There's a well-documented phenomenon in consumer and user research that most practitioners know intuitively but rarely name directly: people don't report their experience accurately.
Not because they're dishonest. Because self-reporting is hard. In the moment of an interview, participants are performing a version of themselves. They round off hesitation. They describe their behaviour more charitably than it actually was. They say "yes, I'd probably use this" when what they felt was closer to "maybe, under the right circumstances, if the price were different." The social pressure of being in a conversation, even with an AI, shapes what gets said.
This is what we call the Say-Do Gap. The distance between what someone tells you and what they actually feel or do.
It shows up everywhere. In concept tests where participants say they love a product they'd never buy. In usability sessions where someone says "this makes sense" while visibly struggling. In brand perception studies where stated attitudes don't match purchasing behaviour.
Respect for not hand-waving that, most launch threads would have. One thing I'd add: per-frame confidence is the model scoring its own certainty, so it won't catch systematic bias. A model can be high-confidence and wrong the same way across a whole population and never flag it. The only check I trust is human-coded ground truth sampled per region, which is painful to collect. Which regions have you actually validated against local human coders versus carried over from the base model?
Does it self-report confidence?
Mira
@paige_lauren1 Yes we can report confidence as well which is majorly driven by tonality followed by facial and textual signals.
Mira
@paige_lauren1 To add to Sumit's answer; in practice this means when a participant sounds uncertain but their transcript reads neutrally, the system surfaces that discrepancy and flags it rather than averaging it away.
The confidence score tells you how much weight to put on any individual emotional data point before drawing a conclusion.
Mira
Head of Sales at Decode here.
The conversation I have most with research and insights teams: "Our studies take too long, and leadership does not trust the findings."
Both problems have the same root cause. The tools being used only capture what people say, and analysis is manual. A 20-participant qual study typically takes 6-8 weeks from setup to report, arriving too late to influence the decision it was commissioned to support.
Mira compresses that timeline. Study setup with templates takes minutes. Recruitment from a 100M+ global panel is built in. Transcripts, themes, and reports are generated automatically. Emotional signal adds defensibility to findings.
If you run an insights function and want to understand what this looks like for your team's specific workflow, feel free to ask below.
The facial coding combined with voice emotion analysis during a test interview genuinely caught moments I would have missed in a regular transcript. Feels closer to sitting with a respondent than a typical AI research tool.
Mira
@srammoa "Closer to sitting with a respondent" is one of the best ways I have heard it described. The transcript tells you what they said. Mira tries to preserve what was happening underneath that. Thank you for testing it.
the facial coding detail is wild, never seen a research tool that actually picks up on what people are feeling while answering. ran a quick test on our team and the voice emotion layer caught hesitation I would have totally missed in a normal transcript.
Mira
@nihalzsaki243g The hesitation catch is where the voice layer earns its place; that pause before the polished answer is usually the most honest moment in the whole session. Glad it surfaced. If you want to run a structured study, the first one is free this month. Please feel free to book a demo at https://www.entropik.io/platform/ai-moderator and mention PH20 on the call :)
Tried Mira for a quick concept test and the facial coding actually flagged a hesitation I never would have caught in a transcript alone. Slightly surreal watching it react to my own face, but genuinely useful.
Mira
@elanurybrp This is exactly the feedback that matters most, thank you, Elanur.
The hesitation flag mid-interview is one of those moments that traditional transcripts completely lose. You see it in real time, the participant moves on, and by the time you are reviewing notes it is gone.
Would love to hear more about what you were testing, if you want to run a more structured study, first one is free this month. https://aimoderator.entropik.io
Tried Mira for a quick concept test and the facial coding picked up subtle reactions I would have completely missed reading a transcript. The automated report came back faster than I expected and the emotional insights actually felt useful rather than just a novelty.
Mira
@krakl9tl6 "Useful rather than just a novelty" is exactly the bar we hold ourselves to. A lot of emotion AI tools add a layer but do not meaningfully connect it back to the insight. Glad the report turnaround and emotional signals both landed as useful for your concept test.