How You Sell Decides How Long They Stay

You can predict how long a customer will stay before they have used your product once. The signal is not their size, their industry, or their score on anybody’s fit model. It’s how they bought.

Start with the number. Customers who receive no initial discount go on to last 2.6 times longer than customers who did. Not 10 percent longer. Two and a half times. The discount you gave to close the deal is the same discount that shortens the customer’s life by more than half. You did not buy a sale. You bought future churn, and you paid for it twice.

Customers who buy with no discount go on to last 2.6 times longer than customers who needed one. The discount didn’t cost you margin. It cost you more than half the customer’s life.

The mechanism is not mysterious. A customer who paid little to nothing for it will do little to nothing with it. Cut more than 25% off list and the customer quietly concludes the product is worth less than you said it was, so they invest less to learn it, less to integrate it, less to make it work — and the number one reason customers churn is that they never got a measurable result. Cheap to buy becomes cheap to ignore.

Not every discount does the same damage. In the bonding study we put the pricing terms against the retention behavior, and the split was clean: a discount used to close a wavering buyer tracks with shorter loyalty, while a discount traded for a longer commitment tracks with longer. Same money off the price. Opposite customers.

That tells you what the price is really measuring. It is not measuring the deal. It’s measuring whether the buyer intends to get a result. Give money away to overcome doubt and you have paid a customer to keep doubting. Trade it for commitment and you have made them declare intent. A customer who commits is telling you they mean to get a result, and that intent is the leading edge of fit.

A discount given to close a wavering buyer tracks with shorter loyalty. A discount traded for a longer commitment tracks with longer. Same money off the price, opposite customers.

Put commitment head to head with the thing everyone else reads at the point of sale — how big the customer is, how big the deal is — and raw size washes out while the commitment signal survives. Serious customers stay longer. Not bigger customers. That is the same finding as the enterprise paradox from a different angle: the segment that posts the best retention on the dashboard bonds essentially none of its customers, because the account was held by procurement, not by attachment.

And it is not only discounts. The same seriousness signal runs through every low-commitment way in. Customers who start on a free trial stay less than half as long as customers who pay to start, a 2.3x gap in customer half-life across 130,000 customers — and the benchmark finds free trials affect retention “similar to” pricing discounts, for exactly one reason: both of them affect customer effort, and effort is what produces results. Opt-outs are worse than they look, too. They add very low-fit customers who rarely stick but consume the same resources, and they force good-fit customers — customers who would have subscribed anyway — into a risk-mitigating cancellation decision at a date you chose for them. An artificial crisis.

So why doesn’t every company already know this? Because nobody measures it separately. I wrote the answer down in 2018 and it has not changed: “The only reason this is not obvious to us is that we don’t measure their churn separately. The way we report churn mixes all groups — and therefore reasons — together.” The discount buyers’ damage hides inside the blended number. You are looking straight at your most predictive retention signal every month and reporting it in a way that cancels it out.

Run it in the other direction and the whole thing inverts. Making a product harder to buy filters out the less serious customers, which gives you better use-case fit, vetting before the deal instead of after, and a clearer business need. I wrote it as a rule that year and I have never had to revise it: increases in risk — making it harder to buy — increase retention at an increasing rate. Even sales-cycle length carries the signal; longer cycles go with longer customer lifespans, and the ratio looks like somewhere between 1:4 and 1:10.

Which means the job of sales is not what your comp plan says it is. “Their job is not convincing but filtering — to cause customers that are interested enough to subscribe but not interested enough to change their behavior to choose NOT to subscribe.” Buying signals are sacred because they are such accurate filters. Remove the price barrier and you have not just let in the wrong customer; you have told the right customer their skepticism was correct, and diminished their own estimate of the value.

Sales is not there to convince people to buy. It’s there to get the customers who won’t do the work to choose not to buy.

One limit, because the wrong version of this argument gets expensive. Commitment is a signal you read, not a cage you build. Contracts mostly select rather than cause: in companies with a customer half-life over four years, multi-year customers stay only about 6% longer — roughly three extra months — and you cannot be sure they stayed because of the contract, since the buyers who sign multi-year terms tend to be the ones who would have stayed anyway. The calibration rule is the one to hold onto: the cost of failure needs to match the difficulty of the change — no more, no less — and the minimum commitment should not be shorter than the time it actually takes to achieve results. Set the opt-out before real success is achievable and the customer is forced to decide before they can possibly have anything to decide with.

The wrong direction wins anyway. Every incentive at the front of the company is pointed at the close. Commission-based comp makes selling cheesy and drives overselling and churn; the fix is paying on the durable book rather than the close. Lazy selling — leaning on a discount, allowing unrealistic expectations, giving the expert services away, selling bad-fit accounts to hit quota — produces lazy customers who don’t change behavior, don’t get results, and leave. The retention lever with the earliest read and the lowest cost sits inside the one process the quarterly number pushes hardest in the wrong direction.

The test to apply is one sentence: any reason to buy is bad for growth if it isn’t also a good reason to stay. A discount is not a reason to stay. A free trial is not a reason to stay. An easy exit is not a reason to stay. Each one closes the deal by removing the exact thing that would have kept the customer.

So go read it in your own data. Take every customer you signed in the last two or three years and split them by how they bought — discount versus no discount, free start versus paid start, opt-out versus no opt-out, short cycle versus long — and run the retention curve for each group separately instead of blended. It costs you nothing but a query, and it uses data you already have. Whatever the gap turns out to be at your company, that is the size of the retention problem you are creating before onboarding ever begins.