Your customers already wrote your best ads.
How we find out what buyers want, count it, check it against what competitors already promise, and turn it into one line every ad and page can hang off.
Written by Pedram Ghozat, founder of MediaBoostr, and PedramAI · Updated
Why most research doesn’t change the ads
Every brand wants ads that sound like the customer wrote them. Most of the research behind them is someone reading ten reviews on a Friday.
Research only pays off when it changes what the ads say. That needs counting and a competitor check, not more reading.
Count the words before you write any
Reading reviews tells you what one person felt. Counting them tells you what most buyers feel, in the words they use. Those words go into the ads unchanged, typos and slang included.
Order matters too. We start with 4-star reviews: people who like the product and still say what almost stopped them. Five-star reviews come later, because “love it” gives you nothing to write with.
Then we hold the counted problems up against every promise competitors already make. A problem many buyers have and no rival owns is where the positioning goes. The rest become support lines.
New brand with few reviews? Start with competitor 1 to 3-star reviews and category reviews on marketplaces.
Six steps from reviews to a positioning line
The research we run before we decide what a brand stands for. It ends with the problems we lead with. Turning them into ads comes next.
Collect customer words in order
Own 4-star reviews first, then 1 to 3-star, then competitor 1 to 3-star, ad comments and forums. Support tickets and return reasons if the brand shares them. AI helps sort and search real quotes. It never writes them.
Count, don’t skim
Tag every quote: the problem, the belief behind it, what made them buy, what they tried before, how they describe themselves. Then count each tag. Keep the original wording. “Bloated by 4pm” sells. “Digestive discomfort” doesn’t.
Find what they tried before
Pull every product, habit or workaround they used first, what it cost them and why they stopped. Most cold buyers have tried something and been let down, so they arrive sceptical. That doubt is often the best opening line.
Find the moment they bought
The event or bad day that made them search this week. The same problem with a different trigger is a different buyer. A gift buyer’s trigger is a date on the calendar, so they count as their own group.
List every promise competitors make
Open the Meta Ad Library for 3 to 5 competitors and write down every promise their live ads make, with the ad ID. Put your counted problems next to that list. A promise three rivals already make can support an ad. It can’t lead one.
Write one positioning line
One sentence: who it is for, the problem in their words, and why this product. Then pick the 2 to 3 problems the ads will lead with. If a rival could sign the line, it doesn’t position you.
How the counted problems become personas, angles and ad sets is on creative strategy. Where to find buyers’ words in depth is in our brand and customer research guide.
What counting shows that reading misses
A made-up example: a greens powder brand, 160 quotes from five kinds of source. The bars show how many quotes mention each problem.
Price has two sources, so it stays a note for now. Mixing has one, so it is a note too. Brand and numbers are made up.
The source mix: 60 own 4-star, 40 own 1 to 3-star, 35 competitor 1 to 3-star, 15 ad comments and 10 forum posts. Example numbers.
The biggest problem isn’t always the one to lead with
The same example, next to what five competitors already promise in their live ads.
Bloating is the bigger problem, but four rivals own it. Taste has enough sources and nobody talks about it, so the ads lead there.
Brands whose ads sound like every competitor
If your ads could run under a rival’s logo without anyone noticing, research and positioning are where we start.
Questions
What is customer research for ecommerce?
Collecting what real buyers say about your product and your competitors, in their own words, then counting which problems, doubts and buying moments come up most. It decides what the ads and pages say.
How many reviews do I need?
We aim for 150 or more quotes for a full piece of research. Under 100 we treat the findings as thin. A new brand can make up the number with competitor reviews and category reviews on marketplaces.
Why start with 4-star reviews?
A 4-star buyer likes the product and still tells you what nearly stopped them. That is the objection your ads need to answer. Five-star reviews are mostly praise, which is good proof but gives you few words to use.
Can AI do customer research?
AI is good at sorting and searching hundreds of real quotes. Asked to describe your customers without real data, it returns generic answers. We only let it work on quotes we collected, and every quote keeps its source and date.
What is a positioning line?
One sentence that says who the product is for, the problem in their words, and why this product and not another. If a competitor could say the same sentence, it does not position you yet.
How is this different from personas?
Research is the raw material. Personas are what we build from it: one buyer defined by a problem, a desire and the moment they buy. Each persona then gets its own angles and ads.