Your Churn Dashboard Tells You Customers Left. It Can't Tell You Why.

A customer canceled on a Tuesday morning. I knew within minutes. Seats went from six to zero. Status flipped to canceled. MRR dropped by an amount I could read down to the dollar.

Every part of that moment was captured perfectly. The one thing I wanted to know wasn’t in there anywhere.

I ran Pulse360, a B2B SaaS for financial advisors. Before that I spent 20 years as a financial advisor myself, so I was building for the person I used to be. We signed customers representing about $1M in demand. We kept $280K. The gap was roughly 80% churn, and the education cost me around $720K.

For a long stretch of that, I had good analytics. Clean event tracking, cohort charts, the whole setup. And I still could not tell you why people left. Not because my instrumentation was bad. Because of what analytics are.

Analytics record events. Leaving isn’t an event.

Your dashboard tracks things that happen at a timestamp. A login. A seat assignment. A button click. A support ticket. The cancellation itself.

The reason someone leaves is almost never one of those. It’s an accumulation of small moments, and most of them are moments where nothing happened.

The report they didn’t run because they weren’t confident the numbers were right. The Tuesday they opened your product, couldn’t find the thing they needed, and quietly went back to the spreadsheet that always works. The export they had to clean up by hand for the fourth week running, which took nine minutes and annoyed them more than nine minutes should.

None of that fires an event. There is no user_gave_up_and_used_excel in your schema. The absence of an action gets recorded as a gap in a chart, with no reason attached.

So a churn dashboard can tell you that someone left, when they left, and usually who. It is structurally incapable of telling you why. That’s not a gap you close by adding more tracking. Perfect instrumentation still only records what your product did. It has nothing to say about what your customer was trying to get done.

The interface I was certain was simple

Here’s mine.

At one point I redesigned a core part of Pulse360 to make it simpler. Fewer things on the screen. Options tucked behind a menu instead of sitting out in the open. I looked at it and thought: clean.

It wasn’t simpler. It was simpler for me. I’d spent 20 years in that workflow and I could move through the screen by muscle memory. Our customers couldn’t. What I read as clean, they read as “where did it go?”

Usage didn’t collapse. That’s the part worth paying attention to. It thinned. People still logged in. They just stopped doing the one thing that made the product worth paying for. On a chart, that looks like ordinary noise. It looked like a slow month. Then renewals came up and it looked like something else.

I found out on a phone call, months later, with a customer who’d already left. It took him about 40 seconds to explain it. No dashboard was ever going to hand me that sentence.

When you don’t know why, you guess. And the guess is always a feature.

This is the expensive part.

A gap in your understanding doesn’t stay empty. You fill it. And the guess is almost always feature-shaped, because features are the part you control. “We lost them because we don’t have X.” That’s a comfortable conclusion. It converts a scary unknown into a ticket.

But people rarely leave over a missing feature. They leave because a job they hired your product to do didn’t get finished.

Advisors weren’t buying Pulse360 for AI meeting notes. They were buying the ability to walk out of a client meeting and be done. Notes written. Compliance covered. Nothing hanging over the weekend. That was the job.

Any break in that chain meant the job didn’t get done. An export that needed cleanup. A CRM field that didn’t map. A summary that was 90% right, which meant they had to read all of it to find the 10%. And a product that gets someone most of the way to done isn’t worth most of the price. It’s a step they now have to remember to finish themselves.

Nobody writes that in a cancellation form. They write “too expensive,” which is what people say when the value didn’t land.

A lot of churn happens before your product does anything

We integrated with Salesforce, Redtail, Wealthbox, and Nitrogen (Riskalyze at the time). Some of my worst retention had nothing to do with the core product at all. It was setup.

Field mapping that needed somebody technical. A sync that ran fine but put data somewhere the advisor never looked. Permissions that needed an admin who was out that week. Those customers signed. They were in the $1M. Some of them never really started.

Here’s the problem: in a dashboard, a customer who never got set up looks exactly like a customer who used your product and didn’t like it. Same thin usage. Same cancellation. Completely different causes, and opposite fixes. One needs months of product work. The other needs a 30-minute onboarding call and a better default.

If you’re guessing, you can’t tell them apart. That’s how a founder spends a quarter building something for a problem that was really a checklist.

So go talk to the people who left

Most founders don’t. It’s awkward. It feels like volunteering to be told off. And there’s usually a quiet fear that they’ll confirm something you’d rather not have confirmed.

Do it anyway. Churned customers are the most honest people you will ever talk to. They have nothing to sell you and nothing to protect. They’ll say things a current customer will never say to your face.

The founders who do make these calls often get little out of them, because they ask the wrong question. The standard one is “what could we have done better?” That question asks your customer to be your product manager. Polite people answer politely, and you get a feature request they invented on the spot to fill the silence. You write it down. You build it. Nothing changes.

Don’t ask people to design. Ask them to remember. People are unreliable designers and surprisingly good witnesses.

The questions to actually ask

Every one of these is about a specific moment in the past. That’s the trick — you want a story, not an opinion.

“Take me back to when you first signed up. What was going on that week that made you go looking for something?” You’re after the trigger. Something broke or changed, and that’s the job they were hiring for.
“What were you doing about it before us?” This tells you what you actually replaced. Usually a spreadsheet, an assistant, or nothing at all.
“Walk me through your first week. What did you do first?” This is where setup friction shows up, at the moment their motivation was highest.
“When was the last time it did what you hoped it would?” Locates the high-water mark. Everything after that is the decline.
“What were you doing the last time you opened it?” Often the exact moment it stopped being worth it.
“What do you do now instead?” Your real competitor. Frequently not another vendor.
“Who else had to touch this for it to work?” Surfaces the admin, the IT person, the ops manager you never knew was in the loop.
“Was there a point where you thought about canceling but didn’t? What happened?” The most valuable answer in the whole call. It’s the near miss, and your current customers are living in it right now.
“When you told your team you were canceling, how did you explain it?” How the decision got justified internally, in their words, not yours.

A few rules for the call. Don’t pitch. Don’t defend — if they describe your product wrong, write it down exactly as they said it, because their version is what they acted on. Twenty to thirty minutes is plenty. Do five to ten, not one; one is an anecdote, five is a pattern. Record it so you’re listening instead of typing. And make the calls yourself, at least the first round. A survey is just a dashboard with sentences in it.

When you ask, don’t offer a gift card. Say you’re trying to understand what happened and you’re not going to ask them for anything. Most people say yes. Some of them will be relieved someone finally asked.

What you’ll usually hear

Two things.

First, the same story from several people in different words. That’s your answer, and it’ll be more specific than anything you would have guessed.

Second, it usually isn’t what’s on your roadmap. It’s an unfinished job, a setup step, or a gap between what people expected and what they got. I’m working with a founder right now whose website promises a little more than the product currently delivers. Customers show up expecting the finished job and get most of it. Nobody complains. They just don’t renew. That gap is invisible in analytics and it comes up in the first two minutes of a conversation with someone who left.

Your dashboard is good at the that and the when. Use it for what it’s good at — spotting the drop, sizing it, finding the cohort. Then pick up the phone for the why, because there’s nowhere else to get it.

Ten phone calls will cost you a week. Guessing wrong will cost you a quarter. I know which one I’d choose now.

If you’re watching customers leave and the numbers aren’t telling you why, that’s what I help founders work through — you can read how I approach it on Reduce Your Churn. Happy to talk it over either way.

← All writing