To test a message before it goes public, show the exact draft to readers from the audience it is meant for and ask three things: what they think is changing, what they would question, and what they would do next. Then fix the sentence their answers point to and check the new version.
Start with the line most likely to be misread. One missing fact usually does more damage than the tone of the whole announcement: a price rise needs the amount and the date, and a change to opening hours needs the hours.
A good message test ends with an edit, not a verdict.
Why do announcements go wrong in public?
Most go wrong in one of three ways. Check each one separately.
They are read differently from how they were meant. "More flexibility from January" could mean easier cancellation, more class times or a new payment schedule. Ask readers to describe the change in their own words. Their answer tells you which meaning your sentence actually carries.
The wrong thing is noticed first. A headline about a new benefit can push a price rise into the background. Readers may understand both and still feel you hid the cost. Ask which detail they noticed first and which one changed their reaction.
The tone lands badly. "Thanks for sticking with us" can sound warm to one reader and smug to another. Ask which words caused the reaction, so the editor gets something more precise than "make it friendlier".
How to test a message in an afternoon
- Write down the decision. "We need to know whether members understand what changes on 1 January."
- Show the draft as it will appear. For an email, include the subject line, the body and the next step. For a social post, show the words and the image together.
- Choose who has to understand it. Existing members and people thinking of joining ask different questions. Record which audience you picked and why.
- Ask what it means before asking whether they like it. Open with "What do you understand is changing?" Follow with "Which part would you want explained?" and "What would you do next?" Keep the first answers before you explain anything.
- Write the edit next to its evidence. Put the exact answer beside the sentence it is about, then decide which fact, order or wording to change and what a second check should show.
Here is a brief you can copy:
Decision: Can members tell what changes, when it changes and where to act?
Audience: Current UK gym members who read membership emails.
Material: The full proposed email, including its subject line and cancellation instructions.
First question: What do you understand is changing for you?
Follow-up: Which part would you want explained before deciding what to do?
Edit rule: If an answer points to a missing practical detail, confirm that detail with the team responsible and add it to the email.
The two examples below are real runs. Both are announcements of a higher price, because that is where a missing reason or a vague benefit costs the most goodwill. To compare several campaign lines instead, use the broader message-testing workflow.
Example: Malta, a bakery price rise
The lesson: the reason did the work. On 2 November 2026, AYA ran a Concept Test of a hypothetical Facebook post, written by us, with six Malta Human Digital Twins built from recorded interviews. The post:
From Monday our pastizzi go from 60c to 75c. Flour, butter and electricity have all gone up, and we'd rather charge more than make them smaller. Thanks for sticking with us.
The question: "A Malta bakery posts this on Facebook. What is your first reaction, and would it change whether you buy there?"
Luigi, a 38-year-old twin from St Julian's:
I would much rather pay seventy-five cents for a proper pastizz than have them cut corners or make it half the size.
Nicholas, a 27-year-old twin from Żabbar:
Look, nobody wants to pay more when site wages haven't moved an inch, but at least they're being straight with people.
The report labelled 5 of 6 reactions positive and 1 neutral. Luigi picked up the trade-off the bakery offered, price against size, and took the bakery's side. Nicholas still felt the cost but credited the honesty.
What I'd change: keep the amount, the causes and the line about size; they are what the twins responded to. Replace "Monday" with the actual date, so a screenshot shared next month still makes sense. That date edit is my recommendation, not something the panel asked for.
Example: UK, a gym membership email
The lesson: a vague benefit cannot justify a price. On the same day, six UK Human Digital Twins saw this hypothetical email, also written by AYA, in a Concept Test:
From 1 January you can cancel your membership any time in the app, no notice period. We're also adding £2 a month to cover longer opening hours.
The question: "Your gym sends you this email. What is your first reaction, and what would you do next?"
John, a 57-year-old twin from Kendal, Cumbria:
My immediate thought is that being able to cancel cleanly in an app without a 30-day notice dance is remarkably fair.
Jordan, a 31-year-old twin from Redlands, Reading:
I'll check whether the new hours mean earlier morning access, but otherwise I'll keep the membership as it is.
Becky, a 27-year-old twin from Sutton Coldfield, Birmingham:
But that extra £2 a month made me wince.
The report labelled 3 of 6 reactions positive and 3 neutral. The good news did its job: John valued the easier cancellation. "Longer opening hours" was there to justify the £2, yet Jordan's next step was to go and find out what the hours actually were, and Becky winced at the fee.
What I'd change: confirm the new timetable, then put the actual hours right next to the £2 ("open from 6am, up from 7am" or whatever is true). Give the cancellation change its own sentence with the route in the app, so the good news and the price rise are not doing each other's work. These are proposed edits for a second check; the run tested the original email only.
What the two drafts teach together. This is my reading, not a comparison of markets: the messages and questions were different. Both asked readers to accept a higher price. The bakery gave a reason readers could see in the product. The gym gave a benefit readers had to go and look up. If your message asks people to pay more, make the reason something they can check from the message itself.
What to change after the test
Change the copy where an answer shows a gap between what the reader needs and what the message gives them. Send objections to the offer itself to whoever owns that decision.
| What the answer shows | What to do |
|---|---|
| The reader cannot tell when the change starts | Put the confirmed date beside the change |
| A benefit could mean several things | Name the practical detail, such as the actual hours |
| The cost is clear but unwelcome | Take the objection to the person who sets the price |
| A phrase sounds dismissive | Ask which words caused that reading, then test a plainer version |
Keep a short record for each edit: the original sentence, the exact answer, the proposed change, who confirms the fact and the follow-up question. If the offer itself is still moving, the concept-testing template gives a wider brief.
For the gym, a good follow-up would be: "What changes for you from 1 January, and what would you check next?" Show the revised email without explaining your edit, then compare the answers with the decision you wrote down in step 1.
Limits
Each example is a small, directional panel of six twins in one market, not a representative sample and not a new live interview. A twin's answer is generated from the interview it was built on; it is not a new statement by the person behind it. Questions about buying or next steps do not predict purchases, cancellations or demand.
We quote excerpts and the report's sentiment labels, not full transcripts, and we have not tested the rewrites suggested here. The two runs used different messages and questions, so they say nothing about Malta compared with the UK.
For a significant price or policy change, take the revised copy to affected customers and the team responsible before release. A message test can show that an explanation is missing; it cannot tell you whether the decision behind it is fair.
Try it on your message
Put your draft and one open question into a free Snapshot and get reactions from six Human Digital Twins in your market. Keep the original wording, and take in the one sentence you are least sure about. If the answers point to a missing fact, get the fact before you polish the sentence.
Disclosure: I built AYA, so weigh this accordingly. The methodology explains how the twins are built from recorded interviews and where they stop. If you are new to the idea, what an AI focus group is covers the basics.
FAQ
How many readers do I need to test a message?
To find misreadings, a handful is enough: if two of six readers take a different meaning, you have a sentence to fix. A Snapshot uses six twins. To measure how widely a reaction is shared, you need a properly sampled survey.
Should I test the subject line or the whole email?
Show the whole email when the question is whether readers understand the announcement. Test the subject line on its own when you need to know what readers expect before they open it.
Can I test a message that has already been approved?
Yes. Agree first which parts can still change, such as the order, the explanation or the next step. Send questions about the policy itself back to its owner.
What if the readers disagree?
Keep both reactions and find the concern behind each. A reader who likes the flexibility and another who resents the cost may both understand the email perfectly; the person deciding needs to see both reasons.
Do I need a finished design?
You need enough design to keep the reading order. If where the price or button sits changes what readers notice, include it in the test.