An advertiser paid for a placement on a channel with 80,000 subscribers and 8% ERR. They got fewer clicks than a follow-up placement on a 5,000-subscriber niche channel. Afterward they checked TGStat: the big channel had seen three vertical subscriber spikes with no matching engagement lift. A proper analysis would have shown 31% of the audience was bot accounts.
Subscriber count and ERR are both visible before you spend money. Neither one is reliable on its own. Here's how to read them correctly, and what to do when the numbers don't add up.
ERR: what the number is actually measuring
Engagement Rate Reach is views divided by subscribers, times 100. Take the average of the last 10 posts — not the top performer, not the pinned announcement. Averages are honest; individual posts aren't.
| Channel size | Healthy ERR | Warning zone |
|---|---|---|
| Under 10K | 20–40% | Below 15% |
| 10K–100K | 10–20% | Below 8% |
| 100K–1M | 5–10% | Below 3% |
| Over 1M | 2–5% | Below 1.5% |
One additional check that takes ten seconds: look at the reaction count on recent posts. A channel with 30,000 views and 6 reactions isn't a healthy channel. Real audiences leave reactions at roughly 0.5–1% of views, often more in active communities. Six reactions on 30K views means almost no one reading is actually engaged.
ERR below benchmarks is a signal, not a verdict. The same low number appears on channels that bought fake subscribers, channels with a burned-out audience from overselling ads, and channels that just publish boring content. The growth chart separates them.
Reading the subscriber growth chart
Organic growth looks like a gradual climb with occasional spikes — the spikes align with real placements, viral posts, or press mentions. Open TGStat or Telemetr for any public channel and look at the 6-month graph.
What bot-purchased growth looks like: a vertical jump of 5,000–20,000 subscribers over 1–3 days, with no visible cause. Views don't move. Then, 7–14 days later, a matching drop — the bot service's "inactive account cleanup" cycle. If you see multiple wave patterns like that — rise, plateau, partial fall — the channel has been buying subscribers repeatedly.
Competitor attacks look similar in shape but different in context: the jump happens right before or during a significant commercial negotiation, and the channel owner usually has no explanation for it. For a channel that's been attacked rather than bought fake subs itself, the attack pattern typically follows a single invite link being exploited.
Subscriber quality: what only a direct scan shows
ERR and growth history tell you something is wrong. They can't tell you exactly what. That requires checking the subscriber list against a database of known bot accounts.
TGuard's channel audit does this: its database covers 10+ million accounts flagged across three years of monitoring bot attacks on connected channels. The audit produces three numbers:
- Bot accounts — subscribers found in the known-attacker database
- Dead accounts — profiles with zero activity for 6+ months
- Deleted accounts — joined the channel, then deleted their Telegram profile
From that, a single quality score from 0 to 100. The median across channels over 10K subscribers is about 7% bots and 15% dead accounts. A channel at 25%+ bots has a real problem. At 30%+, it's not a question of whether something happened — it's a question of whether it was an attack or deliberate inflation.
Some channels run a parallel view-boosting service to mask a high bot percentage — they pay to inflate views so ERR looks normal. ERR analysis misses this entirely. A subscriber-level bot scan doesn't.
When a quick look is enough
For small ad placements — under $100, channels you've worked with before, or audiences where brand fit matters more than scale — a manual check is usually sufficient. Calculate ERR manually, look at the growth chart on TGStat, scroll through 50–100 subscriber profiles and count the obviously empty ones.
A full audit makes sense when: the placement is significant enough that a wasted budget hurts, you're evaluating a channel for acquisition, or you're the channel owner and you want to show a buyer clean numbers. A documented audit report with specific percentages is a different conversation than "our audience is real, trust us."
How to run the analysis
- Open TGStat or Telemetr for the channel — check the 6-month growth graph for spike-and-drain patterns.
- Calculate ERR manually: average views on the last 10 posts ÷ subscribers × 100. Compare to the table above.
- Check the reactions-to-views ratio on recent posts — below 0.3% is a red flag.
- Scroll 50–100 subscriber profiles manually for a rough sense of profile quality.
- For a full audit: add @channel_guardian_bot as an admin, run the subscriber audit, get the bot/dead/deleted breakdown and quality score.
The channel with 80,000 subscribers isn't automatically better than the one with 5,000. A 5K channel at 35% ERR with a clean subscriber list consistently outperforms a 100K channel at 6% ERR with 25% bots. The numbers that matter are the ones most people don't check.
Frequently Asked Questions
ERR = average views on the last 10 posts ÷ total subscribers × 100. Use the average, not the peak post. Healthy benchmarks: under 10K subscribers — 20–40%, 10K–100K — 10–20%, 100K–1M — 5–10%. Below the lower bound, something is suppressing views — could be bots, could be a burned-out audience.
Partially. ERR and the growth chart of a public channel are visible on TGStat and Telemetr without admin access. The actual subscriber list — and a direct bot percentage — requires the channel owner to add an audit tool as admin so it can scan members against a known-bot database.
Based on TGuard data, the median across channels over 10K subscribers is around 7% bot accounts and 15% inactive accounts. Over 20% bots is a red flag. Over 30% means there's been either a deliberate bot attack or the channel has been buying fake subscribers.