Diagnosing underperformance
TL;DR
- Diagnose in layers, outside in: delivery → list → offer → copy. Fixing copy when the problem is delivery is the most common wasted week in outbound.
- The signal that tells layers apart is where the funnel breaks, not how bad the final number is.
- A campaign with no segment-level reporting cannot be diagnosed. Instrument first.
The layer order
1. Delivery. Is the mail arriving at all? Bounces above about 2%, a reply rate that fell while nothing changed, or out-of-office replies continuing while real replies stop all point here [1]. Because placement failure is invisible — no error, no notification — always rule it out first. Fixes are in Deliverability and compliance essentials and Campaign scheduling and pausing.
2. List. Are these the right people? The tells: replies saying "wrong person" or "we don't do that", high unsubscribes with low engagement, or one segment carrying the whole campaign while another is silent. Company size alone moves reply rates threefold [2], so a "bad" average is often a targeting artefact. Fixes: Audience and lists.
3. Offer. Is there a reason to engage? The tell is polite silence — mail delivered, right people, nothing happening — and the diagnosis is that the email asks for something and gives nothing (Offer design). This layer is the most commonly skipped, because silence looks like a copy problem.
4. Copy. Only once the first three are sound. Copy is a multiplier on a base the other layers set, so it is the last place to look and the first place everyone looks.
Reading the funnel
Each layer breaks at a different point, which is what makes the diagnosis mechanical:
| What you see | Most likely layer |
|---|---|
| High bounce rate | List quality (verification) |
| Delivered, near-zero replies, engagement dropped suddenly | Delivery / placement |
| Replies arrive but say "wrong person" / "not us" | List targeting |
| Delivered to the right people, polite silence | Offer |
| Replies arrive but almost none are positive | Offer or segment fit |
| One segment performs, another does not | Targeting or message-segment fit |
| Everything reasonable, results merely mediocre | Copy — and see the sample-size caveat below |
Note the trap in the second row: open-rate collapse is the classic placement signal, and open rate is unreliable (Tracking caveats). Use reply rate against a stable baseline, plus the out-of-office tell.
Instrument before you diagnose
None of this works on a single campaign-level average. You need reply rate broken down by segment, by tier and by sequence step, plus bounce and unsubscribe rates alongside. Campaigns that were not instrumented can only be re-run, not diagnosed — which is why the recording step belongs in Campaign planning workflow.
Change one layer at a time
The temptation when results disappoint is to rewrite everything at once. Then the next campaign either works or does not, and you have learned nothing about why. Change one layer, hold the rest, and give it enough volume to produce a signal — noting that at realistic reply rates, "enough" is larger than most campaigns (A/B testing).
When the honest answer is "the thesis was wrong"
Sometimes the campaign is executed correctly and the underlying belief — that this group has this problem and would value this offer — is simply false. That is a real and useful outcome, and it is only visible if the thesis was written down at the start. Recognizing it early is worth more than another round of copy iteration.
References
- Smartlead — Email deliverability guide (vendor guide, Jun 2026)
- Belkins — Cold email response rates (7.5M emails sent in 2025, updated Jun 2026)
- Up: Playbook contents › Measurement and iteration
- Previous: Benchmarks
- Next: A/B testing