The Month 13 cliff: when 90% retention is just contracts that haven't expired yet

A company sells annual contracts, billed up front. Twelve months of revenue, locked the day the customer signs. For the first year of any cohort's life, the logo retention curve is a flat line near 100 percent. Not because the product is loved. Because the contract hasn't come up for renewal yet. Nobody can leave a room they're locked into.

Then month 13 arrives, the first renewals come due, and the line finds out what the product is actually worth.

Why annual contracts hide churn

Retention measures whether customers stay. An annual contract removes the decision from the customer's hands for twelve months. They can be quietly miserable, log in once a quarter, and route the renewal to procurement with a note that says "find us something else." None of that shows up in a retention figure until the contract reaches its term.

So a cohort of annual customers will report something close to 95 to 100 percent logo retention through month 12 almost regardless of how the product is performing. The number is real. It is also nearly content-free. It describes the contract structure, not the relationship.

This is why an experienced underwriter doesn't ask "what's your retention." They ask "what's your retention on cohorts that have passed their first renewal." Those are two different questions, and on a young company they can have very different answers.

What the cliff looks like

Plot logo retention by months-since-start, cohort by cohort, and a company on annual terms produces a recognizable shape: flat near the top, then a step down at month 13, then flatter again until month 25.

The size of that first step is the whole story. A drop from 100 to 92 at month 13 is a healthy book renewing. A drop to 78 is a product that customers tolerated for a year and then left. Both companies reported "above 90 percent retention" four months ago. Only one of them had renewed anything.

The trap is that the youngest, fastest-growing companies have the least renewal data and the most flattering raw number. If 70 percent of your customers signed in the last twelve months, 70 percent of your book has never made a stay-or-go decision. The blended retention figure is dominated by contracts that simply haven't matured.

How to read it before an investor does

Three checks separate a structural illusion from a real result.

Look at the renewal cohorts in isolation. Take only the customers who have actually reached a renewal date and measure retention on that subset. If that number is materially worse than the blended figure, the blended figure was borrowing strength from unexpired contracts.

Watch the second renewal, not just the first. Some books survive month 13 on inertia and budget cycles, then fall apart at month 25 when the second renewal requires a real decision. A clean first renewal followed by a steep second one is a familiar pattern, and it is worth surfacing yourself.

Separate the count from the dollars. Logo retention and net revenue retention can diverge sharply at renewal, when prices reset and seats get renegotiated. A book that keeps 90 percent of its logos but renews them 15 percent cheaper is not the same book it was, and the dollar curve will say so even when the logo curve looks calm.

What to present

The fix is not to hide the cliff. It is to show it before anyone has to ask. Present retention indexed to months-since-start rather than calendar time, so the renewal step is visible instead of buried in a blended average. Call out how much of the book has passed its first renewal. If the post-renewal cohorts hold, that single chart does more for credibility than any headline number, because it proves the retention you're claiming has actually been tested.

If the post-renewal cohorts don't hold yet, an investor will find that in an afternoon regardless. Better that the founder is the one who frames it, with a view on why and what's changed in the cohorts signed since.

What this means for the founder

A 90 percent retention figure on a book of annual contracts isn't a lie. It's an unfinished measurement, and whether it becomes a strength or a problem depends entirely on the cohorts the headline number is hiding. The founder who can show the renewal step, and stand behind it, is having a different conversation than the one defending a blended average.

Levian reads your transaction data and rebuilds retention the way an underwriter would: indexed to each cohort's own start date, split at the renewal boundary, logo and dollars side by side. So you see your month 13 cliff before you're sitting across from someone who was always going to look for it.