I5 · Ad-intelligence data asset — deep dive

Prepared 2026-10-20 from four research notes and a read of your MediaMagic and LevchykUGS code. Estimates are labelled.

Bottom line

• Keep I5, but sell judgement and production, not data. The one legal performance signal is EU reach, which Meta's API returns for every ad shown in the EU in the past year (Meta). But Meta's Platform Terms ban "selling, licensing, or purchasing Platform Data" (Meta), and Google forbids monetizing its transparency data (Google). So give rankings away; charge for teardowns, kits and community.
• Your current "scraper" isn't an asset. It's a third-party Apify actor ($0.00075 per ad, 46,138 users), one of 10+ selling the same rows for $0.00015–0.0058 (Apify). Meta's Terms ban automated collection "regardless of whether… logged-in" (Meta). Move to the official API; first test whether Meta's ID check accepts you from Ukraine.
• The obvious gaps are already closed. Agent access via MCP (Gethookd from $19/month on annual plans), hook and angle tagging (Atria from $129), an EU-reach filter (AdHeart, $69) and "clone this ad" statics already exist. Still open: finished AI-UGC video remakes of winners, single-niche verdicts, free public rankings, a sub-$40 tier, and MagicBrief's former users (it shut on 31 Jul 2026 — MagicBrief).
• Distribution is the weak link. Audiences here belong to people on camera: Dara Denney has 150K YouTube subscribers after 598 videos and sells an $899 course (YouTube). Faceless teardown channels found have 0–53 subscribers. Without your face or a cloned voice, lead with data nobody else publishes.
• Whop is the money rail, with two rules. It pays to a UAH account, charges 2.7% + $0.30 per card payment and can gate a web tool (Whop), but bans "data-scraping tools" and supplement sales (Whop). Paddle bans "community access" and marketing services: skip it.
• The math (estimate): $10k a month ≈ 180 Scaling Room members at $39 plus ~80 kit sales at $49 average, after fees and tax. At 1% free-to-paid and ~17% monthly churn (beehiiv) that needs ~18k free subscribers. First kit sales in 2–4 weeks; $10k in 12–24 months, if at all.
• Test for 30 days. On 20 Nov 2026, go if kits have earned ≥$500, the free list has passed 300 and ≥15 people have pre-paid for the Scaling Room.

Data you can legally use

Only EU transparency data gives a performance signal (reach); elsewhere you only see how long an ad ran.

Sources: Meta, Meta terms, TikTok, Google, LinkedIn, X, Snap.

What this means for you:

• Compute your own signals from the Meta and TikTok APIs: 7-day EU-reach growth, run length, live variants, country spread. Publish rankings free, each linked to its official library page.
• Sell only what you add: verdicts, hook and angle labels, remix briefs, your own regenerated creatives. Never data rows.
• For agents, the customer brings their own Meta token, so they are the API user; your code adds scoring and labels.
• In videos, show short excerpts with commentary: US fair use and the EU quotation exception cover that (US Copyright Office; EUR-Lex), and YouTube allows "using clips for a critical review" (YouTube).
• Keep scraped and Foreplay/GetHookd data out of anything public or paid.

flowchart TD
    M["Meta Ad Library API - EU reach"] --> Y["Your scoring and teardowns"]
    T["TikTok content API - EU reach"] --> Y
    G["Google transparency data"] --> R["Research only - pick ads to study"]
    R --> Y
    Y --> F["Free rankings with links"]
    Y --> K["Paid kits, teardowns, community"]
    C["Customer's own Meta token"] --> A["Customer's agent runs your toolkit"]
    S["Scraping the web library"] -.-> X["Never in a product"]
    M -.-> Z["Reselling raw rows - banned"]

The market and its gaps

Prices on 20 Oct 2026: Gethookd, Foreplay, Atria, adlibrary.com, AdHeart, Swipekit, Minea, PiPiAds, BigSpy, Motion. The market is mid-sized: Foreplay ≈ $1.2M a year, Motion ≈ $7.4M (modelled — GetLatka, GetLatka).

Gaps worth owning:

1. Finished AI-UGC video remakes of winners. Every "clone" feature found stops at statics, scripts or briefs.
2. One niche, with verdicts. Tools sell a search box; Foreplay's "Experts" page still says "Coming Soon" (Foreplay).
3. Free public rankings. Every competitor gates its data behind a trial.
4. Price. A $39 analysis tier sits under the $59–129 tools.

No longer gaps: MCP access, hook tagging, EU-reach filters, static clones, transcripts. A Skool look-alike, "Admode AI" (155 members), already pitches "AI ads built on Meta winners, made in Claude" (Skool).

quadrantChart
    title Price vs depth of insight
    x-axis Low price --> High price
    y-axis Generic search --> Deep or unique
    quadrant-1 Premium depth
    quadrant-2 Cheap and deep
    quadrant-3 Commodity search
    quadrant-4 Pricey generic
    Gethookd: [0.12, 0.45]
    PiPiAds: [0.15, 0.35]
    Swipekit: [0.18, 0.3]
    AdHeart: [0.25, 0.4]
    Minea: [0.3, 0.3]
    Foreplay: [0.4, 0.65]
    adlibrary: [0.55, 0.7]
    Atria: [0.7, 0.75]
    Motion: [0.95, 0.8]
    I5 free board: [0.05, 0.55]
    I5 Scaling Room: [0.2, 0.85]

The product ladder

flowchart TD
    E["Official EU ad data"] --> S["Scoring and teardown engine"]
    S --> B["Free public board"]
    S --> V["Daily data shorts and carousels"]
    B --> N["Free Monday Scalers email"]
    V --> N
    N --> K1["Niche winner kit - 39 USD"]
    N --> K2["AI UGC remake kit - 79 USD"]
    K1 --> R["Scaling Room - 39 USD a month"]
    K2 --> R
    N --> R
    R --> T["Team plus agent toolkit - 99 USD a month"]
    S --> O["Open-source MCP - free"]
    O --> T
    S --> AP["Apify actor - pay per result"]

Open the mockup: a sample Monday issue, the free board and plans (fictional brands).

Where the AI compute runs

Next steps

gantt
    title First 30 days from 21 Oct 2026
    dateFormat YYYY-MM-DD
    axisFormat %d %b
    section You once
    Meta ID check and developer app :a1, 2026-10-21, 3d
    Whop KYC and TikTok API form :a2, 2026-10-21, 3d
    section AI builds
    EU reach pipeline for 3 niches :b1, 2026-10-22, 6d
    Public board and free email :b2, after b1, 4d
    Niche kit and remake samples :b3, 2026-10-28, 9d
    section Publish and sell
    Daily shorts and carousels :c1, 2026-11-02, 18d
    Monday issues :c2, 2026-11-02, 18d
    Scaling Room presale :c3, 2026-11-09, 11d
    MCP toolkit and Apify actor :c4, 2026-11-11, 7d
    section Decide
    Day 30 go or kill :milestone, m1, 2026-11-20, 0d

I do:

• Week 1: add the official Meta API (and TikTok's, once approved) to the MediaMagic ingestion; scoring and labels; confirm niche 1 (default: EU skincare and supplements, then sleep and AI apps).
• Week 2: public board, free list, issue 1 on 2 Nov, daily posting via the YouTube, Instagram, LinkedIn and X APIs, kit 1 on Whop.
• Week 3: remake samples, Scaling Room presale.
• Week 4: open-source MCP, Apify actor, day-30 dashboard.
• Ongoing: daily posts, Monday issues, quarterly kit refreshes, weekly metrics.

You do once (about 2–3 hours):

1. Confirm your identity at Facebook.com/ID and create a Meta developer app; tell me if Ukraine is refused.
2. Submit the TikTok Commercial Content API form (reply within 2 working days).
3. Whop KYC and a UAH payout account (ask if your FOP account works).
4. Accounts: beehiiv (free), a LinkedIn developer app with "Share on LinkedIn", an X developer account with a few dollars of credit.
5. Choose the face: brand name only, your name, or your cloned voice (30 minutes of recording).
6. Agree to two rules: no resale of ad data, and no live customer chatbot.

You do recurring (about 15 minutes a week): approve claims about results; optionally record a 60-second weekly take (likely the biggest growth lever); check Whop payouts land in your FOP account.

Go/kill thresholds:

• Day 7: you have Meta API access. If the ID check fails, fall back to TikTok data, run length and commentary, and halve the targets.
• Day 30 (20 Nov 2026): go if kits earned ≥$500, the free list is ≥300 and the Scaling Room has ≥15 presales. Kill if kits are under $150 and the list under 100. In between: one more 30-day cycle on a second niche.
• Day 90: go at ≥$1.5k a month with Scaling Room churn under 15% a month; otherwise fold the engine into I1 (Apify) and your own ads in I3.

Risks and open questions

• Meta's ID check from Ukraine is unconfirmed; without the API there is no legal reach signal.
• Grey line: whether rankings derived from API data count as "Platform Data" is unclear (Meta's FAQ wasn't readable). Keep rankings free, linked, with no bulk exports.
• Whop could read the product as a "data-scraping tool" and hold payouts. Describe it as research and analysis; keep scraping out.
• Growing without a face has a near-zero base rate here; a cloned voice is untested.
• Clones: look-alike communities launch monthly; free rankings are easy to copy. Your edge is cadence and remakes.
• Copyright: brands may object to excerpts. Use frames plus commentary, link to the library, remove on request. Content ID on ad clips is untested.
• EU-only signal: US-only brands never show up, and US-focused buyers may discount it.
• Churn ~17% a month in money niches: push annual plans.
• Private data: Foreplay and GetHookd data and the private Kaye corpus never go into products. Remove MediaMagic's secrets and CAPTCHA-bypass code before open-sourcing anything.
• Open questions: Skool payouts to Ukraine, Whop and FOP accounts, TikTok's API terms, X's rules for automated accounts.

Sources

1. Meta: Ad Library API
2. Meta: Archived ad field reference
3. Meta: Platform Terms
4. Meta: Terms of Service
5. Meta: Automated Data Collection Terms
6. TechCrunch: Meta v. Bright Data ruling (2024-01-24)
7. TikTok: Commercial Content API
8. TikTok: Creative Center Top Ads
9. Google: Ads Transparency Center dataset README
10. Google: Ads Transparency Center Additional Terms
11. LinkedIn: User Agreement
12. X: Ads transparency
13. Snap: Ads Gallery
14. U.S. Copyright Office: Fair Use Index
15. EUR-Lex: Directive 2001/29/EC
16. Apify API: curious_coder Facebook Ad Library Scraper
17. Apify Store API: "facebook ads library"
18. Apify Store API: "winning ads"
19. Gethookd
20. Foreplay: Pricing
21. Foreplay: Experts
22. Foreplay: Affiliates
23. Atria: Pricing
24. Atria: API and MCP
25. adlibrary.com: Pricing
26. adlibrary.com: API
27. AdHeart
28. Swipekit: Pricing
29. Minea: Pricing
30. PiPiAds: Pricing
31. BigSpy: Pricing
32. Motion: Pricing
33. Motion: Creative trends report
34. MagicBrief: Shutdown notice
35. GetLatka: Foreplay
36. GetLatka: Motion
37. YouTube: Dara Denney
38. YouTube Help: Monetization policies
39. Skool discovery: "ai ads"
40. Skool discovery: "media buying"
41. Skool discovery: "ad creative"
42. Skool: AI Video Bootcamp
43. Marketing Examples
44. DTC Newsletter
45. Gumroad: Meta Ads Toolkit 2026
46. Gumroad search: UGC ad templates
47. DirectoryGems: How directory sites survive the AI era
48. beehiiv: State of Paid Newsletters 2026
49. beehiiv: Pricing
50. Whop: Fees
51. Whop: Prohibited businesses
52. Whop: Newsletters
53. Whop: SaaS
54. Paddle: What you can't sell
55. Microsoft Learn: Share on LinkedIn
56. X API: Pricing