For scaling brands that are able to run at least $10k+/month in paid media.
What customer research looks like before a single ad gets made.
The voice-of-customer system we run at the start of every account. Sources, mining passes, and how a review phrase becomes a hook. Built for subscription DTC brands.
Most ad accounts fail at the input, not the output.
We once reviewed a batch of forty AI-generated ads for a brand. Every one opened with the product. Not one opened with the customer's problem in the customer's words.
The AI was not the problem. It was prompted with a product description, and a product description contains zero customer psychology. Garbage in is not the model's fault. But nobody in the loop had done the research that would let them recognize it as garbage.
VOC research is that missing input. Not a persona workshop. Not a brainstorm. The customer's actual words, collected, clustered, and turned into direction.
Collect from the five sources.
You are not looking for opinions about your brand. You are looking for language: how real people describe the problem, the moment they went looking, and what almost stopped them from buying.
Your own reviews
The richest source. Sort by detailed reviews, not star rating. Five-star reviews tell you the belief that converted. Three-star reviews tell you the objection that almost won.
Competitor reviews
Their one-star and two-star reviews are your angle inventory. Every recurring complaint is a belief you can sell against with proof.
Support and chat logs
Pre-purchase questions reveal the confusion your ads should resolve. If people keep asking the same thing, your creative is not answering it.
Forums and communities
Reddit threads and niche communities show the problem described with nobody selling anything. This is where the trigger moments live: what happened right before they searched.
Post-purchase surveys
One question outperforms most survey stacks: "What was going on in your life that made you look for this?" The answers are ad openings, verbatim.
Mine for five things. Nothing else.
- First-person pain. The problem in their words, not your category language. "I was tired of X" beats any benefit statement you can write.
- Trigger moments. The event that started the search. Ads that open on the trigger meet the customer mid-thought.
- Objections. What almost stopped the purchase. Price framing, skepticism, past failures with similar products.
- Identity phrases. How they describe themselves. People buy things that match who they think they are.
- Comparison language. What they tried before and why it failed. Failed-alternative stories are the strongest cold-traffic openers.
Use AI for the reading. Synthesis of hundreds of reviews is an hour of machine work, not a week of yours. But the extraction targets above are the prompt. Feed it a product description and you get product ads. Feed it these five targets and you get customer psychology.
Transcribe, don't copywrite.
When one phrase keeps repeating across sources, that phrase is your hook. Verbatim. Not polished into marketing language. Transcribed.
This is the method behind the account we took to a $132k month at 5.75x ROAS. The winning angle did not come from a brainstorm. It came from the review pile. One phrase kept repeating in customers' own words, and it became the hook exactly as written.
Proof of method, not of category: that account was a furniture brand, outside our current subscription focus. The mechanism transfers. The phrase your customers keep repeating is your next hook, whatever you sell.
Why verbatim works: your customer scrolls past a thousand lines of ad-speak a day. A sentence that sounds like their own thought is the only pattern interrupt left.
Cluster into beliefs. Then a human chooses.
Group phrases into belief statements
Each cluster becomes one sentence in the form: "This customer will buy when they believe X." That sentence is an angle candidate. A typical mining pass yields a dozen.
Rank by evidence, not by taste
Which belief has the most verbatim quotes behind it? Which appears in competitor complaints but not in competitor ads? Frequency and gap beat cleverness.
Human picks the one the market believes now
This is the step that does not automate. Knowing which of the twelve candidates your market will believe this quarter is judgment. The research narrows the field. A person makes the call and owns it.
Skipping straight from research to volume is how accounts end up testing seven value props at once and learning nothing. One belief per test. Depth beats volume.
Run this before your next brief.
- Pull your most detailed recent reviews and your top competitor's worst reviews
- Extract the five targets: pain, triggers, objections, identity, comparisons
- Write each cluster as a one-sentence belief statement
- Mark which beliefs your current ads sell against. The gaps are your next angles.
- Pick one. Build the next batch on it, hook verbatim.
One afternoon of this replaces a quarter of guessing. What phrase do your customers keep repeating?
Research fixes the input. Now check the output.
Step 1: run your numbers through the calculator. Two minutes for your aMER, your CAC ceiling, and the monthly amount slipping through the gap.
Step 2: want the research done on your account? Request the audit and we start exactly where this page starts.
Run your numbers Request the full audit