Most stores that fail were lost before the first product was ever listed. The founder picked a niche on affinity — they liked coffee, or camping, or dogs — bought the domain, built the catalog, and only then found out whether anyone in that market actually spends money. By that point the answer is expensive either way.
The honest version of niche validation runs in the other order. Before you commit to a name, a catalog, and a brand, you interrogate the niche with public evidence: what people search for, what advertisers keep paying to reach, what buyers complain about. None of it requires a paid data subscription, and all of it is visible from a laptop in a few evenings.
You will not get certainty. No method delivers that before real orders arrive. What you get is a structured answer to three questions, and a defensible reason to pick one candidate niche over another.
The three pillars: demand, spend, pain
A profitable niche needs three things at once: people who want something, money that already moves, and a problem that hurts. Each pillar leaves its own evidence trail, and each trail is checkable for free.
Two out of three produces the familiar failures. Demand and pain without spend is the passionate-hobbyist trap: a community that talks all day, trades tips generously, and buys homemade, secondhand, or nothing at all. Demand and spend without pain is a solved market: money moves, but every need is already met, so a new store can only compete on price against incumbents with better costs and older ad accounts.
So the work is not proving the niche is good. The work is hunting for the missing pillar, because in most candidate niches one of the three is missing, and finding which one costs you an evening instead of a year.
Evidence of demand
Start with search direction, not search volume. Put the niche's two or three head terms into Google Trends, set the window to five years, and read the shape. Flat-and-alive is bankable. Gently rising is better. A single dramatic spike is not demand — it is a trend, and trends are a different game with different rules. Steady beats spectacular every time, because a store takes months to build and a spike does not wait for you. There is a longer method for reading these charts without fooling yourself in our field guide to Google Trends.
Then check whether the niche has a pulse where its people actually gather. Subscriber counts on a subreddit mean little; posting cadence means a lot. A community producing new threads daily, with real comment depth and recurring questions, is a niche that people live inside rather than pass through. The listening-post method covers how to read this systematically, but even ten minutes of scrolling tells you whether the room is warm or abandoned.
What you are establishing here is direction and vitality. A niche can pass this pillar and still be worthless — which is why the next check matters more.
Evidence of spend
Demand tells you people care. Spend tells you someone is already converting that care into revenue. The fastest public proof is the Meta ad library: search the niche's product nouns and any brand names you have already come across, and see who is advertising. The signal is not the raw count on any given day — it is persistence. An ad that has been running for thirty days or more is, in all likelihood, paying for itself, because nobody funds a losing ad for a month. Sustained advertisers are tenants proving the rent is payable. Reading the ad library in detail is its own discipline, but persistence alone answers the spend question.
Amazon then sets the realistic price band. Search the niche's core products and read the first page: the spread of prices there is roughly what the market has agreed to pay, and your store will live inside that band whether you like it or not.
Two economics filters apply before you accept what you see. First, a niche whose typical order sits under about twenty dollars rarely survives paid acquisition — once ad cost, payment fees, and shipping come out, there is nothing left — unless bundling or multi-packs can lift the order value. Second, repeat-purchase categories compound where one-off gadgets do not. Consumables, refills, and replaceables mean a second and third order with no new ad spend; a clever gadget means buying every single customer at full price, forever.
Evidence of pain
The third pillar is the least measured and the most predictive. Go back into the niche's communities and read complaints, not enthusiasm. What you are grading is recurrence and intensity: does the same frustration appear week after week, and do people describe it in terms of cost — money wasted, time lost, discomfort endured?
Listen for the difference between two kinds of language:
Problem-aware: "My lower back is wrecked by two in the afternoon, every workday. I have tried three chairs."
Solution-aware: "Which lumbar cushion is actually worth it? Deciding between two of the memory-foam ones."
Both are pain, but they demand different funnels. A problem-aware market has not yet settled on a product category, so your job is education — content, comparisons, a store that names the problem before it names the product. That is more work per sale, but you face fewer entrenched brands. A solution-aware market converts fast, but buyers arrive with a shortlist, so you compete on reviews, comparisons, and price against whoever taught them the category. Neither is wrong. Entering one while building for the other is what fails.
Sizing without a data subscription
You do not need a TAM slide. You need "big enough to feed one store," and an honest proxy stack gets you there:
- Community activity. How many active gathering places does the niche have, and how fast do they move? A niche with several communities posting daily is materially bigger than one with a single quiet forum.
- Advertiser count. How many distinct advertisers show sustained campaigns in the ad library? A handful of long-running advertisers means a market proven but not crowded; dozens means proven and contested.
- Review accumulation. Across the first page of Amazon results for the niche's top products, how many reviews have accumulated, and how recent are they? Reviews are receipts — each one is a completed purchase, and recent ones prove the buying never stopped.
Work in ranges and resist false precision. "Thousands of reviews across the category, ten-plus persistent advertisers, communities that move daily" is a real answer; a spreadsheet claiming a $47.3M addressable market is theater. If all three proxies read moderate or better, size stops being your risk. Execution becomes your risk, which is the risk you actually control.
Validate the niche once, then scan the products inside it
A niche is not one bet. It is a portfolio of product markets, and they do not rise and fall together — a healthy niche routinely contains saturated products, dead products, and one or two genuinely open ones. So once the niche clears the three pillars, the unit of repeated analysis shifts to each candidate product inside it. Pick the two or three hero products you would actually launch with and run the same evidence pass on each; the full four-signal method is in how to know if a product will sell.
The two levels of evidence reinforce each other. A moderately sized niche containing one strong open product beats a big impressive niche where every product you check reads crowded or cold.
Comparing niches side by side
Done by hand, that comparison is several evenings of tab-juggling: ad library searches, Trends charts, Amazon first pages, community threads, multiplied across six or nine hero products. Kestrel runs the same checks in one pass — live competitor ads from the Meta ad library, Google search demand, Amazon retail proof, and community chatter — and returns a 0-100 score with a Hot, Promising, or Weak verdict and the evidence behind it. The specimen report shows exactly what comes back, and the free tier's 20 scans, no card required, cover the hero products of three candidate niches with room to spare.
However you run it, the output you want is relative, not absolute. The question is never "is this niche good" in isolation — it is "which of my three candidates shows the strongest evidence on each pillar."
Red flags that override everything
Some findings void an otherwise green scorecard, and they deserve a deliberate check before you commit.
Single-platform dependence. If the niche's entire demand lives on one platform — usually TikTok — you are renting your market from an algorithm. One distribution change and the niche's demand curve is not yours to keep. Demand that shows up across search, communities, and retail at once is demand that survives.
Single-supplier risk. If every listing you find traces back to one manufacturer, one patent holder, or one brand that also sells direct, the niche's margin belongs to them. You would be building a store on someone else's permission.
Regulated and claim-sensitive categories. Ingestibles, anything with health claims, children's products, batteries and chargers — these carry advertising restrictions, payment-processor scrutiny, and liability exposure that a first store is not built to absorb. Not off-limits forever, but not the place to learn.
Any one of these can make a niche that scores well on demand, spend, and pain the wrong niche anyway. The scorecard informs the decision; it does not make it.
What a yes actually looks like
A niche worth building a store around shows all three pillars in public evidence: search demand that is steady or rising, advertisers who have paid rent for a month or more, and a complaint that recurs with real intensity. It sizes to "big enough" on community activity, advertiser count, and accumulated reviews. And it carries no overriding red flag on platform, supplier, or regulation.
That is as close to knowing as anyone gets before the first order. Validation does not remove the risk of building a store — it prices the bet, and it lets you place it on the strongest of your candidates instead of the one you happened to like first. Run the evidence by hand over a few evenings, or let Kestrel assemble it for you across your shortlist; either way, decide on what the market is already showing you, not on what you hope it will do.
