How tgadsspy works — public methodology for Telegram ad indexing

Telegram Ads Spy is an independent OSINT project that indexes every advertising campaign running on the official Telegram Ads platform — both TON-billed and EUR-billed cabinets — in real time. This document is the canonical reference for researchers, journalists, and competitive-intelligence teams who want to verify or cite our data.

What we index#

1. Sponsored messages from the Telegram Ads platform — ads served via the official channels.getSponsored API endpoint, which Telegram exposes to any authenticated user viewing a public broadcast channel.
2. Paid channel placements (TON) — public posts in broadcast channels where the creative is marked with a CTA URL pattern we recognize as paid placement. Detection is based on text patterns + CTA URL structure, not proprietary logic.
3. Channel metadata — subscriber count, avatar, description, primary language, category. Pulled via channels.getFullChannel.

We never index:

• Private channels, invite-only groups, or anything behind a join request screen.
• Private chats or group conversations between identifiable users.
• Content protected by Telegram's flood limits or DMCA flags.

How a channel enters our pool#

Four paths:

1. Batch seeds — hand-curated lists in scripts/batches/, grouped by language and niche.
2. Auto-discovery — the Telegram Ads Spy-discover cron polls Telegram's contacts.search with 202 keyword seeds every hour.
3. User submit — anyone can paste a channel handle at /submit and it enters the queue.
4. CTA extraction — when an ad creative links to t.me/{handle}, we auto-add {handle} to the pool.

Update frequency#

Component	Cadence
Resolver (metadata refresh)	Every 10 minutes
Ingest tier S (top channels)	Every 5 minutes
Ingest tier A (active channels)	Every 30 minutes
Ingest tier B (tail channels)	Every 4 hours
Discover (new keyword seeds)	Hourly
Media mirror (gramesh → permanent URL)	Every 5 minutes

Data dictionary#

Each ad creative in our index carries:

• id — stable CUID, permanent across rebuilds.
• text + title — raw creative copy as served by Telegram.
• paymentSource — TON or EUR_ADS_CABINET.
• niche — one of 18 classifier buckets (crypto, gambling, trading, dating, health…).
• geo — ISO alpha-2 code, detected from creative language + CTA domain.
• lang — detected creative language (ISO 639-1).
• firstSeenAt + lastSeenAt — UTC timestamps, stable.
• reachEstimate + spendEstimate — derived from channel audience × market CPM.
• advertiser — structured entity with { name, domain, tgUsername, slug }.

Full schema: /api-docs.

How we estimate reach and spend#

For each creative:

1. Sum the audience (subscribers) of all channels where we observed the creative.
2. Apply average views ratio from our channel stats (avgViews / members).
3. Multiply by current market CPM benchmark for the creative's niche + geo.

Spend estimates are benchmarks, not bookkeeping. They reflect a reasonable-case market rate for the displayed volume, not the advertiser's actual cabinet spend. Use them for relative comparison between advertisers, not for exact attribution.

What we will never do#

• Re-sell Telegram cabinet access.
• Claim affiliation with Telegram FZ-LLC.
• Pay for private channel entry or insider data.
• Hide creative content behind a paywall (all ads are public by definition; our paid tiers sell convenience — API volume, alerts, team seats — not access to data).

How to verify our numbers#

1. Pick any advertiser on /advertisers.
2. Compare the creative count on their page with the /api/v1/advertisers/{slug} JSON.
3. For any listed channel, open t.me/s/{handle} in an incognito browser — you will see the same sponsored messages we indexed.
4. For disputed data points, contact @dumov on Telegram with the creative ID.

Licensing#

All exported data (CSV, JSON, RSS) is released under CC-BY-4.0. That means you can use it for any purpose — including commercial — provided you credit tgadsspy.com and link to the specific creative or advertiser page.

Example citation:

Source: tgadsspy.com, retrieved on 2026-04-20, https://tgadsspy.com/advertisers/{slug}

Change log#

• 2026-04-20 — initial methodology publication.

tgadsspy team (2026). How tgadsspy works — public methodology for Telegram ad indexing. tgadsspy.com. Retrieved from https://tgadsspy.com/blog/how-telegram-ads-spy-works-methodology