If you could pick one thing that you wanted from potential customers, would it be that they associate your brand with the top attribute from your brand strategy document, or that they buy your product?

You can collect different kinds of data about your customers and brand. These data can be arranged along a ladder of evidence based on how useful they are to you. At the bottom rung are attitudes (what people think), and at the top we find actions (what people do). Together with the intervening steps, the ladder looks like this:

• What people do

• What people say they have done

• What people say they will do

• What people think

You want to climb as high as you can on the ladder of evidence, for two reasons.

The first is business relevance. Actions make us money; attitudes help a lot less. Because the business runs on customer behaviors – when they buy our products, recommend them to their friends, renew a contract, or upgrade tiers. Nobody has ever deposited an attitude into your bank account, so it makes sense to try to get as close as possible to actions.

The second reason is data reliability. Every step down from the top rung adds a layer of translation between customer behavior (which makes us money) and your data. What people did is what they did. What people say they did passes through memory. What people say they will do passes through optimism and social desirability. What people think passes through their limited access to why they do anything at all. Each translation is a place for error to enter. The further down the ladder you go, the less reliable will the business implications be.

So why do people work with attitudes if they are so bad? A big reason is that it is easy. Send out an online survey, ask a few questions, and you get something to put on your slides. The whole process is quick and relatively cheap. (There is also a degree of inertia in the industry: business professors have built careers on intricate theories about attitudes, so they train their students to do the same, which they will continue to use once they graduate.)

Another important reason is that attitudes are not useless – they are just upstream from the actions we are really interested in. A positive attitude towards our brand can tip the balance in our favor in the moment of choice. Fame, warm feelings, attractive attributes, social signaling are all factors to take into consideration. But attitudes are only ever a means to try and influence actions. If your attitude work never shows up in behavior, you have spent a lot of money trying to tilt a choice that never occurs.

Sometimes it is impractical to study actual behavior. It can be hard, time-consuming, sensitive, expensive, or technically difficult to collect the data. When actions are out of reach, take one step down the ladder, not four. Here surveys are not a bad tool – even if we cannot get access to actual behavioral data, we can get closer to it by asking the right kind of questions.

Do not just ask what people think; ask what they did. And, crucially, ask follow-up questions that map the context where they made their decision. So instead of “Do you like oat-milk brand X?” you ask “Have you bought oat milk in the last month?” Followed by questions like: What brand? Which store? What kind of trip – the weekly or a quick top-up? What was it used for – drinking, to put in your coffee, for cooking? Who was it for – yourself, your partner, your kids, the whole family? Who was with you when you made your choice? What happened after – did you use it, and did it taste good?

The context matters because the situation, not the attitude, often decides what action is taken. My colleague Daniel really likes Porsche, but he lives in the city and a bike is much more practical, so he will not buy a Porsche regardless of his feelings about the brand (not to mention, it is outside his budget). Daniel would ace Porche’s brand tracker and never appear in their sales data. If you only ask what he thinks, you learn nothing about what he will do.

Your brand strategy describes what you want people to think. Your bank account shows you what they did. Spend your research budget as close to the top of the ladder as you can get.