Where this data actually comes from

Paid tools like Winning Hunter, PipiAds, Minea, and the rest of the “Spy Essentials” bundle charge $29.99/mo for research that is largely built on top of two sources anyone can access for free: the Meta Ad Library and TikTok Creative Center. This page tells you exactly what we pulled, how, and what we deliberately left out because it isn't real.

Source 1: Meta Ad Library — WORKS (public web, no login/token)

Meta's official Graph API (ads_archive) was locked down in 2026: as of this build, it only returns political/social-issue ads running in EU countries, regardless of what access token you use. We tested this live — a bare call to the API returns an auth error with zero token, and Meta's current developer docs confirm commercial ads are excluded from the API entirely now.

What still works, and is genuinely free with no login: the Ad Library websiteitself (facebook.com/ads/library), which anyone can browse with no account for any keyword/country. It sits behind a JS bot-check that a plain HTTP request can't pass, but a real (headless) browser gets through it in one request — no CAPTCHA, no account, no scraping of private data. We pull: advertiser (Page) name, ad copy, landing-page link, creative image/video, and — the important one — start date / end date / still-running status, which is the exact “last seen to ad spend” signal every paid spy tool sells as its core feature.

What we do NOT show from Meta: a dollar spend figure or an impressions/reach number. Meta only discloses those for political/social-issue ads in regulated countries — never for ordinary commercial products. If a tool shows you an exact ad-spend dollar amount for a dropshipping ad, that number came from a paid, modeled estimate, not from the free public source.

Source 2: TikTok Creative Center — WORKS (public web, no login)

TikTok Creative Center's Top Ads dashboard is free to browse with no TikTok account. TikTok does publish a “Commercial Content API,” but it is restricted to approved academic researchers and qualified organizations only — not available for a tool like this (confirmed against TikTok for Developers docs). Calling the dashboard's internal API directly (no browser) returns {"code":40101,"msg":"no permission"}— confirmed live during this build — because the request has to be signed in-band by the page's own script.

What works: driving the real Top Ads page and searching it exactly like a human would (type a keyword, press enter). That returns genuine data straight from TikTok: ad caption, like count, CTR, and a working video preview.

Honesty caveats on TikTok data:its keyword search is a fuzzy full-text match over captions/hashtags, not a literal product catalog search, so ambiguous terms (e.g. “corrector”) can return unrelated matches — we keep the raw ad caption visible so you can judge relevance yourself. The “cost” TikTok reports is a bucketed spend-LEVEL indicator, not a real dollar figure, so we never show it as currency.

The scoring formula (fully transparent, unlike the black-box competitors)

Meta Ad Library ads are scored out of 100:

Days active (up to 50 pts, scaled to a 1-year ceiling) — how long the ad has been running continuously. This is the single strongest free proxy for sustained ad spend: a dropshipper doesn't keep paying for a losing ad.
Still running today (20 pts) — a flat bonus if the ad is currently active in the Ad Library, vs. one that has stopped.
Multi-advertiser count (up to 30 pts) — how many distinct Pages are running ads that match this niche/keyword. If several independent brands are all selling it, it's much less likely to be a fluke.

TikTok Creative Center ads are scored out of 100:

Likes (up to 55 pts) — real engagement TikTok's own Creative Center reports for the ad, scaled against a 5,000-like ceiling.
CTR (up to 30 pts) — TikTok's reported click-through rate for the ad, scaled against a 3% ceiling.
Presence bonus (15 pts) — a flat bonus for appearing in TikTok's own Top Ads ranking for the selected time window at all.

We never show a dollar spend or an impressions number for these ads — Meta and TikTok only disclose those for political/social-issue ads, not ordinary commercial products. Anyone who tells you an exact ad-spend dollar figure for a commercial dropshipping ad got it from a paid, non-public estimate model, not from the free public source.

How the data gets here (and how fresh it is)

The dashboard reads from a cached database, refreshed by a collector script that drives a real headless browser against both sources for a fixed list of seed niches. This is a deliberate MVP choice, not a limitation we're hiding: running a headless Chromium on every keystroke of a live search isn't something Vercel's standard serverless functions do well, and re-scraping per request would also hammer both sources far harder than is fair to their free public access. Each niche card shows when it was last refreshed. Adding a new niche today means running the collector for that keyword — see the repo README for the exact command; wiring that up as a self-serve “research this niche” button that runs on-demand (via a serverless-compatible headless Chromium) is the planned v1.5.

What's next

v1.5: on-demand niche search (serverless headless Chromium), “find similar” competitor expansion (re-running the same free search against other advertisers selling the same product), and ad-creative transcription. v2: an MCP server wrapper so this tool can be queried directly from Claude Desktop/Claude Code, the same way the paid tools are used in their own demo videos. Monetization plan: a free tier (browse the cached niches, as you can today) and a paid tier for on-demand niche research + the MCP connector — no billing is wired up yet in this MVP.