The Retention Curve Nobody Wanted to See
San Francisco · September 7, 2025
Priya puts the chart on the screen and nobody says anything for about ten seconds. The chart shows our retention curve — the percentage of customers still active at each month after signup — plotted across twelve months. The curve should flatten. Every SaaS company wants the smile: a steep initial drop that levels off as you reach your core users. Our curve doesn't smile. It frowns. It drops steeply in months one through three, flattens briefly around month six, and then drops again at month nine.
"That's a double dip," the CEO says. She says it quietly, the way you'd say "that's a tumor" if you were looking at an X-ray.
"Yes," Priya says. "Month-one retention: 71%. Month-three: 54%. Month-six: 48%. Month-nine: 39%. Month-twelve: 34%."
Thirty-four percent twelve-month retention. That means two-thirds of the customers who sign up are gone within a year. We're filling a bucket with a hole in the bottom, and the hole is getting bigger.
The retention analysis started because of a question the CFO asked during a finance review: "Why is our net revenue retention declining?" Net revenue retention — the percentage of revenue from existing customers that you keep and grow — had dropped from 103% to 94% over three quarters. Above 100% means your existing customers are spending more each year than they did the year before. Below 100% means they're spending less, either through downgrades or cancellation.
Ninety-four percent isn't terrible. It's not great either. The median for B2B SaaS is around 100-105%, and the best companies run at 120%+. We used to be at 103%, which put us in the middle of the pack. Now we're below it.
I asked Priya to dig into the data. I didn't expect what she found.
The double dip in the retention curve was new. A year ago, our curve looked normal — the classic drop-then-flatten pattern. Something changed around month nine, and the change coincided with a product release we'd done in January: a major UI redesign.
"You're saying the redesign caused churn?" the head of product asks, her voice carrying an edge I recognize as defensive.
"I'm saying the timing correlates," Priya says, carefully. "Customers who onboarded before the redesign have a different retention pattern than customers who onboarded after."
Customers who onboarded before the redesign have a different retention pattern than customers who onboarded after. The redesign wasn't the cause. But it was the trigger.
The redesign moved core features around. Things that used to be in the left sidebar were now in a top navigation. The reporting module — our stickiest feature — went from being one click away on the homepage to being nested inside a "Analytics" section that required two clicks. Workflows moved. Dashboards looked different. The underlying functionality was identical, but the muscle memory was broken.
I called twenty customers who had churned between months eight and eleven. Of the twenty, I reached twelve. Here's what they said, condensed:
Five mentioned the redesign specifically. "I spent an hour trying to find the reporting feature." "Everything moved and nobody told me why." "I felt like I had to relearn the product." These weren't casual users — they were power users who'd been with us for over six months. They'd built workflows. They had habits. And we broke those habits without warning.
Four cited a lack of ROI. They'd been using the product long enough to evaluate whether it was worth the money, and they'd concluded it wasn't. This is the month-nine problem: customers who get past the initial onboarding hump and stick around for six months will eventually do a cost-benefit analysis. If the answer isn't clearly positive, they leave.
Two said they'd found alternatives. Not competitors, exactly — one was using spreadsheets, and the other had built an internal tool. They'd outgrown us in the sense that they'd developed bespoke solutions that fit their workflows better than our general-purpose product.
One said he'd been promoted and his replacement didn't use our product. The champion left, and nobody replaced them. This is the silent killer in B2B SaaS: when the person who bought your product moves on, your product often moves on with them.
We held a retention summit — four hours, cross-functional, no phones. The CEO, head of product, head of engineering, head of CS, Priya, and me. We identified three workstreams:
Workstream 1: Fix the redesign damage. We couldn't undo the redesign — too much engineering work had gone into it, and the new UI was actually better for new users. But we could make the transition smoother for existing users. We built a "where did things go?" guide that appeared as an overlay for users who'd been active before the redesign. We added customizable navigation so power users could put features where they wanted them. And we sent a personal email from the CEO to every customer who'd been active for more than six months: "We changed some things. Here's why, and here's how to find what you need."
Workstream 2: Month-six health checks. We created an automated health score based on product usage, support ticket volume, and feature adoption depth. At month five, if a customer's health score dropped below 70 (on a 100-point scale), a CS manager would reach out proactively. Not to sell — to help. "I noticed your team hasn't used the reporting module in three weeks. Is there anything we can help with?"
Workstream 3: Champion tracking. We started tracking the primary user in each account — the person who signed up, who logged in most frequently, who opened the most support tickets. If that person's activity dropped or they left the company (we used LinkedIn integrations to detect job changes), we triggered an outreach sequence to other users in the account.
The workstreams have been running for eight weeks. The data is early and imperfect. Here's what I can see:
The "where did things go?" overlay was seen by 1,200 customers. Of those, 340 clicked through to customize their navigation. Support tickets related to "can't find a feature" dropped 41% in the first month.
The month-six health checks triggered for 89 customers. CS managers reached out to all 89. Of those, 23 responded and 18 ended up increasing their usage. The other 66 either didn't respond or responded with "everything's fine" — which, in my experience, means they'll churn in two months but don't want to tell you yet.
Champion tracking identified 34 primary users who'd either left or gone dormant. In 12 of those accounts, we successfully engaged a secondary user. In 8, we couldn't find anyone else who cared. Those 8 accounts are effectively dead, we just haven't gotten the cancellation email yet.
Month-nine retention for the August cohort: 43%, up from 39% for the May cohort. It's a four-point improvement. I want it to be more. The CEO wants it to be more. But four points is four points, and in a world where the curve was going the wrong direction, any reversal is meaningful.
The truth about retention is that it's the hardest metric to move because it's the sum of everything else. Your product, your onboarding, your support, your pricing, your competition — it all flows into whether someone stays or leaves. You can't fix retention with a single initiative. You fix it with a hundred small things done consistently over time.
You can't fix retention with a single initiative. You fix it with a hundred small things done consistently over time.
The retention curve still doesn't smile. But the frown is getting shallower. I'll take it.