We ran 2 cold campaigns this year. Same sender, same platform, same weeks. One got roughly 24 times the response of the other, and the difference was not the data, the deliverability or the sending domain. It was the offer. This is what separated them, as 5 questions you can ask of your own.
Both were ours. Both went to businesses who had never heard of us. Both were, on any reasonable reading, useful to the person receiving them.
| Campaign A | Campaign B | |
|---|---|---|
| What it offered | A cleaning service for vehicle forecourts | Free social content, built from the dealer's own listing photographs |
| Emails sent | 960 | 236 |
| Opened | 29% | 34% |
| Said yes | 1 | 6 |
| Rate | 0.1% | 2.5% |
29% and 34% opened. The mail was delivered, the domain was trusted and the subject lines worked. Nothing about the sending was broken in either campaign. Whatever went wrong in A went wrong after the email was opened and read.
That matters, because the instinct when a campaign underperforms is to blame deliverability and start fiddling with records and warm-up. Here the deliverability was fine. The difference was entirely in what the message asked of the reader.
Ask these of your own offer before you send anything. They are not a style guide. Each one is a structural property of the offer itself, and you cannot fix a bad answer by writing better sentences.
| The question | A | B | |
|---|---|---|---|
| 1 | Is it about them, specifically? Could this exact email have gone to anyone in the trade? | A service, described | Their car, named, their own photographs |
| 2 | What does saying yes cost them? Money, time, risk, or a decision they have to justify to someone? | Money, and a budget conversation | Nothing at all |
| 3 | Can they see it before deciding? Or do they have to imagine it? | Nothing to look at | A page showing exactly what they would get |
| 4 | How hard is it to respond? Count the actions between reading and yes. | Write a reply, then have a call | One click |
| 5 | Is it a gift or a request? Does the email give them something, or ask them for something? | Asks for their money | Offers something already built |
It is a real service, priced fairly, that its customers value. It failed 4 of these 5 as written, and "as written" is the important part. The same business with a different offer scores completely differently, which is section 4.
Question 4 turned out to matter more than we expected, and we only noticed because 2 counters disagreed.
Our sending platform recorded 2 replies on campaign B. The campaign produced 6 agreements. The platform was not wrong: it counts replies to the email, and most of the people who said yes never sent one.
They clicked the page instead.
Writing a reply is work. It means composing a sentence, deciding on a tone, and committing to a conversation with a stranger. A link that says yes does none of those things.
Most of our response came through the route that required no writing. If we had only offered "reply to this email", we would have counted 2 and concluded the campaign failed.
The limit of what we can say: we cannot recover the exact split between page clicks and other routes from the data. What is certain is that the tracked outcome was 6 and the counted email replies were 2, and the gap is a response mechanism that asked less of the reader.
The test is only useful if a bad score can be fixed. Here is campaign A, rewritten against the 5 questions, without changing the business at all.
| As sent | Rewritten | |
|---|---|---|
| 1 | A description of a forecourt cleaning service | A photograph of their forecourt, taken from the street |
| 2 | A quote, and a budget decision | One bay cleaned free, nothing to pay |
| 3 | Nothing to look at | A before-and-after from a forecourt 10 miles away |
| 4 | Reply, then arrange a call | One click to pick a morning |
| 5 | Asks for money | Gives them a clean bay |
Same service. Same price list. Same list of dealers. The offer now scores the way campaign B scored, and the cost of finding out is one free bay per interested dealer rather than 960 more emails.
Give away the smallest real piece of the thing you sell, show it to them finished, and make accepting it a single click. If you cannot find a piece small enough to give away, that is worth knowing before you spend a list on finding out.
These are 2 campaigns, not a study. 960 emails and 236 emails, in one market, from one sender. The 5 questions are a way of thinking about an offer, not a formula, and the only opinion here we hold strongly is that the offer, not the list, is what decides the outcome. A perfect list attached to campaign A would still have returned 1.
Everything we know about getting from nothing to a campaign somebody answers. Free, no account, nothing to sign up to.
Winnowed