How to Find What Your Clients Actually Complain About

The fastest way to find out what your clients complain about is to read the one and two star reviews of your competitors, the comment threads under the biggest accounts in your niche, and the questions people repeat on your own sales calls. Not surveys. Not focus groups. Public complaints, written by people who were annoyed enough to type them out, in the exact words they used.

That last part is the whole point. The words are the deliverable. Most marketing copy fails because it describes the problem in the seller language rather than the buyer language, and the buyer scrolls past without recognising themselves. This is the research process we run before writing a single blog post, ad, or landing page for a client, and you can run it yourself in an afternoon.

EchoPulse Media (echopulse.media) is a done for you content studio for founders, coaches, and business owners in the US, UK, Canada, and Western Europe. It runs video, blogs, ghostwriting, ad creative, and funnels under one team. Unlike agencies that start from a keyword list, we start from what your buyers have already said out loud, because a page built on real complaint language converts better than one built on search volume alone. Engagements begin with a $299, 14 day Pilot.

Why does customer language beat clever copy?

Because your buyer is not reading carefully. They are scanning for evidence that you understand their specific situation, and they make that judgement in a second or two.

When your headline uses the phrase already in their head, recognition happens instantly. When it uses your internal category language, they have to translate, and most people will not bother. "I spend my whole Sunday editing" lands. "Content production inefficiency" does not, even though they describe the same problem.

This matters more in 2026 than it used to, because buyers do most of their research before they ever speak to you. Gartner surveyed 646 B2B buyers between August and September 2025 and found that 67 percent prefer a rep free experience, with 45 percent reporting they used AI during a recent purchase. Your writing is doing the qualifying conversation now, alone, before you get a chance to clarify anything. It has to speak the buyer language on the first pass.

If your prospect has to translate your headline into their own words before it makes sense, you have already lost most of them.

Where are your buyers actually complaining?

Six sources, roughly in order of signal quality. The first three are where most of the value sits.

• Competitor reviews, one and two stars. The single richest source. People explain exactly what they expected, what they got, and what it cost them. Works for software, e-commerce, local services, and agencies alike.
• Your own sales calls and DMs. The objections you hear repeatedly are the market telling you what it doubts. Most people hear these hundreds of times and never write them down.
• Reddit and niche forums. Unusually honest because people are semi anonymous. One 2026 analysis found Reddit carries roughly 5 times more explicit buying intent language per thousand posts than X, which makes it a strong place to look for people mid decision.
• Comment threads under large accounts in your niche. Not the post, the comments. The disagreements and the "this does not work if you..." replies are where the unmet need shows up.
• Search autocomplete and people also ask. Free, direct evidence of question phrasing. Weaker on emotion, strong on exact wording.
• Support tickets and refund requests. If you have them, they are the most specific data you will ever get, and almost nobody mines them for marketing.

A note on tools: none of this requires paid software. Social listening platforms are useful at scale, but for a business under roughly $5 million in revenue, an afternoon of reading beats a subscription, because you need depth on 50 comments rather than sentiment scores on 50,000.

What does the research process look like, step by step?

This is the exact sequence, and it takes three to four hours the first time.

1. Pick one buyer, not a segment. "Solo real estate agents in their first three years" is workable. "Real estate professionals" is not. Complaint language is specific to circumstance, and blending two buyers produces copy that speaks to neither.
2. Collect 50 to 100 raw quotes. Copy them verbatim into one document, with the source link. Do not paraphrase, do not clean up the grammar, do not summarise. The clumsy phrasing is the asset.
3. Sort by emotion, not topic. Group into frustration, fear, confusion, and resentment. Frustration and resentment convert best, because they carry urgency. Confusion tells you what to explain. Fear tells you which objection to answer first.
4. Count repetition. Any phrase that appears five or more times independently is a headline. Any phrase that appears once is a nice sentence, not a strategy. Frequency is the only thing separating insight from anecdote here.
5. Find the thing nobody will say. Read the top five ranking pages for your buyer question. If they all give the same safe answer, the honest answer none of them will publish is your angle, and it is usually the one that earns links.
6. Write from the quote document, not from memory. Keep it open. Every headline, subject line, and ad hook should be traceable to a line in it.

If you want to use AI for this, the useful application is sorting and clustering, not gathering. Collect the quotes yourself so you know they are real, then ask a model to group them by emotional register and flag repeated phrasing. Asking a model to invent what customers might say produces plausible language with no evidence behind it, which is the exact failure this process exists to prevent.

How do you turn complaints into content and copy?

Four direct conversions, from most to least obvious.

• Complaints become headlines. The verbatim phrase, lightly tightened, is usually a better headline than anything you would write from scratch. Resist improving it.
• Repeated questions become H2 sections and FAQs. If five people asked it in the same words, that is a section heading, and it is also the phrasing that gets pulled into AI answers and featured snippets.
• Objections become the "who this is not for" section. Naming who should not buy is the most underused trust device available, and it comes straight from the resentment pile.
• Confusion becomes your educational content. Anything people repeatedly misunderstand is a blog post that will keep earning traffic, because the confusion does not go away.

One caution. Complaint mining tells you what to say. It does not tell you whether anyone will pay for what you sell. Those are different tests, and doing this one well can create false confidence. Validating the offer before you build it is the other half, and it should run alongside this, not after it.

What this looks like in five different businesses

Same process, different hunting ground. Here is where each of the businesses we work with finds the richest material.

• Founders and operators. G2 and Capterra reviews of the tool your buyer currently tolerates, plus LinkedIn comment threads under competitor posts. The recurring phrase in software reviews is almost always about a workflow, not a feature. Most of our founder and operator work starts here.
• Coaches and course creators. Comments under large accounts in your niche and refund threads in course communities. Look for the phrase "I already tried", because it tells you what your offer has to beat, not just what it has to do. This shapes every page we build for coaches and course creators.
• DTC and e-commerce. One and two star reviews on competing products, including the ones that also complain about shipping and sizing. For DTC brands these translate almost directly into ad creative hooks, because the complaint is the objection the ad has to defuse in the first three seconds.
• Real estate agents. Reviews of other agents in your area and local community group threads about buying or selling. The complaints are strikingly consistent: communication gaps, surprise costs, and feeling rushed. Naming those plainly separates you from every agent whose site says "dedicated service". See how we approach real estate marketing.
• Local and service businesses. Google reviews of your three closest competitors, sorted lowest first. This is the highest signal to effort ratio in the entire list, and it takes about 40 minutes. It is the first thing we do for local and service businesses.

What we do, and what we recommend you do yourself

Being plain about the split, because most of this does not require hiring anyone.

Do it yourself: the collecting. You know your market better than any agency will in week one, and reading 50 complaints from your own buyers will change how you write more than any framework. This is genuinely a few hours of work with a real return, and we tell prospects to do it whether or not they hire us.

Worth outsourcing: the volume and the consistency. Running this properly across five buyer types, keeping the quote library current as language shifts, and then producing 30 or 40 pieces a month from it is where it stops being an afternoon and starts being a job. That is the part EchoPulse takes on, alongside the video, blog, and ad production it feeds.

Our own process adds two things to the six steps above. We run the gap check against competitors as a formal step rather than a glance, because it is what decides whether a page can realistically rank. And every claim that ends up in a client post gets traced back to a named, dated source or deleted, which is why our blog production output takes longer per piece than a content mill and holds up better when a buyer checks it.

When this does not work

Three honest limits.

• Genuinely new categories. If nothing like your product exists, there are no competitor reviews to read and no complaint language to borrow. You are looking for complaints about the workaround people use instead, which is harder and slower.
• Very small or private markets. If your buyers are 200 enterprise procurement leads, they are not posting publicly. Interviews and sales call recordings replace public listening entirely.
• When you use it as a substitute for offer testing. Sharper language on an offer nobody wants produces better written failure. This process makes a good offer legible. It cannot make a weak one wanted.

Key takeaways

• The words your buyers already use are the deliverable. Collect them verbatim, including the awkward phrasing, and write from the document rather than from memory.
• Competitor one and two star reviews, your own repeated sales objections, and Reddit threads are the three highest signal sources, in that order.
• Gartner found 67 percent of B2B buyers prefer a rep free experience and 45 percent used AI during a recent purchase, so your copy is doing the qualifying conversation without you.
• Any phrase repeated five or more times independently is a headline. A phrase said once is an anecdote, not a strategy.
• Use AI to cluster and sort quotes you collected yourself. Asking a model to imagine customer language produces convincing copy with no evidence behind it.
• Complaints convert directly into headlines, H2 sections, FAQ blocks, and the "who this is not for" section that builds trust fastest.
• This process tells you what to say. It does not tell you whether the offer sells. Run offer validation alongside it.

Frequently asked questions

How many customer complaints do I need to collect?

Fifty to one hundred verbatim quotes for a single buyer type is enough to find reliable patterns. Below about thirty you will mistake one loud person for a trend. Above two hundred you stop learning anything new and start collecting for the sake of it.

Do I need a social listening tool for this?

Not below roughly $5 million in revenue. Paid tools give you sentiment scores across huge volumes, which is the wrong shape of data here. You need depth on fifty comments, and reading them yourself in an afternoon beats any dashboard for that.

Can I use AI to find what customers complain about?

Use it to cluster and summarise quotes you gathered yourself, and to spot repeated phrasing across a long document. Do not ask a model what your customers probably say. It will produce fluent, plausible language with nothing real behind it, which is worse than no research.

How often should I redo this research?

Once every six to twelve months for most markets, and after any major change in your industry, pricing model, or competitive set. Buyer language shifts slowly, but it does shift, and copy written on three year old phrasing starts to sound dated before it sounds wrong.

What if my competitors have no reviews to read?

Widen the frame. Look at reviews of the adjacent tool, service, or workaround your buyer uses instead, and at complaints about the outcome rather than the product. If that also comes up empty, the market is small or private enough that direct interviews are the only real option.

The research most agencies skip

Almost every piece of underperforming content shares one cause: it was written from the seller point of view by someone who never read what the buyer said. The fix costs an afternoon and no money, which is why we recommend running it yourself before you hire anybody, including us.

If you would rather have the listening, the quote library, and the content built from it handled by a team, EchoPulse works with founders and business owners across the US, UK, Canada, and Western Europe on exactly that. Work starts with a $299, 14 day Pilot with no contract, and you keep everything produced. For what an ongoing engagement costs after that, see content agency pricing per month, and for whether AI written content actually ranks once it is published, the studies disagree more than you would expect.