Large groups create the illusion of abundant leads. A short source review often reveals that the visible count includes spectators, bots, old members, unrelated roles, and people whose context makes contact inappropriate.
Complete one worksheet per source and compare predicted yield with the reviewed result. Over time, the model becomes an owned sourcing advantage.
How to use this resource
Evaluate a Telegram group before collecting member data. Score topic fit, observable role density, recent conversation quality, commercial appropriateness, administrator rules, data sensitivity, duplication, and expected reviewed yield. Reject sources that cannot support a clear audience hypothesis.
Included in the template
01A weighted source scorecard.
02A sampling procedure.
03Red flags that override a high score.
04A yield-learning loop.
1. Source identity
- Group: [name and canonical link]
- Observed on: [date and reviewer]
- Target hypothesis: [role + problem + offer]
- Access: [public, approved membership, or other legitimate access]
- Rules: [promotion, research, contact, automation, and admin guidance]
2. Score the source
Multiply each 0-to-5 score by its weight, then divide by five. The total is a comparison aid, not permission to contact.
| Dimension | Weight | 0 | 3 | 5 |
|---|---|---|---|---|
| Topic fit | 20 | Unrelated | Mixed | Directly aligned |
| Target-role density | 20 | Rare | Visible minority | Core membership |
| Recent useful activity | 15 | Dormant or noisy | Intermittent | Sustained expert exchange |
| Observable intent | 15 | No relevant signal | General pain | Specific active need |
| Contact appropriateness | 15 | Clearly inappropriate | Uncertain | Reasonably expected |
| Rule compatibility | 10 | Prohibited | Ambiguous | Explicitly compatible |
| Data-risk profile | 5 | High sensitivity | Mixed | Low and business-relevant |
3. Review a bias-aware sample
- Sample recent conversations across several days and times.
- Separate posters from passive member counts.
- Record evidence for eligibility, ineligibility, and uncertainty.
- Estimate duplication with sources already used.
- Have a second reviewer score borderline sources.
4. Apply veto conditions
- Community rules or administrators prohibit the intended use.
- The audience is primarily minors or otherwise vulnerable.
- Relevance depends on sensitive or speculative inference.
- The group is dominated by bots, promotions, or recycled content.
- A lawful and fair use cannot be explained.
5. Forecast and learn
| Forecast | Actual | Learning |
|---|---|---|
| Reviewed observations | [count] | Was access and activity as expected? |
| Eligible rate | [percent] | Which criterion removed most records? |
| Useful reply rate | [percent] | Did source context predict relevance? |
| Qualified outcomes | [count] | Did this source produce downstream value? |
| Control events | [count] | Should the source be paused or excluded? |
Research note
Scoring improves consistency but cannot remove sampling bias or create permission. Validate access, platform terms, community rules, fairness, and lawful processing separately.
Use it in the next review
The best source is not the biggest group. It is the group whose context supports a narrow, testable and respectful acquisition hypothesis.
Keep source evaluation connected to downstream outcomes
TeleBoost links group discovery and reviewed lists to campaigns, conversations, tickets, teams, and analytics.
Continue the operating system