How to Get Your First 100 Pieces of Customer Feedback (Without a Dedicated PM)

Most early-stage founders treat feedback collection like something you get to once the product is more stable. That's backwards. Before product-market fit, feedback is how you find out what to build. Waiting until you have a bigger user base means building longer in the wrong direction.

You also don't need a dedicated PM or a formal feedback program to get started. You need four sources and about 90 minutes a week.

source 1: your support inbox

Every email someone sends asking for help is feedback. If they're confused about how something works, that's a UX problem. If they're asking for a feature, that's a gap in the product. If they're reporting a bug, that's a bug.

The simplest system: create a folder or label called "feedback" and move support emails there as you respond. At the end of the week, skim it. You don't need to write anything down — just notice what shows up more than once.

If you want to go one step further, set up email forwarding so those threads flow into a dedicated tool automatically. You get a searchable archive instead of a folder you'll eventually lose track of.

source 2: onboarding calls

If you're doing onboarding calls (and you should be, at least for your first 50 customers), they're your best feedback source. Users are fresh, they just tried to do something, and they'll tell you exactly where they got stuck.

Before you hang up, ask three questions:

1. Was there anything that took longer than you expected to figure out?
2. Was there anything you expected to be able to do that you couldn't?
3. Is there one thing that would make this significantly more useful for you?

Write the answers down. Or just let the call recording run. The key is asking consistently — not just when you remember to.

Ten onboarding calls with three questions each gives you 30 data points. Patterns start to show after five or six calls.

source 3: churn conversations

People who cancel are your most honest critics. They've already made their decision, so they have nothing to protect. A short call after cancellation ("I'd love to understand what drove you to cancel, purely so we can improve") often produces the clearest signal you'll ever get.

Not everyone will agree to a call. But 10-20% usually will, especially if you ask like a person and not a survey. Five honest exit conversations are worth more than 50 NPS responses.

source 4: a two-question exit survey

For users who won't do a call, a two-question email when they cancel works better than a long form:

1. What made you decide to cancel?
2. What would have made you stay?

Open text, no multiple choice. The answers will surprise you.

If you're using Stripe, automate this with a webhook. If not, a manual email the day after cancellation works fine.

the synthesis problem at small scale

Even 20 pieces of feedback sitting in a doc get ignored.

The collection step is easy to do once. The synthesis step — reading everything, grouping similar items, finding the two or three themes that matter this week — is the one that requires discipline. Most people skip it.

At 20 pieces of feedback, you can do this in 30 minutes. Group the items by theme. Count how many hit each theme. Write two sentences about what it means. That's your synthesis.

Do this every week before your planning session, not after. The planning session should be a decision meeting, not a discovery meeting.

the actual goal

The goal isn't 100 pieces of feedback. It's 100 pieces you read and acted on. Feedback you collect and never synthesize is noise that makes you feel busy without making you better.

A spreadsheet with 100 rows doesn't help you decide what to build next. A weekly 20-minute read-through of 15 items, with a summary at the top, does.

At some point this process doesn't scale. Reading everything manually breaks down around 50-100 new items per week. That's when automated synthesis starts to pay off: grouping duplicates, surfacing patterns, delivering a weekly brief without the manual work.

That's a future problem. For now: four sources, 90 minutes a week, and actually read what comes in.

---

Triagly automates the synthesis step once you've outgrown the manual process. Try it free →