Telegram Ads intelligence — open data for researchers.

Public datasets

Monthly snapshots of every ad creative indexed. Download once, cite freely.

Top verticals in the dataset

Crypto, trading, prop-firm and cashier verticals dominate. Each row links to the full per-niche feed (RSS + JSON-LD).

Geographic coverage

Country pages group the archive by the country an advertiser targets. Each lists local advertisers, niche mix and recent creatives.

Methodology

How we detect and classify ads

We publish how the data is collected because researchers don't cite black-box archives. Below is the four-step pipeline, end to end. Every step is reproducible from the same public Telegram surface anyone with an account in the relevant country can observe.

1. 1

   Continuous polling from real accounts

   Our backend polls Telegram's sponsored-messages API every 30 minutes from a pool of real Telegram accounts. We see ads targeted at users in Russia, the United States, Brazil, Iran, Turkey, Germany, the EU, India, MENA and the rest of the world — including ads that would never appear from a single IP.

2. 2

   Deduplication and provenance hashing

   Each creative receives a content hash so identical reposts collapse into a single record with multiple impressions. Banner photos and videos are mirrored to a permanent content-addressed URL within minutes. Original Telegram media expires after one hour; the mirrored copy survives for the lifetime of the archive.

3. 3

   Niche, geo and payment-source classification

   A regex classifier (open source, see /info glossary) maps each ad to one of 25+ verticals. We tag the country where the ad was delivered and detect whether the advertiser paid in Gram (Gram cabinet, formerly TON), EUR (Telegram Ads Cabinet) or Stars (XTR / cashier vertical). Every classification decision is reversible and auditable.

4. 4

   Permanent permalinks and CC-BY-4.0 export

   Every creative is given a stable /ads/<id> URL that does not change for the lifetime of the archive. Aggregate datasets are exported nightly as CSV, JSON and JSON Feed v1.1. All exports carry a CC-BY-4.0 license block — cite freely with attribution to adsguru.org.

What we do NOT collect

Open research only works when the boundaries are clear. We are not a private-content scraper, not a black-box rebroadcaster of Telegram, and not a brand-monitoring tool that exfiltrates restricted data. The list below describes what is explicitly out of scope.

• ×Private chats, secret chats, or any direct messages.
• ×Invite-only channels that have not opted into ads.
• ×Anything behind a Telegram safeguard (slow mode, premium-only, restricted).
• ×Personal data of users who view ads — we observe ad surfaces, not viewers.
• ×Hidden creatives flagged via DMCA or 152-FZ takedown — they are kept in DB for completeness but excluded from public feed and sitemap.
• ×API endpoints behind authentication that Telegram does not publish.

Built for open research

The same dataset, surfaced through six different lenses. Each use case below uses the same underlying CSV/JSON exports — only the angle of analysis differs.

• Academic citation

  Every dataset has a stable DOI-style URL, author, and CC-BY-4.0 license block — cite it in a paper or report.

• Open-source investigations

  Trace an advertiser across channels, countries, and time. Full permalink history, no data removed.

• Market research

  Cross-country niche comparison, spend estimates, channel ranking. Machine-readable API at /api/v1/ads.

• Regulatory monitoring

  Watch ad-policy compliance per market — gambling/financial/health verticals across regulated geos. Time-stamped permalinks for evidence trails.

• Newsroom investigations

  Follow money: which advertisers spend in which countries, in which languages, around political moments or product launches. Citable for journalism.

• Builders and tooling

  Free public REST API and JSON Feed v1.1. Build dashboards, alerts and analyses on top — same data the production app uses.

How to cite

Suggested citation block for academic papers, regulatory filings or news pieces. CC-BY-4.0 means citation is required; commercial use is allowed.

adsguru.org. (2026). Telegram Ads Creative Archive [Dataset].
Retrieved from https://adsguru.org/datasets. Licensed under CC-BY-4.0.

For dataset-specific DOIs (per-month frozen snapshot), open the corresponding entry on /datasets and use its permalink. The aggregate live archive uses the URL above.

Frequently asked questions

Open data on Telegram Ads — common questions

One dataset — many tools

Pick the tool for your job

Each domain is built around a specific job: finding a single creative, tracking launches live, building a market deck, or piping data into your BI. Same database, tailored UI per role.