Why Your Perfectly Personalised Email Got No Response (Hint: It Was Never Delivered)

A BDR on a team we work with spent forty minutes on one outreach message. She watched three of the prospect's podcast appearances, read a quarterly investor letter, and found a specific product line she thought their sales ops team would love. The message was good — not generically “loved your episode” good, actually good. She never heard back.

Three months later, the same prospect replied to a cold intro from a different rep at a different company, using a one-line Calendly link and zero personalisation. She was devastated. We pulled her send logs. The email had been quarantined by the recipient's Microsoft 365 tenant filter and never surfaced in the inbox, not even Junk. The other rep's sending domain had a clean reputation. Hers did not.

TL;DR

Personalisation is the single highest-leverage reply-rate tactic — but only after your message reaches a folder a human reads. Time spent on research is wasted if the envelope fails the first sniff test at the receiving server.

The math of personalised sends

Let's put actual numbers on it. Suppose you spend 10 minutes per prospect on personalisation. For a 100-prospect sequence, that's 16 hours of deep research.

Generic cold:      list 1000 × inbox 70% × reply  0.5% = 3.5 replies
Personalised cold: list  100 × inbox 35% × reply 4.0%  = 1.4 replies
Personalised cold: list  100 × inbox 90% × reply 4.0%  = 3.6 replies

Same copy, same list, same personalisation effort. The only variable is placement. 35% inbox turns 16 hours of research into fewer replies than the generic blast. 90% inbox makes the effort worth it.

Why the personalised email was the one that got filtered

Ironically, heavily personalised messages often look more suspicious to filters than templated ones:

One-to-one sending pattern without thread history. Filters score first-touch conversations carefully.
Specific entity references (your prospect's company name, title, recent event) that match known phishing patterns.
Unusual subject lines crafted to stand out — the same thing makes them stand out to Bayesian classifiers.
Low-volume sending domain that has not built reputation with the recipient's filter.

A templated Calendly-pitch from a warmed-up domain can slide into Focused inbox while your researched masterpiece is sitting in Quarantine waiting for an admin to release it. Nobody releases those messages. They get auto-purged after 30 days.

Check placement before you research, not after

The standard SDR workflow goes: research → write → send → wait → despair → ask manager for copy review. The deliverability-first workflow inverts two steps:

Warm up your sending domain for 3–6 weeks to a stable baseline.
Run a placement test on your template structure before touching personalisation.
Fix authentication, volume shape and content fingerprints until placement is >85% on Gmail and Outlook.
Now invest the 10 minutes per prospect on research. Now it compounds.

Test the envelope, not just the letter

Inbox Check tests placement with your real sending infrastructure — same domain, same IP, same headers — so you know whether the personalisation effort will actually reach humans. Free, no signup. For automated re-tests on every template change, use the API.

The hybrid that actually works

The highest-ROI pattern we see in 2026 is a hybrid: a short, templated opener (high placement) followed by a personalised second message in the same thread (inherits reputation from the first). The first message gets filtered gently because it looks normal. The second message rides the thread's reputation score and gets to be as bespoke as you like.

This only works if the first message actually gets replied to — or at least opened — which requires the same deliverability discipline as any other cold email.

When personalisation finally matters

Once placement is stable, personalisation multiplies reply rate roughly 2–4x over a templated baseline, depending on ICP specificity. The highest leverage happens at the list-cutting stage, not the writing stage: a 50-prospect hyper-targeted list with a half-personalised message usually beats a 500-prospect generic list with a fully-personalised one, because the per-prospect attention budget is finite and placement costs are fixed.

Signals that you're being filtered without knowing

Open rates suspiciously uniform around 20–30% regardless of subject line (that's bot-scanner noise, not humans).
Zero “unsubscribe” or “wrong person” replies from a 200+ send. Humans reply with those. Filters absorb them.
Sudden reply-rate collapse after a specific date — usually an IP or domain reputation event.
Replies concentrated in one provider (e.g. only Gmail, never Outlook). That is placement asymmetry.

FAQ

How much personalisation is too much?

Enough to be specific and genuine, little enough that the message still scans like a normal one-to-one email. Five sentences is usually the sweet spot. Paragraph-long explanations of ‘why you’ raise filter flags and reader fatigue simultaneously.

Does AI-generated personalisation hurt placement?

Not directly. Filters score patterns, not provenance. What hurts is AI output that accidentally looks templated across recipients (same sentence structure, same closing) — that does show up as a low-entropy batch.

Should I skip personalisation on first touch?

For high placement scenarios, yes, often. Keep first touch short and normal-looking. Personalise touch 2 or 3 when you have thread context and the prospect's filter has scored your exchange.

How do I know if my specific message was quarantined?

Send to your own seed mailbox on the same domain type (e.g. a test Microsoft 365 tenant). Many tenant-level quarantines are invisible to both sender and recipient — you have to own the destination to see them.

Your perfectly personalised email got no response — because it was never delivered