Research · 1 credit
Paste a product name or a product URL. One credit buys a full market analysis across current demand, sourcing, pricing, competition, buyer intent and risk — and one verdict you can act on, with the evidence attached.
Give it a name and it works from that. Give it a URL and it reads the page first to pull the product and its visible specifications, then runs the analysis on a clean product name.
The mix changes with the product, drawn from live TikTok, AliExpress, Google Shopping, Google Trends and Amazon lookups plus web research. Irrelevant checks are skipped rather than run for show — three to five well-chosen ones beat six redundant calls, and the run records which it used.
Not a score out of a hundred and not a wall of charts to interpret. One call, one reason, and the numbers that produced it — including the ones it could not confirm.
5 follow-up questions are included with every research, then 1 credit each. Ask for a faster supplier, a different ad angle, a margin stress-test at a lower price, or a read on another country — and the relevant source is re-run when freshness matters, because a follow-up runs the same engine the opening question did.
Each carries its own limitation, and each limitation is stated on the run rather than buried.
Is anyone actually looking for this right now, and is that interest rising, flat or fading? A twelve-month interest curve puts a product in one of those three buckets instead of leaving it to a hunch.
Search interest is relative, and it is not sales volume.
What a unit costs to buy, what shipping options exist, and how many orders and ratings the listings already carry. This is the number the rest of the economics rests on.
Sampled listings, not a complete count of every supplier.
What comparable products actually sell for in retail today, so the gap between source price and realistic retail is measured rather than assumed.
Sampled results, not every seller in the market.
How crowded the shelf already is, and whether the incumbents are established brands or thin listings. Saturation comes back as low, moderate or high.
Sampled top results; some ranking data is only available when the source exposes it.
Evidence that real people are engaging with this product rather than just scrolling past it — creative that exists, discussion that exists, angles that already work.
Organic evidence from sampled content. It is not paid-ad intelligence.
The reasons a product fails after the money is spent: trademark and IP exposure, fragility or shipping problems, claims you would not be allowed to make in an ad.
Public-web research. Treat it as a prompt to check, not as legal advice.
Five dedicated live integrations, plus public-web research on top. A run calls the ones that fit the product and records which it used. Prices and trends are read at the moment you ask rather than recalled from a model's memory.
| Source | Provider | What it can tell you |
|---|---|---|
| TikTok organic searchDedicated live integration | Apify | Normally up to 20 matching organic videos. View totals cover the sampled videos only, never TikTok overall — and this is organic evidence, not paid-ad intelligence. |
| AliExpressDedicated live integration | RapidAPI | A sample of up to 12 returned listings with source price, orders, ratings and shipping fields. Not a complete count of suppliers. |
| Google ShoppingDedicated live integration | SerpAPI | A sample of up to 12 returned shopping listings, not every seller in the market. |
| Google TrendsDedicated live integration | SerpAPI | Relative search interest over a 12-month curve. Search interest is not sales volume. |
| Amazon searchDedicated live integration | SerpAPI | A sample of the first 10 organic results. Best-seller rank is included only when the upstream response exposes it. |
| Public webPublic-web research | Claude web search | Search-engine coverage varies. Anything found here is public-web research and is labelled as such — it is not a direct integration with any platform. |
The distinction in the first column matters. A dedicated integration queries that source directly; public-web research is a search that may happen to surface a page from somewhere. They are never presented as the same thing, and a run will not describe a web result as a direct integration it does not have.
The evidence lines up: there is demand, the economics survive an ad-spend assumption, and the shelf is not already full. It is a reason to run a small paid test — not a promise of sales.
Something does not add up, or the evidence is too thin for a confident call. When a core source fails or returns nothing usable, the run says so and lands here rather than manufacturing confidence.
There is a clear reason not to spend money on this one — margin that does not work, demand falling away, a market already owned, or an IP problem you do not want.
The fixed shape is the point: two products are comparable because the same seven lines come back for both, and a missing number is visible as a missing number.
A verdict you cannot audit is just an opinion with better formatting. Every completed research carries a private trail recording, per source: whether it ran, when it was checked, how many results it returned, the pages it drew on, and the limitation on the sample it took.
The trail also separates dedicated live integrations from ordinary public-web research, so a number is never dressed up as coming from somewhere it did not. Sources that were skipped, that failed, or that are not currently available are all shown as such — an incomplete run says it is incomplete and lowers its own confidence rather than filling the gap with a plausible figure.
How coverage, confidence and margin are calculatedPlans start at $29 a month for 25 credits, and unused credits roll over. Want to sanity-check the economics first? The profit margin calculator is free and needs no account.
