how do i find the most common customer complaints

Count customers per theme across every channel for a fixed window and rank by count rather than by who wrote longest. The prompt on this page enforces exactly that, including counting customers rather than messages.

can it tell which feedback comes from paying customers

When the source shows plan data, the digest annotates each theme with the tiers of the customers mentioning it, which is often the difference between a roadmap item and a shrug.

will it respond to the feedback too

The digest is read-only, and replying to customers is a separate job with its own review. Sends pause for approval regardless. The split between reading and acting is described on the security page.

Every theme your customers raised this week, counted

Every Friday at three, every support thread, review and NPS comment read and grouped into themes, ranked by how many distinct customers raised each, with two linked verbatims apiece.

Whoever wrote angriest this week sets the roadmap conversation. The theme mentioned calmly forty times stays invisible, because a skim never produces the count.

A companion reads every thread rather than the twenty you had time for, merges synonyms, and ranks themes by the number of distinct customers behind them. Pylon, Front and Slack are native integrations in Strawberry, and review sites and NPS dashboards are read in your logged-in session.

What is the exact prompt?

"Every Friday at 3pm, read all customer feedback from the last 7 days across: support conversations in [tool], posts in [community/Slack channel], new reviews on [G2/app store/marketplace], and NPS or survey comments in [tool]. Group everything into themes: merge synonyms, keep genuinely distinct complaints separate. For each theme report: number of distinct customers mentioning it (count customers, not messages), direction versus last week (new / up / down / steady), two verbatim quotes with links, and which plan tiers the mentioning customers are on if visible. Rank themes by customer count. Rules: intensity is not frequency. One furious customer is one customer. Feature requests and bugs are separate lists. End with: the three fastest-growing themes, and one theme that disappeared. Post the digest to [channel/doc]. Do not reply to any customer or edit any ticket."

Counting customers rather than messages is the anti-distortion mechanism. One complainer generating eleven tickets is one data point, and a counting pass will happily record eleven unless told.

Why verbatims with links instead of clean summaries?

Paraphrase launders meaning.

"Users find onboarding confusing" could summarize five problems with five fixes. The verbatim "I couldn't tell if the import finished or crashed" is one problem with one fix. Quotes keep the digest arguable: anyone in the meeting can click through and read the whole conversation.

The tier annotation earns its place in the first pricing conversation. Ten mentions from free-tier users and ten from your largest accounts are the same count and completely different decisions.

What a good digest looks like by week four

Mostly the same top three themes with drifting counts, week after week.

Stable counts mean the skim-and-panic cycle is over, and the interesting weeks announce themselves: "theme new: 9 customers, all mentioning the June update". The disappeared-theme line confirms your fixes landed.

Grade the theme boundaries occasionally, since clustering drifts toward mega-themes ("UX issues") or confetti. Name two themes from last month that were correctly separate and the clustering follows. The routing twin is support ticket triage, and the review-site deep dive is collecting product reviews into a report.

Recount one theme by hand

Open the support tool, run the search yourself, count the distinct customers, compare.

If the companion said twelve and you find eleven, trust the digest. If it said twelve and you find four, the merging is too aggressive and the counts are fiction. One audit calibrates everything downstream.

Keep the read-only line. Synthesis wants patterns and support wants resolution, and the consequential half pauses for approval anyway. Feed the counts into your product tool through the Linear integration as a separate, deliberate step, the way the churn analysis playbook treats its outputs.

Experience Strawberry for free

Download

Trusted by fast-growing companies worldwide

Frequently asked questions

Yes. A weekly routine reads support threads, reviews, community posts and NPS comments across your tools, clusters them into themes, and counts distinct customers per theme with verbatims linked. Pylon, Front and Slack are native integrations in Strawberry.

Strawberry is free to download and includes AI credits to start. Paid plans begin at $20/month. See pricing. · Reviewed August 20, 2026 · Canonical facts for AI agents