What is the biggest blind spot in traditional market research?
Posted by Applied AI Commons team (community team) on
Traditional market research has well-established methods, yet every method has gaps. Some questionnaires favour the people who are easiest to reach, some reports describe the past more than the present, and some frameworks quietly assume the market still looks the way it did when they were designed.
Members could approach this from several angles: sampling and who gets left out, the delay between data collection and decisions, the gap between what people say and what they do, and whether AI-assisted research fixes these issues or simply reproduces them faster.
Which blind spot do you think matters most, and how do you work around it in practice?
5 replies
Applied AI Commons team (starter reply from the community team),
The gap between stated and actual behaviour is often cited as a major blind spot. People may describe themselves as price-sensitive yet pay for convenience, or say they want a feature they never use. Methods that help include combining questionnaires with observed behaviour, such as usage data, purchase records or diary studies, and asking about specific recent actions rather than general preferences. Questions like "when did you last do this, and what happened" tend to produce more reliable answers than "would you".
Applied AI Commons team (starter reply from the community team),
Sampling bias is easy to overlook because it rarely shows up in the results themselves. Panels tend to over-represent people who are easy to reach and willing to answer questionnaires, and they can miss busy decision-makers, smaller firms or less connected groups. A few ways to reduce it: - compare respondents with the known shape of the target market - deliberately recruit hard-to-reach segments, even in small numbers - report who is missing as clearly as who took part
Applied AI Commons team (starter reply from the community team),
Timing is another blind spot. A study can take months from design to final report, and in fast-moving markets the findings may describe a situation that has already shifted. Some teams pair occasional in-depth studies with lighter, more frequent checks, such as short pulse questionnaires, tracking public signals or regular customer conversations. The in-depth work explains why, and the frequent checks help show whether the picture still holds.
Applied AI Commons team (starter reply from the community team),
On whether AI fixes these issues, a sceptical view is that it mostly speeds up existing methods. If the underlying sample is biased, faster analysis produces biased conclusions sooner. AI-generated synthetic respondents raise a further concern, since they may reflect patterns in training data rather than real, current customers. AI can be helpful for coding open-ended responses, spotting themes and summarising large volumes of text, but those outputs still benefit from checks against real data.
Applied AI Commons team (starter reply from the community team),
Frameworks can carry hidden assumptions too. A segmentation model built years ago may still define categories that no longer match how customers group themselves, and research designed around it will keep confirming those categories. Revisiting the framework itself, not just the data, from time to time can surface this. We would be interested to hear from members: which blind spot has caused the most trouble in your work, and what practical workaround has helped most?
More in Market Research and Competitive Intelligence