2 times can be in trouble. What tells you which is how it compares with your other channels, and whether it’s getting better or worse.
02
Work it out from what customers paid
A quick formula built on an average bill and an average cancellation rate is fine for a dashboard. A budget decision deserves the real history.
How to work it out
Group customers by the channel they came from and the month they started. For each group, add up the profit they brought in, month by month, until it covers what it cost to win them. The month that happens is that group’s payback time.
Working from real groups of customers allows for slow starts, discounts, customers whose spending goes up and down, and customers who leave early. It takes longer to set up the first time, and you only set it up once.
Do this for each channel, and for each kind of customer if you serve very different ones. A single company-wide number can look healthy while one channel drags it down and another carries it. That’s exactly what System 02 found in the home-services example.
03
Decide what happens to each channel
1
Grow it. It pays back quickly, its customers stay profitable, and there are plenty more customers to win there. The clearest case for a bigger budget.
2
Hold it. Payback and loyalty are fine, and the channel is already reaching most of the customers it can. More money would mostly buy pricier clicks.
3
Fix who it brings in. Each new customer costs a reasonable amount, but too many of them leave early. The channel is probably attracting the wrong customers. Tighten who it targets, or add a screening step, before you cut it.
4
Test it before trusting it. Some channels take credit for customers who were coming anyway. Ads that show when people search for your business by name are the usual example. Run the simple switch-off test in Where the Next Dollar Goes before deciding either way.
5
Cut it. Slow payback, customers who don’t stay, and no sensible way to attract better ones. The channel is buying customers the business is better off without.
This is the step where an AI assistant will reach for a published target if you let it, so the last message below matters as much as the first.
Do this step with an AI assistant
Paste in the all-in cost per channel from System 01 and the customer groups from System 02. Then send these three messages, in order.
“For each channel, work out how long its customers took to pay back what they cost to win, using the profit they brought in. Only use the months I have data for.”
“Give each channel one decision: grow it, hold it, fix who it brings in, test it before trusting it, or cut it. Say which number drove each decision.”
“Point out any channel that might be taking credit for customers who were coming anyway, and say what test would settle it. Compare the channels with each other, and leave published ratios and payback targets out of the decision.”
04
A worked example
Back to the home-services company from System 02 and its four channels, now with a decision for each.
The four decisions
Cold calling: grow it. Its customers brought in the most profit over their first six months, and the sales team could handle far more.
Local service ads: hold it. It pays back well, and it already reaches most of the local market. More budget would buy pricier clicks and few extra customers.
Referrals: fix who it brings in. It costs little and brings plenty of customers, but profit is thin because many referrals are small one-off jobs. Add a minimum job size to the referral reward before deciding whether to grow it or cut it.
Ads on the business name: test it before trusting it. Nobody else is bidding on the name right now, which is when research finds these ads mostly take clicks the free listing would have got anyway. Switch them off in one area for four weeks and count every visit from people searching the name, paid and free, before moving the money.
Ranked only by what each new customer cost, this company would have kept pouring money into its lowest-cost channel and worried about its most expensive one. Ranked by what those customers went on to earn, the decisions run almost exactly the other way.
Systems 01 to 03 are one method in three passes: add up the full cost of winning customers, judge the customers each channel brought in, then decide where the next dollar goes using your own numbers. The complete audit puts them together so you can run it start to finish, with the data to gather first and a prompt to paste into whichever AI assistant you use.
