Offer And Ad Intelligence — Deep Dive
How to see which affiliate offers and which ads are rising right now, before the crowd piles in — what data is available, how fresh it is, what it costs, and exactly how to build the system. Everything from our conversations plus fresh verified research, in one place.
2–6 weeks
Ad signals move before gravity does
24 hours
Maximum lag on Meta Ad Library data
$40–110
Total monthly run cost
$0
Servers to rent — all serverless
Summary
The whole idea in five lines.
Winners announce themselves before the crowd sees them: affiliates keep re-launching ads that make money, ramp their spend, and pile onto rising offers. Nobody can see profit data — but everybody leaves behavioral fingerprints. Our tool collects those fingerprints daily, scores them, and hands us a short list of what is breaking out this week — plus exactly where to apply to promote it. Cost: roughly $40 to $110 a month, no servers, built on data feeds that already exist.
- The edge is speed, not volume. By the time an offer is obviously "hot," the ad costs are bid up and the edge is gone. The money is in velocity signals — spotting the rise while few affiliates are on it.
- Use existing scrapers, build only the thin layer. Proven, maintained scrapers already exist for both data sources (21 million runs, 99.8 percent success). We build only the scheduling, storage, scoring, and dashboard.
- Two-stage targeted pulling, never mass scraping. The ad world is hundreds of millions of creatives — a full sweep is impossible and pointless. A cheap broad snapshot plus targeted pulls on candidates gives the same intelligence for a few dollars a day.
- Phase 1 first: ClickBank offer trends, with the breakout-ad data model designed in from day one. It proves the pipeline in about a week and every later phase plugs into it.
What The Tool Does
The pitch video we transcribed describes three jobs: find winning offers, create winning ads, and run the media buying. We are deliberately building the first two — the intelligence layer — and skipping the media buyer for now (more on that below).
Stage 1 · SnapshotDaily cheap sweep of ClickBank marketplace stats and ad-count fingerprints
→
Stage 2 · DetectVelocity scoring: gravity slope, advertiser ramp, hook re-creation, survival rate
→
Stage 3 · DeepenTargeted pulls only on candidates: full ads, creatives, landing pages
→
Stage 4 · ActWeekly report: rising offers, breakout ads, and where to apply for each
The real asset is the snapshot history
Every day the system runs, it accumulates trend data that nobody sells and nobody can buy. After three months, our database knows which offers rose when and which advertisers ramped when — that is the moat. The scrapers are replaceable; the history is not.Corrections First
Four assumptions worth fixing before anything gets built, because each one changes the design.
One: Facebook gives nobody results data on ordinary ads.
No spend, no sales, no return on ad spend. The Facebook Ad Library only shows that an ad is running, when it started, what it looks like, and where it points. Spend and impression figures exist only for political ads. Everything we call "winning" is therefore inferred from behavior — ads kept alive for weeks, creative variants being tested, third-party estimated spend. That inference is genuinely useful, but it is a signal, not proof of profit. Only running an ad ourselves ever proves profit.
Two: ClickBank has no marketplace application interface and no official assistant connection.
Their official APIs return only our own account data — our sales, our orders. Nothing about what is selling platform-wide. The one community-built MCP server wraps those same account APIs, so it cannot find trending offers either. Market-wide data lives on public marketplace pages that anyone can see — scraping is the only route. (This was verified twice independently: against ClickBank's own developer documentation and by live endpoint testing.)
Three: gravity is a lagging confirmation signal, not a discovery tool.
Gravity is a recency-weighted count of unique affiliates earning commissions over a rolling ninety-day window — it moves days to weeks after real activity. Practitioner research puts ad-spy creatives two to six weeks ahead of gravity. Discovery happens in the ad data; gravity is what confirms we were right (and warns us when to exit — falling gravity means others are pulling budget before our earnings-per-click drop).
Four: some "known" marketplace metrics no longer exist.
The current ClickBank marketplace shows Average Dollars Per Conversion (90 days), Conversion Rate (30 days), Earnings Per Click (30 days), and Gravity. The legacy fields (Initial Sale, Average Sale, Average Rebill, Refund Rate) are gone from the current interface — refund rate survives only through third-party tools like CBEngine. Our first scrape run should verify exactly which fields come back before we design around them.
Freshness Scoreboard — Where The Data Comes From
"Most up to date data possible" is the core requirement, so here is every source ranked by how quickly it tells us something new.
| Source | What it shows | Data lag | Access route | Cost |
|---|---|---|---|---|
| Meta Ad Library (public web) Primary | Every currently running ad on Facebook and Instagram: start date, still-active status, image/video creative, landing page link | Within 24 hours of an ad's first impression (Meta's own published standard) | Public pages, no login — scraped through the maintained Apify actor | $0.75 per 1,000 ads |
| Meta Ad Library API | Only ads reaching the European Union plus political/issue ads worldwide — ordinary United States commercial ads are excluded | Same 24-hour standard | Developer app, identity verification, roughly 1–2 weeks approval, about 200 calls per hour | Free, but wrong coverage for our purpose |
| ClickBank Marketplace Primary | Per-offer stats: gravity, average dollars per conversion, conversion rate, earnings per click | Listings live within 24 hours; gravity is recency-weighted and effectively updates daily | Public pages, no login — scraped through the maintained Apify actor | $3.88 per 1,000 offers |
| Launch calendars (MunchEye, JVNotifyPro) Ahead Of Everything | Upcoming product launches with price and commission percentage — up to six months before they go live | Negative lag — pre-launch notice | Public pages | Free |
| Google Ads Transparency Center | Ads from verified advertisers across Google and YouTube | 48–72 hours after go-live | Public pages, no application interface | Free |
| TikTok Commercial Content Library | All paid TikTok ads — but only for Europe, Britain and Switzerland | Within 24 hours (legal commitment under the Digital Services Act); reach figures lag about 7 days | Public pages; a Commercial Content API exists on application (about 2 working days) | Free |
| TikTok Creative Center Top Ads | Curated showcase of winning ads by industry — editorial selection, not the full inventory | Not published — treat as a weekly inspiration feed | Public pages | Free |
Freshness verdict
The two primary sources both deliver within 24 hours and both are free to see — the only cost is collection. Running the system daily (with one full ClickBank sweep per week) puts us at most one day behind reality. Nothing on the commercial market is materially faster. The only faster signal in existence is MunchEye, which runs the opposite direction: weeks to months ahead, before an offer even exists.Spy Tools Compared — What They Add Beyond Free Data
These tools aggregate the same public ad data and add two things we would otherwise compute ourselves: an estimated spend figure (theirs is a third-party guess, never Meta's real numbers) and a days-running metric (our survival signal). Buying one can shortcut Phase 2 while we build.
| Tool | Coverage | Days-running metric | Estimated spend | Application interface | Price |
|---|---|---|---|---|---|
| WinningHunter Best value | Facebook, TikTok, Pinterest | Yes — filterable | Yes (paid tier) | No | $49 / $79 / $249 monthly |
| Minea | Facebook, TikTok, Pinterest | Yes — launch date plus days running | Yes | No | $49 / $99 / $199 monthly |
| BigSpy | 11 platforms | Yes — duration plus last seen | Partial | No | $0 – $399 monthly |
| AdSpy | Facebook and Instagram only | Seen dates only | No | No | $149 monthly flat |
| AdClarity (Semrush add-on) | Multi-platform | Yes — campaign duration, first/last seen | Directional | No | $129 – $349 monthly plus $69 per market |
| Foreplay | Swipe file and ad research | Yes — sort by time running | No | Yes — real API and MCP connection Only one | $59 / $175 / $459 monthly |
| CBEngine | ClickBank only — gravity deltas, rank history, best gainers, new and removed listings | Not applicable | Not applicable | No | $27 monthly, 7-day trial |
| CBTrends | ClickBank performance history | Not applicable | Not applicable | No | Free — but data appears stale (last updated banner shows April 2024) |
PlaybookSpy no longer exists — its domains do not resolve. Vendor freshness claims (for example "120,000 updates per day") are self-published marketing with no independent benchmarks; the practical floor is Meta's own 24-hour lag, so expect same-day to 48-hour data everywhere.
Smart shortcut
WinningHunter at $49 to $79 a month already does most of our Phase 2 research for any niche, and CBEngine at $27 does Phase 1 deltas today. Running one or both alongside our build costs little and gives us instant access plus a live benchmark to check our own data against. What none of them do — and what becomes our unique tool — is combine ClickBank offer trends with ad activity in one view.Finding Breakouts — Using The Data In Our Favor
The hard question we worked through: affiliates rotate creatives constantly, so "oldest ad still running" alone would flag old survivors and miss every fast-scaling winner. The answer is to analyze at the right level. Here is the full detection design.
Watch hook families, not individual ads
A hook family is every ad pointing at the same landing page with the same hook or script pattern, no matter how many ad identities it spawns. Creative rotation then becomes visible as iteration: a winner keeps getting re-shot and re-launched; a loser dies once. The two family metrics are family age and re-creation count. "This hook has been continuously re-launched for four months across thirty ad identities" is the breakout signature. Text similarity plus landing-page matching does the grouping.
Watch advertisers like a stock ticker
Snapshot every tracked advertiser daily: active ad count, new creatives launched per week, and survival rate — what share of their ads live past seven days. Disciplined advertisers' accounts are full of dead tests; anything they keep alive is the winner. An advertiser suddenly ramping launch volume is scaling — that is a breakout in progress.
Solve "new profiles" with discovery feeds, not a watchlist
New advertiser profiles appear constantly, so a fixed watchlist always misses the freshest money. Three feeds catch them regardless of who they are:
- Offer-driven search — take the rising ClickBank offers from Phase 1 and search the Ad Library for their names and brand keywords. Whoever is promoting a hot offer shows up, and the age of their profile is irrelevant. This is the moat: nobody else connects these two datasets.
- Domain clustering — every ad maps to a landing-page domain. A domain suddenly appearing with 20-plus active ads is a breakout attempt, no matter who runs it.
- Keyword and category sweeps of the Ad Library weekly, plus TikTok's free top-ads feed as a second breakout source.
The breakout score — five signals, one ranking
| Signal | What it means |
|---|---|
| Re-creation count plus family age | They keep iterating on this hook = it keeps making money |
| Advertiser launch acceleration | They are ramping spend — scaling a winner |
| Survival rate of their creatives | Selective keepers, not spray-and-pray — their survivors are real |
| Active ad count and estimated spend trend | Real money is being put behind it (spend is a third-party estimate — directional only) |
| Fresh advertiser velocity on the same offer | New promoters piling in weekly = the earliest breakout signal there is |
Why this beats the crowd.
Most people look at static marketplace numbers that update slowly, or browse spy tools manually. A daily automated score across these five signals surfaces the rise days to weeks earlier — and because gravity follows ad activity by two to six weeks, our ClickBank trend view doubles as the confirmation and exit layer.
Finding The Offer And Where To Apply
Intelligence is only useful if it converts into action. Every breakout leads to two questions: what exactly is the offer, and where do I sign up to promote it.
From ad to offer in three hops
AdCreative and copy from the Ad Library pull
→
Landing pageEvery scraped ad carries its destination link
→
OfferProduct name found on the page → matched to the marketplace listing
Because the scrapers capture the landing page for every ad, matching ads back to offers is a lookup, not a hunt. The domain-clustering feed from the breakout design plugs straight into this.
Where to apply — by network
| Network | Approval friction | Commission shape | Best for |
|---|---|---|---|
| ClickBank Instant | Instant "Get Affiliate Link" for most offers — no approval needed. Exceptions: vendors who switch on approval whitelisting (contact the seller and get added), and private invite-only offers | Percent of sale, often with recurring rebills; 60-day cookie; cost-per-action available by negotiating with the vendor | Everyone — the default starting point |
| Digistore24 Instant | Mostly instant; some vendors want one-click approval (~24 hours); content-link approval about 24 hours | Up to 100 percent on front-end offers; 180-day cookie | Beginners alongside ClickBank |
| JVZoo | Per-product request to the vendor — write a short promo plan with real traffic URLs | Percent of sale, often high on launches | Launch promotions once you have some results |
| WarriorPlus | Vendor-only approvals; beginners are often ignored | Percent of sale | Later, when you have a track record |
| MaxBounty | Application plus compliance review plus a short phone call | Cost per action (flat per lead/sale) | Once you are doing real volume |
| ClickDealer | Application — they want to see cost-per-action experience | Cost per action | Experienced media buyers |
| Awin (absorbed ShareASale) | Per-advertiser approval | Varies by advertiser | Physical and retail products |
The pre-launch advantage
MunchEye and JVNotifyPro — see offers up to six months early
Launch calendars list upcoming products with price and commission before they exist publicly. Verified today: MunchEye's calendar runs months ahead. This is the only genuinely "ahead of everyone" source — the offer is not even promotable yet, so we can prepare creative and be live on launch day while everyone else discovers it in the gravity numbers weeks later. JVZoo also has a built-in launch calendar where approval can be requested pre-launch.Reading the offer signals
- Gravity level: ClickBank's own guidance names 50 to 200 as a "sweet spot"; practitioner forums favor 10 to 50 — high enough to prove conversion, low enough to avoid saturation.
- Gravity direction beats gravity level: week-over-week change is the trend line. Rising 8 (from 2) beats flat 30 every time. Falling means others are pulling budget before our earnings-per-click drop.
- Earnings per click and conversion rate (both now shown per listing) tell us what traffic is worth.
- Refund pressure is invisible in the current marketplace — use CBEngine's refund column, and treat suspiciously generous recurring offers with care.
Mass Versus Targeted — Is It Too Much Data?
Short answer: mass scraping is both impossible and pointless. Targeted two-stage collection gets the same intelligence for pocket change. Here is the math that settles it.
| Approach | Monthly cost | Freshness | Maintenance |
|---|---|---|---|
| Mass-scrape everything | $1,164+ for ClickBank alone; ad world infeasible | Daily | High |
| Targeted two-stage (recommended) | $25–55 | Daily | Low |
| Do-it-yourself headless browsers plus proxies | $20–100 plus development time | Daily | High — block pages, layout churn |
| Official APIs only | $0 | Politically themed ads only — impossible for our purpose | |
Why mass fails
- The ad world is unbounded. The Meta Ad Library holds hundreds of millions of creatives — the political subset alone shows over 17.4 million on its live counter. At $0.75 per 1,000 ads, sweeping 100 million ads costs $75,000 per sweep. Meta's own free API caps at roughly 5,000 records per hour — 348 days of nonstop calling for one sweep.
- ClickBank could be swept, but should not be. The catalog is only "thousands" of live offers (their own wording) — a full sweep is cheap per run, but daily full sweeps of mostly unchanged gravity data run about $38.80 per day, roughly $1,164 a month. A weekly full sweep is about $155 — and daily top-candidate pulls give fresher signal at lower cost.
- Most data is noise. Ninety-nine percent of ads die in days. What we need is the one percent that survives and iterates — which only targeted pulling finds.
Why targeted wins
The two-stage pattern
Stage 1 (cheap, broad, daily): the whole ClickBank stats snapshot (one weekly sweep plus daily candidate refresh) and light ad fingerprints — counts, start dates, domains — for tracked names and keywords. Cents to a few dollars a day.Stage 2 (targeted, deep, daily): full pulls — creative, copy, landing page — for roughly 50 candidate ads per day: the hook families on the move, the domains suddenly loading up, the offers climbing. About $28 a month at list price.
Same intelligence as a mythical full sweep, at roughly 0.5 percent of the cost.
Existing Tools Or Build Our Own
This was the direct question: is the scraping using existing tools that already have access, or do we do it on our own? Answer: existing tools, unambiguously. We already pay for an Apify account with working API access (today's research ran through it). Two battle-tested actors cover both primary sources.
1
curious_coder/facebook-ads-library-scraper Proven
21.6 million runs, 46,786 users, rating 4.79 out of 5. Last 30 days: 2,120,577 succeeded against 1,302 failed — 99.8 percent success. Returns start date, still-active status, creative images and video, and the landing page link per ad. Facebook, Instagram, and more.
21.6 million runs, 46,786 users, rating 4.79 out of 5. Last 30 days: 2,120,577 succeeded against 1,302 failed — 99.8 percent success. Returns start date, still-active status, creative images and video, and the landing page link per ad. Facebook, Instagram, and more.
$0.75 per 1,000 ads
2
bovi/clickbank-marketplace-scraper Proven
946 of 946 runs succeeded in the last 30 days. No login required. Returns gravity, sale figures, rebill, and earnings per click by keyword or category. There is also a purpose-built cauldo/clickbank-offer-monitor for before/after change detection — worth evaluating for the trend layer.
946 of 946 runs succeeded in the last 30 days. No login required. Returns gravity, sale figures, rebill, and earnings per click by keyword or category. There is also a purpose-built cauldo/clickbank-offer-monitor for before/after change detection — worth evaluating for the trend layer.
$3.88 per 1,000 offers
3
Alternates standing by Backups
memo23 (from $0.30), s-r/meta-ads-library ($0.46 per 1,000), corner_cutter (1.09 million runs), data_xplorer for TikTok (51,500 runs). Bright Data sells a Facebook ads dataset at $2.50 per 1,000 with a $250 minimum — worse economics. Scrapingdog has no Ad Library product at all.
memo23 (from $0.30), s-r/meta-ads-library ($0.46 per 1,000), corner_cutter (1.09 million runs), data_xplorer for TikTok (51,500 runs). Bright Data sells a Facebook ads dataset at $2.50 per 1,000 with a $250 minimum — worse economics. Scrapingdog has no Ad Library product at all.
$0.46–2.50 per 1,000
What we build ourselves — and only this
A thin orchestration layer: the scheduled jobs, the storage, the delta and velocity scoring, the hook-family grouping, and the dashboard. That is a small, maintainable codebase — exactly what Codex writes well. The rule: never build a scraper when a maintained one exists. When ClickBank or Facebook changes a page layout, the actor's maintainer fixes it (both actors are actively updated) — that ongoing repair is the entire reason to rent scrapers instead of owning them.
Where do-it-yourself scraping actually stands
- ClickBank: public and login-free, but plain scripted requests are dead — legacy endpoints and even the contract page return block pages to simple requests. Headless browsers through clean infrastructure work fine (946 of 946 on Apify proves it). The failure mode is a page-layout change, not an account ban.
- Meta Ad Library: publicly searchable with no login; the official application interface covers political ads only, so commercial affiliate ads are scrape-only territory.
- Legal posture (research-input use): the leading case — Meta versus Bright Data, January 2024 — held that terms of service could not bar logged-out scraping of publicly available data, and Meta dropped the appeal. The Ninth Circuit's hiQ versus LinkedIn ruling similarly treats public-data scraping as outside the anti-hacking law. Scraping public data for our own research sits in the low-risk zone; the caution is to treat it as research input, never a resold data product. ClickBank's own contract wording is unverifiable (their legal pages are behind the same bot wall) — flagged, not quoted.
One more advantage worth naming
Apify has its own MCP server — meaning Hermes and Codex can call these scrapers as first-class tools during research and builds. Today's entire research run used exactly that connection.Tech Stack And Costs
Everything runs serverless. There are no servers to rent, patch, or babysit. Everything in the "already owned" column is cost we pay regardless.
| Layer | What it does | Tool | Monthly cost | Status |
|---|---|---|---|---|
| Data collection — ClickBank | Pulls every offer's stats: gravity, conversion, earnings per click | Apify actor (bovi) | $3.88 per 1,000 offers | Apify account already owned |
| Data collection — Facebook | Pulls running ads: start date, creative, landing page | Apify actor (curious_coder) | $0.75 per 1,000 ads | Already owned |
| Scheduled jobs | Runs the collection daily, computes what is rising | Cloudflare Workers cron | Free — 100,000 requests a day included | Already owned |
| Database | Timestamped snapshots — week-over-week comparison is what makes "trending" exist | Convex | Free — 0.5 gigabytes included | Already owned |
| Backend logic | Velocity scoring, hook-family grouping, offer matching | Cloudflare Workers | Free tier covers it | Already owned |
| Dashboard | Tables, trend arrows, filters — the tool we actually look at | Cloudflare Pages with our standard report layout | Free | Already owned |
| Ad creation (Phase 3) | Hooks and scripts from winning ads, generated images | Existing models plus RunningHub image interface | About $10–30 | Already owned |
| Build tooling | Writes the code | Codex plus Hermes | $0 new | Already owned |
| Login (only if selling access) | User accounts | Clerk | Free until roughly 10,000 users | Already owned |
What we actually pay each month
$25–55
Scraping — targeted two-stage pulls
$10–30
Image generation for ad batches
$0
Hosting, database, jobs, front end
$40–110
Total new monthly cost
One real storage caveat
The Convex free tier's 0.5 gigabytes is fine forever for a latest-state table with deltas (about 10 megabytes at 10,000 rows) — but keeping full daily snapshots forever would breach it in about two months. Design rule: store deltas and prune full snapshots after 30 to 60 days. The trend value lives in the deltas anyway.How the pieces connect, in plain English
NightlyA scheduled job wakes up and tells the scrapers to collect
→
LandStructured rows arrive — one per offer, one per ad — timestamped
→
CompareToday against yesterday turns raw rows into "rising"
→
ScoreBreakout score per hook family and per offer
→
ReportDashboard plus a weekly "test these three" digest
Hard Limits — What This Tool Can Never Do
It can never tell us whether a winner-looking ad is actually profitable.
No tool can, because that data does not exist anywhere public. "Breakout" means "behaving like a winner" — survival, iteration, ramping spend, fresh promoters. The only proof of profit is running the ad ourselves. This tool's job is to shrink the number of expensive tests we have to run, not to eliminate testing.
- Estimated spend figures are third-party guesses, never Meta's real numbers. Directional only.
- Scrapers can break when page layouts change — expect a few days of downtime a couple of times a year. Because we store our own snapshots, a missed day gaps a trend line instead of destroying anything.
- Terms-of-service exposure is real but low for logged-out public data used as research input — never resell the data.
What Is Next
Every path laid out side by side, with where each one stands today.
| Option | What it gives | What it misses | Ready to run |
|---|---|---|---|
| Build our own two-stage pipeline | Combined ClickBank-trends-plus-ad-activity view nobody sells; our own history moat; breakout scoring; scales to Phase 3 | One to two weeks of build time before first output | Spec → Codex, this week |
| WinningHunter subscription only | Instant ad research with days-running and estimated-spend filters across Facebook, TikTok, Pinterest | No ClickBank trend data, no combined view, no owned history, no Phase 3 | Live today, $49–79 |
| CBEngine subscription only | ClickBank gravity deltas, best gainers, refund column — $27 with a 7-day trial | ClickBank only — no ad intelligence at all | Live today, $27 |
| Do-it-yourself headless scraping | Total control | Block-page whack-a-mole, constant maintenance — worse data for more effort | Not recommended |
| Official application interfaces only | Free and sanctioned | Coverage is wrong: own-account ClickBank data and political ads only | Impossible for this goal |
Top pick
Build Phase 1 now — with the breakout data model baked in from day one
A weekly ClickBank gravity-change report (what is rising, what is dying) that stores its snapshots in the same shape Phase 2 will need: hook families, advertiser fingerprints, and landing-page domains as first-class fields. It is self-contained, proves the whole pipeline in about a week, and every later phase plugs straight into it. Meanwhile, run WinningHunter's cheap tier or CBEngine's free trial as a live benchmark to check our numbers against — and cancel whichever we outgrow.
Phase 1 — Offer trends
Weekly ClickBank snapshot plus gravity-slope ranking, hook-family-ready storage. About a week to first report.Phase 2 — Ad intelligence
Ad Library pulls for candidates, hook-family grouping, breakout score, offer matching. The moat.Phase 3 — Ad creation
Generate hooks and images from proven patterns using our existing generation stack. A few dollars a batch.Decision checklist
- Confirm the Phase 1 scope: weekly ClickBank trend report, hook-family data model included
- Decide on the benchmark subscription (WinningHunter trial, CBEngine trial, or both) — optional but useful
- Verify in the first scrape run which marketplace fields actually come back (refund rate especially)
- Hand the Phase 1 spec to Codex; Hermes owns the scoring design and the report generation
- Defer the media-buyer question entirely — revisit only after Phases 1 and 2 are producing weekly reports
Sources
Official documentation and platform data
- Meta for Developers — Ads Archive reference (coverage: European Union ads plus political worldwide; 24-hour appearance standard)
- Meta Ad Library API (access requirements, identity verification)
- Meta Transparency Center — Ad Library tools
- Meta Ad Library report — live political ad counter (17.4 million-plus ads)
- ClickBank — marketplace metrics article (current fields: average dollars per conversion, conversion rate, earnings per click, gravity)
- ClickBank — ClickBank APIs (own-account coverage only)
- ClickBank blog — gravity score explained (recency weighting, 50–200 sweet spot)
- ClickBank Marketplace (instant affiliate links, no approval for most offers)
- Cloudflare Workers pricing and Convex pricing (free-tier figures)
Collection tools
- curious_coder/facebook-ads-library-scraper (pricing, run volumes, success rate)
- bovi/clickbank-marketplace-scraper (pricing, returned fields)
- Bright Data — Facebook ads dataset (pricing comparison)
- WinningHunter, Minea, BigSpy, AdSpy vendor sites (tiers and features — vendor-published)
- CBEngine and CBTrends (ClickBank trend tools)
Offer discovery and application
- MunchEye (launch calendar — verified running months ahead)
- JVNotifyPro (joint-venture launch notices)
- Digistore24 (beginner-friendly instant approval network)
Legal context
- Meta versus Bright Data analysis (January 2024 — logged-out public scraping not barred by terms of service)
Vendor pricing and feature claims are vendor-published and labeled directional where they are marketing figures. Platform statistics (ad counters, run volumes, success rates) were live-verified during research. Operational scraping costs shown at list prices.