How to Do Amazon Product Research: A 5-Gate Judgment Chain from Demand to Risk

Scrolling the Best Seller list is not product research

Here's how most sellers do Amazon product research: open the Best Sellers list, find something with high sales that "looks doable," order samples, commit to inventory. When it fails, it rarely fails on the numbers they looked at. It fails on the dimensions they never checked — a top 10 locked up by the same veterans for years, a pain point in the reviews they can't actually fix, a CPC that eats the entire margin at their price point.

Good Amazon product research is elimination, not treasure hunting. You're not trying to find "a winner." You're pushing a candidate category through five gates — demand, competition, pain points, differentiation, risk — and killing it the moment any gate shows a hard defect. Whatever survives all five deserves your money.

Here's what each gate looks at, what counts as a pass, and what should make you walk away.

Gate 1: Demand — is there actually room for you?

Three things: market size, monthly sales, and trend. And they need to come from real listings — the category's BSR Top 100 — not gut feel.

For the width of demand, check the category's keyword pool too: the ABA search terms buyers actually type. Demand hanging off one keyword is a very different market from demand spread across dozens of search intents.

Pass signals: the middle of the ranking (say positions 20–50) still moves steady volume, meaning the market feeds more than just the top sellers; the trend is flat or climbing.

Walk-away signals: sales are bunched at the very top with nothing left for the middle, or the trend line points down. A declining category doesn't care how good your product is — you'd be buying someone else's exit.

Gate 2: Competition — monopoly kills faster than weak demand

High sales don't mean opportunity. This gate looks at three things: how concentrated the top of the market is, how few brands own it, and the review moat.

The review moat is the most brutal one. If the established players sit on thousands of reviews and you launch at zero, ads have to carry the entire gap — and they usually can't at a survivable cost.

Then there's an angle most sellers never check: the overlap between the New Releases list and the BSR Top 100. If none of the current New Releases Top 100 has cracked the BSR Top 100 — an overlap of zero — new products simply aren't breaking in. Incumbents own the shelf. Walk away. An overlap of five or more means the market is still churning and newcomers are landing.

It also pays to know who you're actually up against: which country the Buybox sellers operate from (CN, US, HK and so on), and whether one seller is running the same product across multiple storefront brands. A matrix operation changes the nature of the fight completely.

Gates 3 and 4: Pain points and differentiation — your ticket is in the 1-star reviews

The pain point gate ranks negative-review themes, and every theme has to be backed by actual buyer quotes — not a vague "quality issues." Say you're evaluating pet grooming clippers (hypothetical numbers): the top complaint tag is "motor noise scares the animal," it shows up in roughly a third of negative reviews, and a buyer literally writes "the second I turn it on, my cat is under the bed." That's specific and fixable — a quiet motor is a differentiation direction sitting in plain sight.

The differentiation gate asks the follow-up: put the incumbents' selling points and feature sets side by side. Can you build the thing buyers are begging for in the reviews that nobody has on the shelf yet? If the top products are near-identical and the complaints cluster, that's a pass. If the complaints are "shipping was slow" and "too expensive" — things you can't engineer away — the gate fails.

Gate 5: Risk — if you can't verify it, leave it blank

Two kinds of risk. First, compliance: does this category carry certification requirements or regulatory restrictions? The only honest way to answer is researching real published sources and summarizing only what's actually there. Found nothing? Then the answer is a blank, not a guess. A made-up compliance section is worse than none.

Second, a trend circuit breaker: even if the first four gates pass, a trend signal bad enough should trip the breaker and downgrade the verdict on its own.

And one more thing hiding inside risk: price segments and ad costs. A $19 product and a $39 product in the same category are two different wars. Each price band has its own monthly volume, its own competitive density, its own reasons buyers love or hate what's in it. A useful check is gross margin headroom divided by real CPC — call it ad resilience. Suppose (again, hypothetical) CPC eats most of the margin headroom in the $15–20 band. Then that band's volume is irrelevant to you as a new entrant, whatever the Best Seller list says.

Turning five gates into go, wait, or walk away

The five gate results roll up into an opportunity score and a one-word verdict: enter, watch, or pass. "Watch" isn't a cop-out. It usually means one gate came back ambiguous — a trend that just turned up, a sample still too thin — which is a reason to monitor and re-check next cycle, not a reason to buy inventory today.

Running this chain by hand — pulling rankings, reading hundreds of negative reviews, checking regulations — takes the better part of a day per category. That's exactly why most people fall back to scrolling Best Sellers. If you want to see the full chain run on live market data, you can generate a free product research report on Sellerside.ai: type in a category keyword and get all five gates plus the enter/watch/pass verdict in one report.