A voice check-in

Having seen clients fail to connect with research when it is delivered, I have learned that numbers are not enough. An insight moves between people when it carries a story: one person's experience, supported by the wider evidence. It also has to remain recognisable as it is retold from one person to the next. Packaging is therefore part of the research job, not something that happens after the analysis.

Qualitative evidence gives us experience, language, explanation, and contradiction. Before asking how common something is, it is reasonable to understand the range of ways it happens for individuals. A score such as a 0–10 recommendation rating can be useful to track, but it deliberately compresses the reason behind the answer. If we never collect that explanation, no amount of later analysis can restore it.

The obstacle has always been the work involved. Focus groups require careful facilitation and can place a social burden on someone who disagrees with the first confident answer. One-to-one interviews give people more room, but demand more collection and analysis time. Transcription, repeated reading, coding, recoding, comparison, and checking another researcher's interpretation all take serious effort when the work is done properly.

The economics are changing. Automated transcription, search, assisted organisation, and new analytical tools make larger bodies of qualitative material easier to work with. They do not remove the obligations to protect participants, preserve the source material, trace interpretations back to what people actually said, and check the analysis. They do make it more practical to collect human explanation alongside structured measures instead of treating it as an occasional luxury.

That matters even more in an environment filled with generated answers. Reviews and first-person discussions are valuable because people want to know what another person actually experienced. Research can ask for that experience directly and in relation to the specific decision at hand. Synthetic evidence can suggest what to ask; a recorded human response tells us how the matter appears in a life.

This concept is a short voice check-in: a small series of purposeful, semi-structured questions that people answer aloud. It is not intended to replace structured data collection. It provides nuance, gives colour to the measured pattern, and tests whether a claim still makes sense when people explain the experience in their own words.

I would begin with voice rather than video because it asks less of the participant while preserving tone and expression. That is a design hypothesis to test, not an assumption about every participant. Transcription quality also needs to be established rather than presumed. A short warm-up can help someone settle into the task and confirm that their speech is being captured accurately before the substantive questions begin.

The structure can remain light. A good written prompt and simple rule-based follow-up may be enough: if someone describes a good experience, ask what made it good; if they describe a poor one, ask what happened. AI can assist with transcription, organisation, and analysis in the background. The participant should still be able to see what was captured, correct it, and understand how it will be used.

The richest version is repeated and close to the moment of experience. Diary and experience-sampling methods collect reflections within people's lives over days or weeks, reducing reliance on one retrospective account and revealing how experience changes (Csikszentmihalyi & Larson, 1987; Stone & Shiffman, 1994; Bolger, Davis, & Rafaeli, 2003). Short voice check-ins fit that shape. They also ask more than ticking a box, so the exchange needs enough time, a clear purpose, and fair, predictable payment (Church, 1993; Singer & Ye, 2013).

This method will not suit every question, and a purposive sample should not be presented as representative of everyone (Baker et al., 2013). Its value is more fundamental: understand the spectrum of human experience before trying to estimate how widely each part of it is shared. The visual concept is not built or validated yet. The next step is to test whether people will use it comfortably, whether the recordings are accurate, and whether the resulting material improves the decisions made from the research.