Message testing: turn confusion into better copy

Message testing reveals what buyers understand, find relevant and believe in your copy. Test it where they will encounter it. Use the responses to choose the words to keep, the confusion to remove and the proof to add.

A customer can love your headline because they’ve imagined a product you don’t sell. That makes an enthusiastic response worth looking at as closely as an objection.

The useful question is what the customer thinks will happen next. Their explanation can expose a promise you never intended to make, or show you exactly what needs more proof. This guide uses one fictional meal-planning service to show how that feedback changes the next draft.

Look for the product people have pictured

By the time a campaign is ready to test, the team knows the product well. A few words can remind them of features, benefits and demonstrations they have seen many times. A new customer has only the words and images in front of them, and may fill in the gaps very differently.

Ask them to describe the offer and what they expect to receive. An answer such as “I’d use this because it would save me time” becomes more revealing when you know what they think will save that time.

That explanation gives the writer something to work with. If a customer likes a headline because they expect delivery, and the product helps them plan, the next draft has a specific misunderstanding to resolve.

Start with the offer

Consider Weektable, a fictional meal-planning service. You choose five dinners, save the plan and get one grocery list. You buy the ingredients and cook.

The team wants to reach households trying to organize weekday dinners around work and family. It is choosing a headline for an ad: should the copy lead with the relief of having dinner decided, or show how the service works?

Both versions will carry the same supporting line: “A dinner plan and grocery list for your week.” This gives people enough information to react to the offer while leaving the headline to do its own work.

Compare the benefit with the explanation

Show each person one version first, beside the Weektable name and the shared supporting line. Keep the layout and other information the same. Collect that first reaction before showing the alternative; otherwise, the second version may supply an explanation the first one lacked.

For a landing-page test, include the actual section people would see. A screenshot of the meal planner may explain the service better than either headline. Save the complete version with the response so a recommendation refers to the copy and images that produced it.

Let people explain it before you explain it

Start with “What do you think this service does?” and “What would happen after you signed up?” Let people answer in their own words before showing a list of benefits. The options may explain the service for them, making it harder to tell what the copy communicated on its own. AAPOR’s questionnaire guidance covers the effects of question order and response options.

Before reading the answers, agree what understanding the offer means. For Weektable, someone needs to recognize dinner selection and a grocery list without expecting meals or ingredients to arrive at their door. Keep answers that mix those ideas separate from answers that describe the service accurately.

Then ask how it would fit into their week, what seems helpful, what they doubt and what they would want to see. Someone who understands the offer may still have a good reason to pass on it.

If an answer is vague, follow up. “It sorts dinner for me” could mean several things. Ask what they expect to receive, and keep their original words beside your interpretation.

The words can be right while the expectation is wrong

Here are two invented first reactions to the Weektable ads. They illustrate how to read an answer; no study was run.

Teaching example: the original message and a revised candidate

First encounter

Dinner, without the daily debate. A dinner plan and grocery list for your week.

Invented interpretation: “A grocery list comes with the meals they deliver.”

Revised candidate

Pick five dinners. Get one organized grocery list. Plan the week once, so dinner is easier to decide each evening.

Fictional Weektable copy and response. The revision names the service and keeps the emotional benefit; it has no measured conversion result.

The first answer contains the phrase “grocery list,” but it also contains a promise Weektable cannot fulfil. If you count that as understanding, the readout looks reassuring while concealing the problem.

The second customer already understands the service. Explaining meal planning again would miss their question. They want to know whether the list handles something that would otherwise take them time.

These individual examples do not tell us which headline wins. In a real readout, show how often each interpretation occurred, among whom and after seeing which version. Keep the explanations with the results so the team can see why a change is being recommended.

Turn the misunderstanding into better copy

The next version can keep the relief of A while making the service concrete:

Pick five dinners. Get one organized grocery list, and one less decision each evening.

Or keep the shorter headline, “Pick five dinners. Get one organized grocery list,” and add: “Plan the week once, so dinner is easier to decide each evening.” A view of the planner and its grocery list would show what the words describe.

These are candidates to test. The reason for trying them is that they address the misunderstanding; we have no result showing they improve conversion.

If customers understand the service but prefer their current routine, ask what would make it worth changing. Perhaps the proposed benefit is too small, or the idea fits a different household. The answer may lead to a different promise or audience rather than more explanation.

The same problem can arise in business software. “Close faster, with less chasing” might make an accountant think about reconciliations and a finance lead picture getting reports finished. Find out what each person has pictured, make the actual workflow clear, and show the proof relevant to their work.

Where Sapien can help

You can put candidate messages to a simulated audience in Sapien and ask follow-up questions about its reactions. Explain the offer, who the campaign is for and what they would see. Past research and customer data can add context to the audience.

Look at the answers behind a recommendation: what does the audience think the product does, what appeals to it and what remains unclear? That gives a brand team or agency strategist material for the next draft and an explanation they can share with a colleague or client.

These are modeled reactions. Keep them distinct from responses collected from people, and ask for relevant validation if you are using them to support a prediction.

Put the revised message back into the campaign

Try the revision where customers will encounter it. A search ad has less room to explain Weektable than a landing-page section. The test should leave you with a specific version to use in that setting.

Then look at the expectations people bring with them. If enquiries still ask where the meals are delivered, the campaign is carrying the same misunderstanding forward. More clicks would not resolve it.

For help comparing your messages, bring Sapien the current copy, audience and campaign context. Talk through the messages you want to test.

The Weektable examples use invented responses, not findings from a completed study.