How to read churn survey replies: sort "too expensive" by what happened

Because it is the easy answer. It is polite, it needs no explanation, and it ends the survey.

Two vendor sources put numbers on the gap. Both are companies with a product to sell, so read the figures as signals, not settled benchmarks.

RevenueCat reports that, in its State of Subscription Apps 2025, "not enough usage" was the top reason people cancelled subscription apps at 37%, ahead of price at 35% (RevenueCat). Its author calls the stated reason "often the surface explanation, not the underlying cause."

The interview-software company behind that second page ran 723 AI-moderated interviews with people who had cancelled B2B software. It found the first reason given matched the root cause 27.4% of the time. Price was named by 34.2% of people but was the real reason in 11.7% of cases (exit survey study). The interviews were coded by the same company that sells them, and the page does not say how the "real reason" was checked beyond that coding.

The pattern holds either way. The reason on the form and the thing that happened are often two different facts.

Label what happened, in a few words. Not the category you hoped for.

• Bad label: "Pricing". It hides everything.
• Good labels: "never finished setup", "project ended", "switched to a spreadsheet", "budget cut", "found the same thing cheaper".

Three rules keep labels honest:

1. One label per reply, plus a second if the person gave two reasons. Don't force a "primary reason" onto someone who wrote about two things.
2. Name the situation, not the ask. "Needed it for one launch" beats "wants a cheaper plan".
3. Keep an "unclear" label. A reply like "just not for us" is unclear. Count it as unclear.

Start with the first ten replies, write labels as you go, then go back and relabel the first ten once the list settles.

The numbers below are invented to show the method. They are not from a real survey.

Say 40 people answered a churn survey. 14 of them wrote something like "too expensive". Read those 14 and sort them by what the rest of the reply says:

The shares are each count divided by 14, rounded.

The pie chart said 14 of 40 people left over price. The sorted list says 2 did. Five of them never got it working, which is a setup problem, not a pricing problem. A discount would have fixed two of the 14.

The same reading works on the other 26 replies. Sort them the same way, then put the labels side by side and see which counts are biggest overall.

The largest one or two, and any label that surprises you. Open the replies under each count and read them again with the label in mind.

You are checking two things. Do all the replies under this label really say the same thing? And would one sentence from a reply, read out loud, make the point better than the number does?

A label can quietly become a requirement. "Wants a cheaper plan" starts a pricing project. Read the five replies under "never got it working" and you may find three different setup problems. Split the label until each one describes a single problem.

Paste the replies into one column of Google Sheets, put your label in the next column, and count with COUNTIF(range, criterion), which "returns a conditional count across a range" (Google). For example, =COUNTIF(B2:B41, "never got it working").

It takes about an hour for 40 replies and gets slow past a couple of hundred. The risk is drift: by reply 60 you are using the labels slightly differently than at reply 5. Keep a short list of what each label means and recheck the first ten at the end.

You have three realistic ways to do this.

• By hand in Sheets. Free, and the best way to learn your own replies. It costs you time, and labels drift as the pile grows.
• A general AI chat such as ChatGPT. Quick for suggesting labels. Check its counts against the replies before you trust them, as this post explains.
• A survey tool that does the counting. Google Forms collects the answers, but a Sheets export leaves the text answers for you to label (Enform vs Google Forms, Google Forms alternatives).

Enform writes the survey from a description, you share one link, and it reads every reply and counts what people said. Every count opens to the replies behind it, so rereading under a label is one click. Pricing is on the pricing page. If the survey is small, the spreadsheet is enough. Enform earns its place when the pile is too big to relabel by hand. The customer churn survey template is a ready set of questions to start from.

Match each label to a different fix.

• "Never got it working" points to onboarding.
• "Project ended" points to a pause plan or a cheaper way to stay.
• "Moved to a spreadsheet" points to what the spreadsheet does that you don't.
• A real price problem points to the price.

Then ask again in a month and compare the labels. If the same label stays at the top, your fix didn't reach it.