Workflow essentials
A deterministic matching hierarchy.
Rules for preserving multiple source observations.
Conflict handling and client isolation.
A merge audit record that keeps suppression trustworthy.
Overlap is not dirty data. The same person appearing in several relevant communities may be one of the strongest signals available. Deduplication should remove duplicate identities while preserving the overlap as evidence.
The dangerous implementation is a spreadsheet Remove duplicates on username. Usernames can change, names collide, blanks are common, and client contexts may be intentionally isolated. A reliable process needs identity rules and ownership boundaries.
Operating rule
Deduplicate Telegram prospects with stable platform identifiers inside a defined ownership scope. Merge observations, not source history. Preserve every group, extraction time, qualification reason, campaign event, and objection. Similar usernames or names are review candidates, not automatic matches. When evidence conflicts, keep records separate because a false merge can corrupt contact history and suppression.
Define the scope before the match
Identity may be global, but ownership and purpose are not. A technical match must not automatically grant one client access to another client's lead, conversation, or source history.
| Scope | Typical rule | Reason |
|---|---|---|
| Personal workspace | Deduplicate within the owner's data | Avoid accidental mixing with team data |
| Team workspace | Deduplicate within the active team and its permitted sources | Preserve shared contacted state |
| Agency client | Do not merge across client workspaces by default | Protect contractual and campaign boundaries |
| Global suppression | Match only the minimum identifier needed for objections | Enforce preference without exposing full lead data |
Use a matching hierarchy
Telegram's API models users through identifiers while usernames are profile attributes that can be updated. That is the correct architectural clue: use a stable ID for identity and treat visible labels as versioned evidence.
- Exact stable Telegram ID: automatic match within the authorized scope.
- Trusted import mapping: match when a verified mapping links an external record to that Telegram ID.
- Current username plus corroboration: manual review only, because usernames can change or be reused.
- Name, biography, organization, or photo similarity: never enough for automatic merge.
- No reliable identifier: keep separate and record the uncertainty.
Merge the person, append the observations
Last-write-wins is especially dangerous for suppression and ownership. A newer import should never turn an objected contact back into an eligible lead. A merge should choose the most restrictive applicable contact state until a documented review changes it.
| Data type | Merge behavior |
|---|---|
| Stable identity | One canonical lead ID inside the scope |
| Source groups | Append each source with first and last observed time |
| Qualification | Keep decision history and criteria version |
| Contact events | Append immutable campaign and message events |
| Suppression | Use the strongest applicable objection state |
| Owner and stage | Resolve by explicit policy, never last-write-wins silently |
| Conflicting profile fields | Keep time-stamped values and flag for review |
Handle collisions as incidents, not cleanup
Quarantine the records, stop relevant campaigns, preserve the merge log, and repair downstream contact state. Do not delete the evidence that explains how the collision occurred.
- Two records share a username but have different stable IDs.
- One record appears in workspaces that must remain isolated.
- Contact history belongs to different people after an earlier false merge.
- A suppression state has been lost or narrowed.
- A merge combines conflicting organizations or roles with no reliable explanation.
Measure deduplication quality
- Automatic matches by stable ID.
- Manual match candidates and reviewer agreement.
- False merges discovered.
- Duplicate contacts prevented.
- Suppression conflicts detected.
- Cross-client matches blocked.
Useful output: report both unique people and source observations. The first controls contact. The second preserves the market signal that several relevant communities independently contain the same person.
Research note
This guide describes identity resolution inside authorized data scopes. It does not recommend combining client datasets or using visible profile similarity as proof of identity.
Make the handoff operational
Change the deduplication question from Which rows look the same? to Which observations refer to the same stable identity inside this permitted scope?
Good deduplication reduces repeated contact without erasing the evidence that made a prospect interesting.
Keep one contact history without flattening sources: TeleBoost lead management connects lists, contacted state, campaigns, and replies inside personal or isolated team contexts.
Continue the operating system