A client asked last month: "Which channels are talking about my brand, and who starts the conversation?" Sounds like a five-minute Google Alerts answer. It took four hours - and the four-hour shape is the actual product, so here's the walkthrough.
Hour 1 - find the surface, not the mentions. Google Alerts covers web-visible text; Telegram conversations live under it. I seeded a candidate channel list from three public sources: (1) the brand name and two common misspellings in Telegram's public search, (2) channels linked from the brand's own web pages and press kit, (3) channels found via one known mention's forward header (forwarded posts carry t.me/<channel> provenance in their HTML). Seeded 47 channels; 9 turned out to have genuine brand conversation in the window.
Hour 2 - quantify without reading everything. 9 channels x ~90 days via the public t.me/s/ preview layer = a few thousand posts. Reduced every post to (timestamp, channel, brand-mention?, sentiment-cue, forward-origin?) with a keyword dictionary. Result: 312 brand-mentioning posts. The raw count is worthless; the structure is the product.
Hour 3 - who starts, who amplifies. Forward-origin is the key column: 312 mentions collapsed into 28 origin events (the post that first put a topic into the ecosystem) with 284 amplifications. The top 3 origin channels - not the biggest channels - drive every brand narrative in the corpus. One has 4,100 subscribers; the biggest channel, 240k, never originates, only re-broadcasts. That asymmetry is the answer to the client's real question: "who do we talk to."
Hour 4 - narrative clustering + the brief. Origin events cluster into 5 narratives (price complaints, a counterfeit scare, a feature praise wave, a competitor comparison, one geopolitical spillover). The counterfeit scare is the actionable one: originated in channel #2, amplified within 6 hours, peaked in 2 days, and is still alive in the corpus 40 days later - meaning new readers keep arriving at it via search. Each narrative: origin post, amplification timeline, top amplifiers, live/dead status. That's the brief.
What the client paid for is exactly that shape - origin/amplify separation plus narrative lifecycle - which no mention-count tool produces. The pipeline (seed sources, forward-header provenance parsing, origin-detection rules, brief template) is the core of the Telegram & Web OSINT Bundle ($5). A free sample brief shows the format; 48-hour custom runs of exactly this walkthrough are the custom report ($25).
Runs on public sources, free GitHub Actions cron - no server, no paid APIs.
How did your mention tool handle the amplify channel with 240k subscribers? It counted 284 mentions. The ecosystem counted 28 events.
Top comments (0)