Women don’t know what they want. Men don’t know what they want. We all pay a good buck to gurus and therapists to tell us who we are.
If we were smart, self-observant creatures, we would never believe commercials persuading us that buying a new sweater will make us finally complete.
Understanding our real needs and wants is a task for a lifetime. So, we might as well agree that asking people about their opinions in surveys might be ineffective.
It has been proven that people find themselves more moral than they are. So when you ask them about their inclination to buy stuff during Black Friday? They’ll most likely reply: “Of course not! Me and Mary buy only products we need, no matter the bloody sale.” And that’s a lie.
How can we, as marketers, get to the truth? Simply. Instead of asking people and sending surveys, do experiments. Put people in the situation you’re trying to examine.
Suddenly, you might discover that your customer is very much prone to be seduced by Black Friday sales.
Actually, they bought an amount of toilet paper for the next decade and a new TV that day. But if you ask them why they bought the products, they will still be stating that it’s just a necessary expense.
Of course, I have an experiment to prove this. In 1999, Adrian North, then a psychologist at Leicester University. For 14 days, he focused on wine-buying customers at a local supermarket.
During the first week, he played French harmonic music in the wine aisle, and French wine suddenly accounted for 83% of wine sales. The second week, he played German music in the aisle, and German wine made 65% of the sales. The results speak for themselves, right?
However, then North started asking people why they bought French/German wine today. He realised that only 2% of them admitted that the music had something to do with their decision.
According to the experiment, people are heavily influenced by the background music while making their wine-shopping decisions.
But according to the surveys, people are buying wine that they like/because the label was “cute”, just because they were in the “mood”, or they’re simply experts on wine and know what’s best!
Well… just stop asking questions, if you want to know the truth, I’d suggest.
This article is inspired by The Illusion of Choice. A book written by Richard Shotton about cognitive biases in marketing and everyday life. Read another article from this series.
Create an Experiment Step by Step
How to create a successful experiment? I’ll share a simple seven-step routine that might help you succeed.
Write down your problem in one sentence
If you can’t put it simply, you don’t understand it enough. Trust me on this. Pick only one problem, one perspective and leave the rest for another experiment. It’s the participant's job to add noise to your research. You have to stay as simple as possible. Also, don’t be too vague.
An experiment is a great means to test your assumptions, stereotypes, and “common sense“ sayings.
How do you put the whole problem into one sentence? This form could work for you:
[specific target group] tends to do [specific action] when [tested object] is present.
Let’s apply it to the North’s research I mentioned earlier:
[The wine-buying customers in this supermarket] tend to [buy French wine] when [French music] is playing.
Visit the library
I know it’s boring. But first, we always have to consult the existing research.
Read important articles published on the topic you’re interested in. Maybe you’ll find out that the statement you so carefully wrote down in the first step was already disproven.
Or, you can find similar papers supporting your findings!
As for our supermarket experiment, I would recommend Robert Cialdini’s book Pre-Suasion. It’s a good start.
However, don’t create your experiment just to support the positive findings of others. You might unconsciously influence the whole experiment just to get the answer you want.
Is your experiment necessary?
Compare your one-sentence problem with found research. Do you have a perfect match? No? Then, it makes sense to run your experiment.
It could be the tiniest detail. Previous research was on beer, not wine. It was done outside at a farmer’s market, not in a supermarket. Everything counts.
Do you need a monadic test?
Now, it’s time to choose how your experiment will look. Do you still have to include a survey? Do a monadic test instead to get better results. I’ll show you how.
Find a way to ask the same question in multiple ways. For example, instead of asking:
“Which insurance seems like a better deal? Paying £4,57 a day or £32 a week?“
(both values make up the same price at the end, of course)
Divide people into two groups, and each group ask one of the following questions:
“How would you rate insurance that costs £4,57 a day?“
“How would you rate insurance that costs £32 a week?“
This way people will answer a simple question and would not try to be smart about it. Which will improve the results immensely.
Now, evaluate. Let's pretend that according to the results, people rated the £4,57 a day insurance as a better deal. Probably because the number was smaller.
So the monadic test proves that if you want to sell insurance, it’s more likely people will buy it if you show the price per day rather than the price per week or month.
Do you need a field experiment?
If you choose the field experiment, you get to ditch all the surveys and monadic tests entirely. Let’s get to it!
First, create two scenarios in a naturalistic setting or, even better, a real world. Make sure that everything is the same in both scenarios, except one thing.
For example, do you want to know whether the actual weather forecast influences people’s opinion about the weather anchor? Show two groups of people an identical weather forecast with the same anchor, except that one group will be promised sunny weather and the second heavy rain.
Second, compare the results. If the first group liked the anchor better than the second, you know the weather forecast influenced their judgment.
(People hating anchors for bad weather is a real thing, by the way. Let's give them some love.)
Prepare your experiment
Keep it simple. One metric and one problem at a time.
Choose a representative sample of participants. 5 might be too low, and 1000 might be unnecessary. As for most small experiments, the 100 is the magic number.
The participants must know they are part of an experiment but don’t tell them what your research is actually about. Otherwise, we’ll be back at the survey problem. They will unknowingly compromise their behaviour to look better in front of you.
Let's show how to do that in our wine experiment:
In the end, you have to share that you’re running an experiment before asking why they bought a French wine, right?
So this is how to do it right:
“Hello, I’m researching wines. Can I ask you one question?”
And this is how to do it way wrong:
“Hi, I’m researching how French music played in the background in this
supermarket influences French wine sales. Do you have a moment?”
Don’t forget to diversify your participants properly. You want all kinds of people there. As for our supermarket experiment, the music in the background has to play during the whole opening hours. Not just in the evenings or in the morning.
7. Run a bigger, real-world experiment
Your first experiment should be smaller, so you won’t spend much money if it fails. It can also be done in artificial conditions. In your office, at the university.
The first experiment is there just to support or decline your initial assumption. Then you have to go all out.
In the second experiment, you can go bigger by prolonging its duration (play French music in the supermarket for a month instead of a week), getting more participants (500 instead of 50) or making the conditions even more real-like (if you tried the first experiment at the uni with students, go outside and play with real people this time.)
Running experiments are extremely important. Even though we might be book-smart and know a lot about psychology and behavioural sciences, we are never able to predict how people will behave in that one exact scenario.
Good luck!
Source: Richard Shotton | The Illusion of Choice
Read more: Make your numbers look bigger
Comments