Effort is the scarce resource in outbound, not contacts. A personally researched approach to one company costs something real; a templated blast costs almost nothing and is worth about as much. Tiering is how the system decides which accounts deserve which.
Why tiers exist#
Without tiers there are only two strategies available, and both are bad. Treat every account like it matters and you will contact eight companies a week. Treat none of them like they matter and you will contact eight hundred, badly.
A tier is an answer to a narrow question: given what we can prove about this account right now, how much of a human’s attention is it worth? It is recalculated as evidence changes, so it is a current judgement rather than a permanent grade.
The four tiers#
| Tier | State of the evidence | Position |
|---|---|---|
| A | Window open, three or more conditions met, at least two strong records | Contact this week, personally, with the evidence in the message |
| B | Window open, two conditions met, evidence mostly moderate | Contact this month, sequenced, with one specific claim per message |
| C | Fit confirmed, no open window | Do not contact. Watch for a trigger. |
| D | Fit not confirmed, or an exclusion applies | Out of scope. Held for audit, not worked. |
What each tier gets#
| Treatment | A | B | C | D |
|---|---|---|---|---|
| Scout dossier | Automatic | On request | No | No |
| Message drafting | Per account, from that account's evidence | Per account, one claim reused across steps | — | — |
| Sequence length | 2 steps, then a human decides | 3–4 steps | — | — |
| Human approval | Always | Above the autonomy threshold only | — | — |
| Re-qualified | Daily | Every 3 days | Weekly | Monthly |
| Counts against send caps | Yes | Yes | No | No |
Two of these are worth dwelling on. A-tier sequences are short. Two steps, then a person decides what to do next — because if the evidence was strong enough to justify a personal approach, the follow-up decision is worth a person too. C-tier accounts do not consume send capacity. Nothing is sent, so nothing is spent, and the capacity stays available for the accounts that earned it.
Promotion and demotion#
Accounts move between tiers on evidence, and every move is recorded with the record that caused it.
- C to B — a new signal stacks with an existing one and a window opens. The commonest promotion, and the one the whole watch list exists to catch.
- B to A — a further family fires, or a moderate record is upgraded to strong after verification against a better source.
- A or B to C — the window decayed. No rejection is recorded; the account simply stops being urgent.
- Anything to D — an exclusion matched, or you rejected the account with a reason. Reasons generalise; a single name does not.
- D back to C — only when the exclusion that put it there expires, such as a competitor contract reaching its renewal date.
Movement is damped. An account that has moved tier within the last 72 hours will not move again on a single weak record, because tier churn produces the worst possible outcome: contacting a company twice in a week from two different campaigns because it crossed a boundary and crossed back.
{
"account_id": "ACC-19022",
"from": "C",
"to": "B",
"at": "2026-07-24T06:02:11Z",
"caused_by": ["EV-2308"],
"reason": "Expansion evidence stacked with existing hiring evidence; window opened.",
"window": { "state": "open", "estimated_close": "2026-10-22" },
"damped_until": "2026-07-27T06:02:11Z"
}Tier caps#
Caps stop the system flooding you with A-tier accounts it cannot justify. They are limits on how many accounts may occupy a tier at once, not limits on sending — sending limits are separate and live in sending.
- A tier
- 20 accounts per seat, per week
- B tier
- 120 accounts per seat, per week
- C tier
- uncapped
- Damping
- 72 hours between tier moves
- Cap behaviour
- lowest-confidence account demotes to B
When the A cap is reached, the lowest-confidence A account is demoted to B rather than the newest one being refused. The result is that A always contains the twenty best-evidenced accounts available this week — which is, deliberately, the same twenty the product exists to name.
Next: qualification — the argument that produces the verdict a tier is assigned from.