Beyond the Say-Do Gap

For the past 30 years, the insights industry has been operating under the knowledge that what people say and what people do rarely align.

We have built entire methodologies around it, from behavioural economics to implicit association testing to ethnography, each one another attempt to bypass what people say and get closer to what they actually do.

And yet the question that sits at the heart of almost every research project I have ever worked on remains the same...

Did we get the right answer, or just a good-looking one?

Because somewhere between what respondents said and what they were really thinking, something slipped. The insight feels a little too tidy, a little too much like what everyone expected to find, and that feeling is not paranoia. It is social desirability bias, and it is one of the most consequential and least discussed challenges in research, particularly when the topic is sensitive.

The industry has been solving the wrong problem

The say-do gap assumes the problem is with the individual respondent; that they intend to do one thing and end up doing another. But what we have been finding, consistently, across some of the most sensitive research we have run at Bolt, is something more structural.

The gap is not primarily between what people say and what they do. It is between what people privately believe and what they publicly permit, between individual conviction and collective permission. We call it the Honesty Gap. And the reason conventional research misses it is that it is only ever measuring one version of the person.

When you ask someone in a survey whether they support a particular value or behavior, they give you a genuine answer, but what they are giving you is the considered, socially calibrated version of their view.

Allen Edwards formally identified this in 1953: respondents systematically skew toward the socially acceptable answer not because they are dishonest, but because the desire to present oneself favourably is a deeply embedded social instinct.

As long as someone feels they are performing for an audience, they will perform. What we needed was a way to change the conditions, not just the questions.

What you find when you look harder

Earlier this year we ran a multi-methodology study on LGBTQ+ acceptance in professional football. If acceptance is as high as stated data suggests, why has not a single active elite male player come out in over a hundred years of professional football?

We surveyed 3,000 respondents across the UK, US and Brazil, tracked 89,500 social media mentions during the World Cup period, and ran 150 AI-moderated qualitative interviews with semi- and pro-footballers, LGBTQ+ individuals and fans using Bolt Intelligence.

By removing the human audience that respondents typically perform for, we created conditions where honesty came more naturally. What came back were not one gap but five distinct paradoxes. 92% of fans agreed a footballer should feel safe to come out, but only 15% would actively encourage him. 89% were personally comfortable with a gay player at their club, but only 28% believed other fans felt the same.

66% of people who said they would publicly support a gay player privately advised against it, driven not by hostility but by concern for his mental health and personal wellbeing.

All five paradoxes share a common root... the masculine code. 77% of fans expect male footballers to be tough, emotionally controlled and strong. The men's game has become the gold standard of a particular kind of masculinity that the women's game was never asked to perform. That explains why the paradoxes exist in the first place.

The qualitative sessions surfaced reasoning the quantitative instrument could not access. Not "I support him" but "people think it's just about courage, but it's about whether you want your whole life turned upside down." These are not details that add colour to a finding. They are the mechanism.

What this means for brand research

The football finding is not football-specific. The same pattern shows up wherever stated values diverge from lived behaviour. Stated purchase intent for sustainable products consistently outperforms actual behaviour. Purpose alignment scores outperform brand loyalty when a brand is tested on a genuinely difficult issue.

In every case, the reflective self is more progressive than the social self, and the gap between them is where the strategic question lives.

The say-do gap is a symptom. The Honesty Gap is the diagnosis. Getting to it requires methods built to reach below the stated answer, and the experience to know what you are looking for when you get there.

That is what running AI-moderated research across 15+ million consumers, in 120+ countries, on some of the most sensitive topics in research has given us. Not just a technology. A framework. And a consistent finding that the gap between private belief and public permission is not a failure of individual honesty. It is the predictable outcome of an industry that has spent decades measuring the wrong self.

You can find out more about Bolt Intelligence here.