The problem with "warm"
Walk into any UK B2B sales conversation in 2026 and the word "warm" gets used to mean five different things — sometimes in the same sentence. Some teams call any prospect who has been emailed "warm". Others reserve it for prospects who have replied. Most sit somewhere in between, with a vague "they've shown interest" definition.
The fuzziness costs money. If your sales team treats opens as "warm" and reaches out via phone to every opener, they waste hours calling bots and inbox security scanners. If they only treat replies as "warm", they miss strong signal from prospects who clicked but did not type.
The bot-open problem
Here’s the inconvenient truth in 2026: roughly half of every "open" your cold-email tool reports never involved a human eyeball. Modern enterprise email security (Microsoft Defender for Office 365, Mimecast, Proofpoint, plus increasingly Gmail's Apple-Privacy-style image proxies) opens incoming email from unknown senders before the human ever sees it, to scan for malware and phishing.
From the cold-email tool's perspective, that scan looks identical to a human open. Same tracking pixel hit, same timestamp, often even a click on the unsubscribe link as the security tool tests for malicious redirects.
This means: if your cold-email tool reports a 25% open rate in 2026, the actual human open rate is probably 12–15%. And if it ranks prospects by "opened the most", the top of the list is sometimes a Microsoft tenant's automated security scanner, not a real prospect.
The 4-tier warm-scoring model
Here is a clean way to rank engagement signals, ordered from strongest to weakest:
| Tier | Signal | What it means | Bot risk |
|---|---|---|---|
| 1 | Direct reply | A human typed words back. Strongest possible signal. | Near-zero |
| 2 | Multi-touch engagement | Same prospect engaged with 2+ campaigns over time. | Low — pattern-matched out |
| 3 | Single click on a content link | Human chose to click an in-body link to read more. | Low — bot scanners avoid most non-unsubscribe links |
| 4 | Open without click, or click on unsubscribe only | Possibly human, possibly security scanner. | High — these need filtering |
A clean morning shortlist promotes Tier 1 to the top, surfaces Tier 2 prominently, includes Tier 3 with context, and either filters out Tier 4 entirely or ranks it last with a clear "low signal" tag.
How AI Email scores warm
AI Email applies bot-filtering at the open level (we cross-reference open patterns with known security-scanner signatures and remove them before scoring), then ranks the remaining prospects on the 4-tier model. The 10 prospects in your morning report at 7am are the 10 highest-scoring after filtering.
In practice, on a typical day for a typical UK B2B service business: 3–5 of the 10 are Tier 2 or Tier 3 (multi-touch or content-clicked), 1–3 are direct replies, and 2–4 are filtered Tier-4 opens with strong job-title fit that we surface even at lower engagement because the persona match is strong.
What this means for how you work the list
- Direct repliers — call or reply within 24 hours. These are the most perishable leads on the list.
- Multi-touch engagers — reach out via LinkedIn first (less invasive), email or call within the week.
- Content clickers — these are research-stage prospects. Treat them as nurture-list, not as immediate phone calls.
- Filtered opens with strong persona match — a soft-touch follow-up email or LinkedIn connect is appropriate. A phone call usually is not.