Most searches for a winning product start the same way: copying a product someone saw in an ad. By the time you see that ad, the product is already in the crowded phase — the first-mover advantage is gone and acquisition costs have risen.
The method that works is duller and more reliable. It looks for three things being true at once: real demand, competition you can enter, and margin that survives paid acquisition. If any one fails, the product does not work however appealing it looks.
Start by excluding, not searching
Begin by narrowing scope rather than hunting candidates. Every category you rule out early saves hours later.
Fragile and bulky products. Shipping cost and damage rates eat margin quietly. Whether a product fits a standard parcel is one of the earliest signals of whether it can be sold profitably.
Anything needing permits, documentation or certification — cosmetics, supplements, electronics, children's products. Not unsellable, but you need the cost and timeline before you order, not after.
High variant counts. A product with thirty variants is thirty times the inventory risk, and not a risk to carry on a first product.
Categories with structurally high return rates. In parts of apparel, returns alone can remove the margin.
Validating demand: where are you on the curve?
Search volume is not enough on its own; where the volume is heading matters more. Two products with identical volume are entirely different investments if one curve is rising and the other falling.
Three shapes are worth distinguishing. Rising: volume growing, competition not yet settled — the best entry point and the hardest to spot. Plateau: stable volume, established competition — enterable but requires differentiation. Declining wave: volume still looks high but the peak has passed. This is where most mistakes happen, because historical data still looks good.
Do not confuse seasonality with trend. A peak that repeats in the same month every year is a season, not a trend. Seasonal products are sellable but they have a calendar, and missing it costs a year.
Reading competition: how many, and for how long?
Competition existing is not a bad sign — zero competition usually means zero demand. What matters is saturation and age.
If many sellers have been running the same product with the same creative for a long time, entry cost is measured in ad budget and a small budget will struggle. A product sold by few sellers for a short time is either a new opportunity or something tried and abandoned; separating the two means checking how long those sellers persisted.
Look at their pricing spread too. If everyone sits at the same price there is no room to differentiate; a wide spread means there is space to position.
The margin maths: the real test
This is the most skipped step and the most decisive. What the margin has to carry for a product to be sellable with paid ads is more than most people assume.
Include: product cost, inbound shipping, outbound shipping, payment fees, the cost of your return rate, packaging, and estimated customer acquisition cost. The last is the largest and the most underestimated.
A practical check: the gap between selling price and product cost should sit clearly above your estimated acquisition cost. If it is equal or only slightly above, the product cannot be scaled with advertising. Low-margin products can work through organic traffic and repeat purchase, but that is a hard route to choose for a first product.
Validating supply: do not decide without a sample
The price on a supplier listing is a starting point, not a commitment. What needs confirming: real unit cost, minimum order quantity, lead time, quality and packaging.
Sampling looks expensive and slow, and skipping it is far more expensive. A decision made from photographs comes back as a different material or a different size.
For overseas sourcing, factor in customs and lead time. A forty-day lead time means running out of stock on a product whose ads are working.
Record your rejections
The most valuable output of the research is not the chosen product — it is the reasoning behind the rejected ones. If the same product comes up again in three months and you cannot remember why you rejected it, you are repeating the work.
One line per candidate is enough: product, reason, date. Over time this record builds a judgement about your own category that no tool provides.