# Connect does not guess email addresses

No pattern-generated contact is ever produced, tested or sent to. There is no `first.last@domain` construction, no permutation list handed to a verifier, and no setting that turns one on. An address is found in a public source with that source recorded, or the prospect has none and gets a different next action. The rule costs volume on purpose.

- **Status:** Available
- **Audience:** both
- **In the app:** #/prospects
- **Last verified:** 2026-09-10
- **Canonical:** https://connectbyjbrh.com/docs/prospects/no-guessed-email/

## What the rule forbids, precisely

Guessing has a recognisable shape, and it is worth naming each part of it so there is no ambiguity about which parts are absent.

| Common step | Present? | What happens here |
|---|---|---|
| Build candidates from a name and a domain | No | Nothing is built from a name |
| Score patterns by what usually works at that company size | No | There are no patterns to score |
| Send each candidate to a verification service | No | Nothing unfound is verified, because nothing unfound exists |
| Keep the one that did not bounce | No | A bounce is not treated as a research method |
| Send to the survivor | No | Outreach requires a sourced address and a passed compliance check |

The fourth row is the one that does the damage in practice. Verification by sending is still sending: the mail leaves your mailbox, lands somewhere, and is judged by a receiving system that keeps a memory of your domain.

## Why the rule exists

Three reasons, in the order they bite.

1. **The cost lands on your mailbox, not on the tool that guessed.** Invalid recipients, spam traps and complaints attach to the sending domain and the sending mailbox. A guessing tool produces a longer list and leaves the reputation damage with you, months after the list stopped being interesting.
2. **A guessed contact is a claim with no evidence, which this section does not permit.** Every other statement on a prospect record carries the source it came from. A string derived from arithmetic on a name has no source; admitting it would mean the record has one field where the evidence rule quietly does not apply.
3. **A wrong guess reaches a real person.** `first.last@` at a common domain shape frequently belongs to somebody real — often somebody at a different company, sometimes a former employee, occasionally a shared alias that a dozen people read. Cold mail to them is not a near miss; it is mail to an uninvolved stranger.

> **Note** This is recorded in the capability registry as `no_guessed_email` and in the list of claims no page in this documentation may make. It is a product decision, not a limitation waiting for a release.

## What is done instead

1. Research reads public material for a contact the organisation itself publishes, and records where it was found.
   - Result: If one exists, the prospect is contactable and the provenance is visible on the record — see [Evidence for a contact address](/docs/prospects/contact-evidence/).
2. If none exists, the prospect keeps everything else — the research, the judgement, the fit reason, its resolution against your relationships.
   - Result: Nothing is thrown away, so the same business does not have to be researched again from scratch.
3. A next action is set on a channel that does exist: a call, a published enquiry form, an introduction, or a wait for re-research.
   - Result: The prospect appears in the follow-up queue with a reason attached rather than sitting in the list as an unexplained dead row.
4. Freshness brings the record back later.
   - Result: Businesses publish contact details they did not publish last quarter, and a prospect can become contactable without anybody re-running the search.

## The trade, stated honestly

This costs volume. A workspace that runs the same criteria through a tool that constructs addresses will show more rows with a green tick beside them, and on the first afternoon that feels like the better product. The difference shows up later, in three places: the bounce rate on your mailbox, the share of replies that begin *you have the wrong person*, and how much anybody trusts the list a quarter after it was built.

**What you lose** — Rows. Sometimes a majority of them, in trades that do not publish addresses.
**What you keep** — A sending domain nobody has trained a filter against, and a list where every address on it belongs to the business beside it.
**What you can still do** — Every other channel. An uncontactable prospect is a phone, form or introduction task, not a discarded one.
**What never changes** — There is no override. No plan, allowance, mode or support request enables address construction.

If the count of contactable prospects is too small to work with, the productive move is a different population or a different channel — not a different sourcing standard. [A good prospect has no address](/docs/troubleshooting/prospect-no-email/) walks through the specific case where the business is right and nothing reachable was ever published.

## Questions

### Can I paste a contact I found myself?

Yes, and it is stored as your statement with you as its origin. The rule is about Connect not inventing one; a person who knows where to write is a legitimate source and is recorded as such.

### What about a verification service that checks one without sending?

The problem is upstream of verification. A verifier can only grade a string somebody constructed, and there is no construction step here for it to grade. Where a contact was found in a public source, its provenance is the record — not a third party's opinion of the string.

### Does this apply to phone numbers too?

The same principle: a number is recorded when a source publishes it, never derived. Nothing about a business's name or website produces a number, and no range is dialled to find out which one answers.

### Is a role mailbox like info@ acceptable?

If the organisation publishes it, yes — it was found, not constructed, and it carries its source. The distinction the rule draws is between something somebody published and something somebody computed.

## Related

- [Contactability](https://connectbyjbrh.com/docs/prospects/contactability/)
- [Evidence for a contact address](https://connectbyjbrh.com/docs/prospects/contact-evidence/)
- [Prospecting in Connect](https://connectbyjbrh.com/docs/prospects/)
- [A good prospect has no address](https://connectbyjbrh.com/docs/troubleshooting/prospect-no-email/)
- [Evidence-first prospecting without guessed emails](https://connectbyjbrh.com/research/evidence-first-prospecting/)
- [Email in Connect](https://connectbyjbrh.com/docs/email/)

## What this page is based on

- docs-source/sources/CHANNELS.md §4 — Connect does not guess email addresses
- `docs-source/facts.py` — CAPABILITY_STATUS no_guessed_email; FORBIDDEN_CLAIMS
- docs-source/sources/CHANNELS.md §4 — a prospect with no address keeps a different next action