Spotting Recurring Complaints in a Sea of Posts

Two hundred saved posts and no decision. That is the usual end state of a research session. You read for hours, everything feels important, and then you close the tab without a clear answer.

The way out is not more reading. It is looking for recurring complaints: the same problem, described by different people, in different words, over and over. Those repeats are the closest thing to a reliable signal you can get without spending money.

Why Repetition Beats Volume

One detailed post can be fascinating and completely unrepresentative. The author may have an unusual setup, a strange tool stack, or a strong opinion that nobody shares.

A problem that returns across many separate people is different. Each repeat is a small independent confirmation, and independent confirmations are what make a claim hold up.

There is a second reason repetition matters. When a problem keeps coming back over a long stretch of time, it means people have lived with it, tried the obvious fixes, and still not solved it. A problem that survives years of available tools is either hard to solve or genuinely worth solving.

Recurring Is Not the Same as Loud

The loudest complaint in a thread is often the least useful one. Loudness comes from the thread’s mood, not from the problem’s weight.

• Loud, not recurring. A single furious post about a price increase. Everyone agrees in the comments. Nobody has a workflow that depends on it.
• Recurring, not loud. Someone mentions in passing that they rebuild the same report every month. Three other people do the same thing in different threads, without complaint.
• Both. The same workflow described as painful by several people across several months.
• Neither. An interesting idea with no one behind it. Common in discussions about tools people wish existed.

Your best leads usually come from the middle row. Quiet repetition is easy to miss and harder to fake.

How to Group Complaints Without Flattening Them

Grouping is the part that turns a pile of posts into knowledge. Do it on paper or in a text file before you write any summary.

1. Copy every complaint out word for word, with a link. No paraphrasing yet.
2. Write a short label next to each one. Keep the label close to the words people used. A label like “chasing approvals in a shared inbox” carries more information than “inbox problems”.
3. Merge labels that describe the same task, not the same feeling. “Frustrating” is not a category. “Exporting from the old system” is.
4. Keep the variation. If one person needs this weekly and another needs it daily, note both. The difference often shows you the real buyer.
5. Count distinct people, not posts. One person commenting five times is one person.
6. Sort the groups by how many separate people appear in each.

Six groups from two hundred posts is a good outcome. If you end up with forty groups, your labels are too specific and you should widen them.

A Grouping Pass You Can Run Today

Pick a topic you already care about. Say it is scheduling shifts for small cafés. These searches surface complaints rather than vendor pages:

• “making the rota” pain
• “schedule keeps breaking” staff
• “how do you plan shifts” spreadsheet
• “no one wants the closing shift” scheduling
• “staff swapped shifts and no one told me”

Then run the pass:

1. Save every post that describes a situation, not just an opinion. Aim for forty or so.
2. Search inside your saved text for phrases like “every week”, “takes hours”, “by hand”, “I still”, “we gave up on”. These words usually sit next to the real problem.
3. Build your groups with the six steps above.
4. For the top three groups, note the workaround each person uses. The workarounds are your competition.
5. Write one sentence per group that answers: why has this not been solved yet?

The last question is the one that separates a real opportunity from a wish. If the answer is “the tools exist and people use them happily”, you have learned something valuable.

What Recurring Complaints Cannot Tell You

Repetition tells you the problem is shared. It does not tell you the problem is worth paying for, and it does not tell you how many people exist outside the communities you read.

Two things stay invisible in a manual pass:

• Who is not posting. People who are not in these communities have the same problem sometimes, and you will not see them at all.
• What people already pay for. A complaint can be completely real and still not change anyone’s spending.

Treat repetition as the strongest signal you have, then check it against what people already buy. Recurring complaint plus existing spend is a much better combination than either one alone.

FAQ

How many repeats make a complaint recurring?

There is no threshold that applies to every topic. What matters is that the repeats come from separate people and separate threads, and that they describe the same underlying task rather than the same emotional reaction.

Should I count comments as separate complaints?

Only if a different person wrote them. Replies that say “same here” are useful confirmation, but they are one voice unless the reply adds its own detail.

What if a complaint repeats only in one community?

That is still a lead, but a weaker one. Try the same search in another community before you assume the problem is broader than the room you found it in.

How do I know when to stop grouping?

When new posts keep landing in groups you already have. That is the point where more reading adds detail but no new direction.

---

If grouping two hundred posts by hand sounds tedious, that is the part the research tool automates: it pulls discussions from Reddit’s public archive, Hacker News, Stack Overflow, and GitHub, merges near-duplicates into clusters, reports how many distinct people each problem involves, and keeps only quotes it can verify against the source text. A report takes two to four minutes and costs 2.99 EUR once. Related reading: Counter Evidence: The Step Most Founders Skip and How to Score Pain Intensity in Customer Research.