Same product. Same month. Same connection request.
Campaign A: 45 sent, 16 accepted. 36%. Campaign B: 411 sent, 242 accepted. 59%.
The copy was identical. The difference was who was on the list.
Campaign A was built from a conference attendee list. Regional, mixed titles, half of them vendors. The profile sending it had never touched that market.
Campaign B was built from a tight ICP: security leaders, one segment, one geography. The profile sending it had been posting into that market for months.
Twenty-three points of acceptance rate, and not one word of the message changed.
Most teams treat acceptance rate as a copy metric. Rewrite the note, test the opener, add a first name. That work is worth two or three points.
List quality and sender warmth are worth twenty.
So before we test a message, we test the list: does the title match, has the sender shown up in this market before, would this person recognise the company name.
If two of those are no, no opener is going to fix it.
How was the test set up?
| What it was | |
|---|---|
| Control | One message. Same opener, same note, same follow-up cadence on both profiles. |
| Variable 1: the list | A = conference attendee export, mixed titles, half vendors. B = one segment, one geo, built to the ICP. |
| Variable 2: the sender | A = profile new to that region, no prior activity there. B = profile posting into that market for months. |
| Result | 36% vs 59%. 23 points apart, zero words changed. |
How did the two campaigns compare?
| A: Conference list | B: ICP list | |
|---|---|---|
| How it was built | Event attendee export the week before the event | Sourced to one profile, one segment, one geography |
| Sent | 45 | 411 |
| Accepted | 16 | 242 |
| Acceptance rate | 35.6% | 58.9% |
| Replies | 2 in 5 days | 80 to the opener |
| Titles | CISO to "freelance for gov"; roughly half vendors and consultants | Security leaders only, one buying title band |
| Sender history in that market | Never posted into that region | Active in that market for months |
| Company recognition | No mutual connections to speak of | Company name already seen via events and partners |
How do you test a list before you test the message?
Run this before any copy test. Two "no" answers and the list goes back to sourcing. The opener is not the problem.
- Does the title match the ICP?
- Has the sender been active in this market in the last 60 days?
- Would this person recognise the company name?
When the list passes, then and only then A/B test the opener. Copy tests move 2–3 points, which is still worth doing.
What is acceptance rate actually measuring?
Working estimates from our accounts, not a benchmark.
| Lever | What it means | Rough weight | Where teams spend time |
|---|---|---|---|
| List quality | Title match, one segment, one geo, no vendors in the list | ~20 points | Rarely. "The list is fine." |
| Sender warmth | Has this profile shown up in that market in the last 60 days | ~12 points | Almost never. Treated as a content problem. |
| Profile maturity | Age, connections, premium, activity | ~6 points | Sometimes, when onboarding a new rep |
| Message copy | Opener, note, first-name token | ~3 points | Most of the week |
What did we change after the test?
Campaign A
- Paused the connection-request flow to the attendee list.
- Switched that audience to paid awareness (minimum 300).
- Kept the 16 accepted in a manual follow-up.
Campaign B
- Kept the list, kept the copy.
- Fixed the step after acceptance, which is where B leaked.
- Reported the 59% with the list criteria next to it.
Every campaign
- The list check runs before copy review.
- Acceptance rate is reported with the list source.
- "Rewrite the opener" is the last fix, not the first.