Most product research is not research. It is browsing.
You scroll Etsy, something looks nice, you get a feeling about it, you make a version. Sometimes it works. Usually it does not, and you have no idea why either way, so you cannot repeat it.
Here is a method that is actually repeatable. It is not exciting. It is mostly about ignoring things.
The core principle
Look at what has sold, not at what looks good.
That sounds obvious and almost nobody does it, because what has sold is often boring and what looks good is fun to look at.
A design that has been quietly selling 40 units a month for two years is worth ten designs that went viral on Pinterest last week. The first one is a proven, durable demand pattern. The second is a moment that has already passed by the time you can act on it.
Boring and proven beats exciting and unproven. Every time.
Step 1: Start from an identity, not a product
Do not start with "I want to make mugs." Start with "who am I selling to."
Pick a narrow identity — narrow enough that it feels slightly absurd. *Night shift nurses. Bernese Mountain Dog owners. People who have just retired from teaching. Sober people at ten years.*
Then ask the question that decides everything: what are these people actually going to buy?
If the answer is "a shirt, and I suppose a tumbler", stop. That is not a niche, it is a product. You want an identity that justifies eight or ten different things.
Step 2: Check the evidence
Now go and look at what has genuinely sold in that space.
Three sources, roughly in order of usefulness:
Historical sales data. What actually moved units, ideally over multiple years so you can see whether it is durable or a spike. This is by far the best source and the hardest to get, which is exactly why we built the ProfitBusters archive — around nine thousand proven winners going back several years, filterable by niche and category, sortable by date. One free winner a day publicly, ten a day for members.
Etsy's own search results, read properly. Search your phrase. Look at listings with high review counts relative to how long the shop has existed. Reviews are a rough proxy for sales — usually somewhere between 1 and 5 percent of buyers leave one, so 200 reviews is a lot of units.
Etsy's autocomplete. Type your niche into the search bar and look at what it suggests. Those are real searches, ranked by volume, given to you for free. It is the cheapest keyword research available and people ignore it because it looks too simple.
Step 3: Read the signals properly
Not everything that looks like demand is demand. Here is how to tell.
Real signals:
- High review count on a shop that has not been open long — that is velocity, not accumulation
- The same product concept appearing across several *different* shops, all with sales — means there is a market, not just one lucky listing
- Steady sales across multiple years rather than one seasonal spike
- Recent reviews. A listing with 400 reviews where the last one was 14 months ago is a dead product with a good history
Fake signals:
- High favourites, no sales. Favourites are free. People favourite things they are never going to buy, constantly. It is the single most misleading number on Etsy
- One shop doing well and nobody else. Could be a great product, could be that they have an audience from somewhere else entirely
- Trend articles. By publication, it is over
- Something that looks beautiful. Your taste is not the market
That favourites one deserves emphasis because it fools people constantly. Favourites measure *appeal*. Sales measure *demand*. They are not the same thing and the gap between them is where a lot of wasted weekends live.
Step 4: Check you can actually compete
You have found something with real demand. Now, can you get in?
- Search results count. Under 5,000 is genuinely interesting. Over 50,000 and you need a real angle
- Who is on page one. Big shops with 20,000 reviews, or a patchy mix? Patchy is your opening
- Price range. If everything is $18, there is no margin. If there is a spread from $22 to $45, there is room to position
- Quality of what is there. If page one is mediocre, that is the best signal on this list. Bad competition is an opportunity that shows up nowhere in the data
Step 5: Test small, read the result
Do not make forty designs. Make three to five for one identity, list them, and leave them alone for six to eight weeks.
Then read it honestly:
- Impressions but no clicks → the concept is being found but the first image is not selling it. Photo problem, not product problem
- Clicks but no sales → price, delivery estimate, reviews, or the listing page is not closing. Fixable
- No impressions → SEO problem. Title, tags, attributes. Also fixable
- Impressions, clicks and sales → make ten more like it immediately
That last one is the whole game. Most sellers find something that works and then move on to a completely different idea because they are bored. Do not do that. When something works, drain it. Every colour, every product surface, every close variation, a personalised version.
Finding a winner is hard. Squeezing a winner you have already found is easy, and it is where most of the money actually is.
The part where you have to be patient
Six to eight weeks feels like forever. It is not optional.
A new listing has to be indexed, shown to some people, and evaluated on what they do. Etsy's algorithm weighs click-through, add-to-cart and purchase rate heavily now, and it cannot weigh anything until it has data.
Judging a listing after nine days is like judging a plant after nine days. Nothing has happened yet.
Keep listing while you wait. That is the trick — the waiting is only painful if you are sitting there watching.
A realistic weekly rhythm
If you want this as a routine rather than a project:
- Monday: 20 minutes. Look at proven winners in your categories. Note three ideas
- Tuesday to Thursday: design and list. Two to four a day if you can
- Friday: 20 minutes. Check what is six weeks old and read the results
- Ongoing: when something works, make ten more like it
That is it. No secret. Just the same loop, run consistently, with the research step based on evidence instead of feeling.
The one thing I would change if you take nothing else
Stop starting with the design.
Start with the person, check what that person has already bought, then design. That single reordering is the difference between a shop that grows and a shop that is a folder of nice-looking files nobody wanted.
Short version: Start with a narrow identity, not a product. Check what has actually sold rather than what looks popular — reviews and multi-year durability, not favourites. Confirm you can compete before you commit. Test three to five designs, wait six to eight weeks, and when something works, make ten more of it.
