Ask your heaviest users for feedback first

Heavy users feedback means finding the people who use a feature most and asking them, personally, to try the next version before everyone else does. Use it when you are rebuilding or extending a feature that already has real usage. Light users cannot tell you much about a feature they barely touch. Heavy users know its rough edges, and they usually care enough to answer.

How to do it

1. Define "heavy" with one event and one number. Pick the event that means the feature was used (saved a report, favorited a song, ran an export) and a threshold over a recent period, such as a set number of uses in 30 days. Start narrow: a list you can email by name beats a list too big to contact.
2. Pull the list from your analytics. In Amplitude, create a cohort with a "performed event" rule, choose Count, set an operator and value, and pick a date range (Amplitude docs). In Mixpanel, build a cohort of users who did the event more than a set number of times, then download the list as a CSV from the Cohort Builder (Mixpanel docs). Or run the same count as a database query.
3. Write to each one personally. Say what you noticed ("you used export every week this month"), what you are changing, and what you want: 20 minutes on a call, or a few written answers. Send it from a founder or product person, not a no-reply address.
4. Give them early access. Let them use the new version before launch, and ask what is missing or broken. Keep a log of what each person said.
5. Close the loop. Tell them what you changed because of their input, and thank them by name.
6. Ask for a word at launch, only if it fits. Some will volunteer a quote or a post. If they get early access or any perk in return for posting, they should disclose it (see below).

For a related way to turn loyal users into proof on your site, see featuring your customers.

Examples

Superhuman used a similar filter when it hunted for product-market fit. As described by First Round Review, it surveyed users who had "used the product at least twice in the last two weeks" and asked how they would feel if they could no longer use it. It studied the "very disappointed" group to see what they loved, then looked at what held back the "somewhat disappointed" users who valued the same thing. It ignored feedback from users who would not be disappointed (First Round Review). That survey covered the whole product rather than one feature, but the logic is the same: listen most to engaged users.

What it looks like in practice

Say you run a music app and are building version 2 of song favorites. You pull the people who favorited the most songs in the last 30 days. You email each one: "You are one of our heaviest users of favorites. We are rebuilding it. Would you try it this week and tell me what is missing?" You book short calls with those who reply, and take written answers from the rest. You ship fixes before launch and write back to each person with what changed.

Limits and when it backfires

• Heavy users are not your next customers. They already love the feature. They can miss problems that block newcomers, so pair this with a few tests from new users.
• They ask for more of what they have. Their requests lean toward advanced options. Weigh them against what drives signups.
• Small lists mislead. A handful of replies is not a vote. Look for patterns across several people before you change direction.
• Disclose perks. The FTC's endorsement guidance says to disclose material connections, including payments or other incentives, when a significant minority of consumers would not expect them (FTC). If early access is exchanged for a public post, tell readers.
• Respect how you got the list. Check that your privacy policy covers using product usage data to contact users, and stop emailing anyone who asks.

Questions people ask

Who counts as a heavy user? The users with the most uses of one feature in a recent period. Define it with one event and a threshold, then adjust it until the list is a size you can contact.

Why ask heavy users before launch? They know how they use the feature today, so they can say whether the new version helps or gets in the way.

Sources

• Amplitude: Define a cohort, read Oct 8, 2026
• Mixpanel: Cohorts, read Oct 8, 2026
• First Round Review: How Superhuman built an engine to find product-market fit, read Oct 8, 2026
• FTC: Endorsement Guides, what people are asking, read Oct 8, 2026