Market research before targeted ads: 5 steps so you don't burn the budget
The usual launch looks like this: a few audiences, a few creatives, a budget, and the hope that the algorithm will find the buyers. It often does find people, but it can only optimise the message you gave it. If that message talks about a problem your buyers do not feel strongly, the budget is spent learning that, one ad set at a time.
Why blind testing is expensive
Ad platforms are good at finding people who react to an ad. They are not good at telling you which problem your product should be sold on. That knowledge sits elsewhere: in what your customers wrote after buying, in what your previous campaigns already showed, and in what the wider audience says when you ask them directly.
Research before the launch moves most of the learning to these cheaper places. Instead of five random angles, you go to the auction with two or three pains that real customers named, phrased the way they phrase them. Testing does not disappear. It becomes smaller, and the hypotheses are better.
The five steps
- Learn the product in detail: what exactly is sold, to whom, in what price range, how it is delivered, which alternatives the buyer compares it with and which questions people ask before buying.
- Analyse previous launches: audiences, offers, creatives, landing pages, cost per lead and per sale, comments under the ads and the conclusions the team drew. The point is not to pay twice for the same mistake.
- Collect customer reviews, written, audio or video, and pull out the pains: the situation before, what the person tried, what changed after. Keep the exact quotes next to each pain.
- Turn the pains into a short survey and ask the audience, through a quiz link or by phone, which problem is the most acute for them. Rank the pains by the answers.
- Launch ads on the top pains, in the customers' own words, with one pain per ad set so the results stay readable.
Steps 1 and 2: the product and its history
Write a product card of one or two pages: what you sell, who buys it, the price range, the main alternatives, the promise and the proof behind it. Talk not only to the owner but also to whoever answers customer questions every day. They know which doubts come up before a purchase and which complaints come after it.
For past campaigns, collect screenshots of results and creatives, exports of ad sets and the story behind them. Look for patterns rather than single numbers. Did leads look cheap but rarely buy? Did one angle work on warm audiences and fail on cold ones? Did people ask the same question in the comments again and again? Each of these is a lesson that someone has already paid for.
Step 3: reviews are a source of pains, not of stars
A rating says the customer is satisfied. A sentence says why. In each review, look for three things: the situation before (what was wrong), the trigger (why they started looking) and the result (what changed). Label the pain in a few words and keep the verbatim quote next to it. The quotes are the raw material for ads later.
Video and audio reviews are often the richest, because people speak more freely than they write. Transcribe them first, then read the transcript the same way. Reviews from marketplaces, Google Maps, social media comments and support chats all count, as long as they come from real buyers.
Step 4: check the pains with the audience
Reviews come from people who bought and cared enough to write. A survey widens the view to people who look like your buyers but have not bought yet. Keep it short: the list of pains from the reviews, written in plain words, and one task, to choose what matters most.
Do not use a rating scale where every pain gets a five. Ask people to choose: the more important of two pains, or the most and the least important in a small set. This method is called best–worst scaling, or MaxDiff, and it gives a clear order even when everything sounds important. A quiz link works for people who came from ads or a mailing list; a phone call works for existing customers and B2B buyers who will not fill in a form.
Step 5: launch on the most common pains
Build one ad set per top pain, with headlines and texts that reuse the customers' words. Give most of the budget to the pain that ranked first, but keep a smaller share for the second and third, because the survey shows what people say and the ads show what they do. Compare the ad sets by cost per qualified lead or per sale, not by clicks.
Worked example (illustrative numbers)
An online accounting service for small businesses plans its first paid campaign. The team reads 40 reviews and finds six pains. Three appear most often: fear of fines for late reports (14 reviews), evenings lost to paperwork (11 reviews) and a previous accountant who did not answer (8 reviews). A low price is mentioned in only two reviews, although it was going to be the main message.
A best–worst survey of 150 respondents puts fear of fines first by a clear margin, slow answers second and paperwork third. The team launches three ad sets and splits the budget 60/25/15. The headline of the first one is almost a quote: I stopped waking up at night thinking about the tax deadline. After two weeks the team compares cost per qualified lead across the three ad sets and moves budget accordingly.
How it works in OrgLines
In OrgLines this process is the Market research add-on module, and our team runs it inside your cabinet. The project holds the product card, files and screenshots of past campaigns and competitors added manually or by link. Reviews go in as text, audio or video, recordings are transcribed automatically, and the AI picks out pains together with verbatim quotes. Each stage can be extended when new data arrives: the AI processes only the new material and keeps manual edits.
The survey goes out as a quiz link with pairwise or best–worst ranking, or by phone with the question about the most acute problem and a log of every call. You receive the finished report with the ranked pains, without names, phone numbers or voice recordings, and the top pains can go straight to Creative Studio for ad creatives. Terms are on the pricing page: https://orglines.com/en/pricing/.
Common mistakes
- Starting with creatives and audiences before anyone has read what customers actually wrote.
- Ignoring past campaigns because they were run by another person or agency.
- Counting stars instead of reading the text of the reviews.
- Rewriting customer quotes into polished marketing language and losing what made them convincing.
- Asking the audience to rate each pain on a scale, so every pain looks equally important.
- Putting several pains into one ad, so the results cannot tell you which one worked.
- Treating the survey ranking as final and never checking it against real ad results.
Checklist
- A product card exists and the people who talk to customers have read it.
- Past campaigns are collected with results, creatives and the team's conclusions.
- At least a few dozen real reviews are read, including audio and video.
- Every pain has a short label and at least one verbatim quote.
- The survey asks people to choose between pains, not to rate them.
- Each top pain has its own ad set and its own measure of success.
- There is a date to compare survey ranking with ad results.