5-step pipeline
of top media buyers

When you look at public courses on Telegram cold outreach, you get the standard set: "scrape a list, write spintax, send". That works at a baseline level — 2-5% conversion, constant rework due to bans.

What the teams that consistently convert at $80K-200K/month do is a fundamentally different pipeline. Below is a step-by-step breakdown based on real agency cases that ran through TG:ON (with their permission to publish anonymously).

Nicheing: 3-5 channels instead of 50

The most common mistake — "grab as many channels in the niche as possible". Scraping 50 channels → 50K leads → mass outreach → 2-3% conversion.

What the top operators do: pick 3-5 competitors with maximally overlapping audiences. The list comes out at 5-8K instead of 50K, but conversion is 10× higher (20-30% at this stage).

How they pick those 3-5:

1. Their product and ICP (ideal customer profile) are as niche as possible — not "everyone interested in crypto", but "smart-contract auditors on Solana"
2. They find channels where that audience is already gathered — usually small (3K-30K subscribers) vertical communities
3. From those, they pick the 3-5 liveliest ones (regular discussions, lots of comments, no bots)

Pre-warmup through activity in target channels

Before starting outreach, the sender account is present in the target channels for 2-3 days: comments on posts, leaves reactions, occasionally writes something relevant in discussions.

Why: when this account DMs a lead, the recipient sees mutual channels. Telegram shows "member of channel X, Y, Z — 3 mutual with you". That's a strong trust signal:

• Recipient sees "we're in the same community" → sharply higher opens
• Telegram sees social graph overlap → lower probability of "stranger spamming" flag
• If you've been commenting on posts — your handle may already have been in front of the recipient

Pipeline into a channel, not DMs

The most counter-intuitive step. Most people send outreach straight into direct messages. The top operators do it differently:

1. Create a warmup channel with niche content (5-10 quality posts, so it doesn't look empty).
2. Invite the scraped leads into that channel (using auto-inviter, but with limits).
3. Wait 2-3 days — leads either read the channel's content or unsubscribe. Those who leave — clearly not interested, instant filter.
4. After 2-3 days, DM the ones who stayed: "Noticed you're in my channel, wanted to ask — [offer]".

The effect: a first contact from a familiar channel is much warmer than "off the street". Response rate 3-4× higher than cold DMs.

AI-triage: Qualifier + Closer

When leads start replying, the classic problem appears: most replies aren't worth human time. "Tell me more", "too expensive", "not interested", "not ready yet". Each takes 2-3 minutes — × 50 replies = 2-3 hours of low-value work.

The top operators use an AI agent on the front line:

• Qualifier — an LLM agent that runs the first dialog. Asks clarifying questions, uncovers the need, filters out non-targets.
• Triage rules — if replies match the ICP (budget, role, timeline), the Qualifier escalates to human chat. If not — it closes politely ("we're not a fit right now, we'll let you know when we have something relevant").
• Closer — either a human or a more advanced AI agent that handles price objections and pushes to payment. Works only with triaged leads.

The effect: a 2-person team handles the inbound from 500-1000 messages sent per week.

48h retargeting (not 24)

Leads who didn't reply to the first touch come back with a second touch from a different angle. Timing is critical:

• Not at 24 hours — too pushy, reads as spam
• Not at a week — the lead has already forgotten the first message, no context
• At 48-72 hours — the sweet spot. Reminds them enough to restore context, not enough to feel pushy

The second touch must come from a different angle: if the first was "we have a service for X", the second is "saw you're doing Y, thought we might overlap on Z". Not a reformulation of the first, but a fresh thought.

Conversion on retargeted touches is usually 50-80% of the first — meaning if the first got 20 replies out of 100, retargeting adds another 10-16. Free leads, from the same list.

Numbers for the full pipeline

These numbers are the median across cases with an active pipeline. Variance: higher in crypto, lower in edu-tech. But the proportions between steps are usually similar.