How to validate new positioning before committing campaign spend

By the time you see the first campaign concept, a lot has already been decided.

Someone has chosen the customer need. The brief has turned it into a promise. The agency has built a story around it. Now there are people, money and opinions attached to the answer.

That is an expensive moment to discover you picked the wrong problem.

Positioning deserves a test much earlier. While you can still change the brief. While a different answer would change what you build, what you say and where you put the budget.

Here's how to do it.

Start with the need you're betting on

Imagine a food-storage brand planning its next campaign. We'll use a fictional example.

The team has three ideas about what customers need:

• Keep food fresh so less of the weekly shop ends up in the bin.
• Bring some order to a fridge where nobody can find anything.
• Get lunch to work without it leaking through the bag.

Each could be a compelling reason to buy the same product. Each leads to a different campaign, a different demonstration and potentially a different product priority.

Before choosing the words, decide which of those needs deserves the investment.

Write down the decision the experiment will change: "Which customer need should lead our next campaign for this audience?"

Also write down the team's favorite. It's useful to know what you believed before the market had a say.

Give each idea a fair shot

Turn each need into a clear proposition:

• "Keep tomorrow's lunch fresh."
• "Make room in your fridge."
• "Lunch in your bag. Not all over it."

Use the same product, price, offer, image treatment and call to action. Keep the destination consistent. You want to understand the response to the proposition, so give each one an equally good execution.

If the freshness version has a discount and the organization version has a different product, you've made the answer harder to use. You'll be debating what caused the result instead of improving the brief.

Look at the ads side by side before launching. Is each promise easy to understand? Does the product deliver it? Have you given one idea a much better visual?

Decide whose answer you need

Be specific about the people this campaign is for.

Commuters carrying lunch and households organizing a weekly shop might respond very differently. That difference could be the most useful thing you learn.

Choose the market and segment, then use the same audience definition across the variants. Check the targeting and delivery settings on the platform. When the results arrive, look at who actually saw each option.

If two segments matter to the decision, keep their results separate. That gives you something useful to act on: a proposition for each audience, or a reason to concentrate on one.

Put the propositions into market

In Heatseeker, start with your question and the decision you need to make. Select the product and audience, review the recommended experiment, approve the ads and set the budget. Live experiments run through connected Meta, LinkedIn or TikTok accounts. Read how an experiment gets designed and run.

Real people then see the competing propositions in market. Their responses give you evidence about which need is earning attention and action.

Behind every live experiment, Heatseeker also runs a synthetic experiment. We compare the paired results to measure how closely the model's responses correlate with what happens in the market.

That comparison is built into the process. Each live test gives you a decision to work with and a direct check on the synthetic response.

Before launch, check the whole customer experience, including the destination and tracking. For an upcoming product, make its availability clear.

Choose the signal that answers your question

Decide what you'll measure before the results come in.

For a proposition test, you might compare click response. For a lead-generation test, you'll want to see who takes that next step. If the decision depends on purchases, set up the purchase measurement.

Then read the strength of the result alongside the ranking. Heatseeker's results show both relative performance and how conclusive the signal is. See how to read your results.

A close result is useful too. It may tell you the propositions need sharpening, that two options deserve further development, or that you need more evidence before committing the budget.

Agree how you'll handle that before launch. Otherwise it's very easy to keep asking the question until the answer agrees with the room.

Use the answer to improve the brief

Suppose organization leads in our fictional example.

Now the creative team has a much sharper job. Show how this product makes a crowded fridge easier to live with.

You can test the next layer: is it seeing what's inside, stacking containers neatly, or finding lunch quickly? You can explore whether the same need matters to another segment. And you can follow the response through to purchase.

You've moved from a broad argument about what customers want to a specific question worth answering next.

Keep the learning available

The value of an experiment continues after the campaign brief.

Heatseeker's customer models let teams explore the next question using the evidence they've built. The paired synthetic and live experiments keep giving them a way to compare the model's response with the market.

Our published study of general AI models examines a separate question: how their choices compared with results from 11 selected live experiments. Read the study and method. The ongoing paired testing is part of the product itself.

Where this method comes from

There is an established research base for putting choices to the test. In a field experiment with real loan offers, Bertrand and colleagues varied advertising content and measured its effect on demand. It is a useful model for turning a debate about a proposition into a question the market can answer.

The consequences of a choice matter too. Ding, Grewal and Liechty found that incentive-aligned conjoint improved out-of-sample choice prediction in their studies. Schmidt and Bijmolt's meta-analysis found systematic differences between hypothetical and real willingness to pay for consumer goods.

Good design makes the response useful. Gordon and colleagues compared observational estimates with randomized Facebook experiments and found that the observational approaches often missed the experimental effects. For a proposition test, that means checking allocation, keeping the comparison fair and measuring the action the decision depends on.

Before committing the bigger budget, ask your team: which customer need are we building this campaign around, and what have we learned about it?

Bring that question to Heatseeker. Test it while the answer can still change the work.