No, Surveys Aren’t Enough to Understand Your Customer

Problems with traditional surveys

We can summarize the main problems in the following points:

Structural factors

• Language issues: Traditional surveys are based on a verbal report. As we all know, language per se has many limitations, from the use of biased words in questions, to a complete misunderstanding of what is written (and how people read it, therefore).

  Look at the following example: «Where do you enjoy drinking beer?» This is a skewed question. We assume that the respondent drinks beer and forces us to respond in a way that doesn’t necessarily reflect our opinion. You can find a list of the most common mistakes when writing up questions for surveys here.

• Response Rate: People not only tend to fill out surveys partially, but many also don’t answer them at all. The average response rate is 33% (1 in 3 respondents) so we’ll need three times as many respondents to reach our initial goal. That’s a significant waste of time and resources! Now, there are differences between survey recollection techniques, but it’s still a significantly low rate:

Infographic from SurveyAnyplace

• Response quality: various studies reveal that respondents take shortcuts to reduce the effort and attention when filling out a survey. Not only do they not read the questions carefully, but the answers given are unreliable too.

  For example, in multiple-choice questions, participants might decide to always fill in one end of the column (or always click on the answer A) regardless of the question’s statement. Many ill practices put into question the reliability of the information we can get from these surveys.

Psychological factors

• Honesty: people lie in the surveys they take. A recent study found that only 16% of people are completely honest in them. Considering this low percentage, we might think twice about how useful surveys are. Plus, it’s very difficult to detect lies in them.
• Social Desirability: people want to show a positive self-image and they seek to look good in front of others. People seek social approval so they’ll choose responses that they believe others expect of them. This is more evident in societies that focus more on the collective well-being than on the individual, as it happens in Asian or Latin American countries.

  For instance, people often agree with phrases like «I would never lie to another person» or «I always keep my promises, no matter what happens.» Personality questionnaires have introduced specific scales to detect this phenomenon, but common surveys don’t seem to use them.

Surveys have many limitations but we still use them as our main source of customer information. Don’t you think it’s time for a change?

When they’re used without resorting to other sources of information, there is a high risk of providing unreliable data. And we all know what happens when we make decisions based on faulty information, right?

The future of data collection

While traditional surveys have contributed to decision-making for decades, they’re becoming increasingly outdated. Many shortcomings decrease the survey’s usefulness which leads us to ponder a much-needed change in data collection.

There’s a bright future in other techniques such as eye-tracking, facial recognition, and other methods applying AI and neuroscience to the equation. I wrote an article about this if you want to check it out.

The main difficulty of surveys is that they’re subject to language interpretation such as biases, reading comprehension, vocabulary, and expressions varying from one generation to another.

An analysis of people’s facial features can avoid this problem.

Facial Recognition Software at my firm Sounditi

Our main emotions are reflected on our faces. Each emotion involves an activation of a distinctive set of muscles that we can read with specialized software. This software is constantly learning, contrasting it with thousands of facial traits in its database. When it comes to recognizing the emotional state of the respondent, AI has an unmatched level of precision.

Emotions are short term reactions. We express them without consciously controlling every muscle on our face.

Look at the image on the left. For a smile to be genuine it must activate the zygomaticus major muscle (located near the corner of the lip) and the orbicularis oculi muscle (near the eye). An artificial intelligence engine can distinguish this and many other reactions with great precision.

There are other advantages if you use a facial recognition system:

• It allows access to the respondent’s genuine emotions. These are accessible in real-time, unlike in traditional surveys. Surveys cannot collect information at such a depth.
• The level of intrusion is minimal. You just need a video or image to run an analysis.
• The probability of falsifying data is significantly reduced since facial reactions tend to be involuntary.
• Since it can recognize a person’s prominent features, it decreases the probability of having identity theft which traditional surveys cannot control so effectively.