Most insight decks are full of facts. A real insight explains the behaviour and changes the decision.
What is a consumer insight?
A fact describes. An insight explains and points to action. "Sales of compact models rose" is a fact. Why they rose, and what you should change because of it, is the insight.
For a midsize consumer goods brand, manufacturer or retailer, an insight earns its place by moving a choice. It should change where people, budget or capacity go.
A useful test has three parts. It is true, shown by more than one source. It is not obvious to the team. And someone would decide differently because of it.
Fact: Interest in small kitchen appliances is rising. Observation: People compare them on storage, not on power.
Insight: Buyers treat counter space as the real constraint, so they judge a product by where it goes between uses. Implication: Lead with size and storage.
What are the main sources of consumer insights?
| Source | Shows | Weakness |
|---|---|---|
| Surveys | Stated attitudes at scale | People report, not act |
| Interviews | Motivation in depth | Small samples |
| Purchase panels | What was bought | Only after the decision |
| Social listening | Public conversation | Skews to people who post |
| Search and AI orientation | What people look for | Shows the what, not the why |
| Synthetic respondents | Fast hypotheses | Lose rare answers |
Use at least one source that asks and one that observes. Agreement raises confidence. Disagreement gives you the next question.
Nielsen Norman Group compared 120 synthetic respondents with a survey of 604 professionals. Both reached the same overall conclusion, but the synthetic panel lost the rare answers.
Rare answers are often where change starts. Use synthetic panels to sharpen a question. Do not use them as proof.
Why do people say one thing and do another?
That difference is the say-do gap. Economists separate stated preferences, which people report, from revealed preferences, which show in actual choices. Both are real. They answer different questions.
A 2026 industry commentary says the gap is "not a flaw in consumers". It calls the gap "a signal about how decisions are really made".
When people say they want something and do not buy it, ask what stops them. Price, habit and availability are common answers. Often that barrier is the real insight.
People are honest about what they intend. Their behaviour is honest about what happens next.
How do you build an insights process that lasts?
- Start from the decision. Write the choice down before you commission anything.
- Watch before you ask. Use behavioural data to find what is moving. Then research why.
- Triangulate. Let no single source carry a major decision.
- Keep it current. Replace one-off studies with a running view where you can.
- Record outcomes. Note what you acted on and what happened.
The second step is often reversed: teams survey first and look at behaviour later. Starting with behaviour makes the research cheaper and sharper. It also shows you which questions are still worth asking. Consumer Intelligence, a perspective within Demand Intelligence, supports that first look.
Where is the insights industry going?
Toward software and continuous data. ESOMAR shows research software growing 11.5% in 2024, against 4.8% for market research. The share of turnover from firms actually conducting research fell by about 3 percentage points against 2023.
Faster tools do not remove the need for judgement. They raise the value of knowing which question to ask. To see the wider picture, read our guide to Market Intelligence.
Frequently asked questions
What is the difference between consumer insights and market research?
Market research is the activity of collecting evidence. A consumer insight is a conclusion drawn from that evidence, with an implication for a decision.
What is the difference between customer and consumer insights?
Customer insights usually concern your own buyers. Consumer insights cover the wider market, including people who have not bought from you.
How many sources does an insight need?
At least two of different kinds, ideally one that asks and one that observes. A single source gives you a finding, not yet an insight.
Sources
- Say-do gap, Marketing Research and Insight Glossary (Quirk's)
- Annex 8: stated and revealed preferences (UK FCA, MS17/1.2)
- Why consumers say one thing and do another (NIQ, 24 Mar 2026)
- Synthetic Users vs. Human Respondents (Nielsen Norman Group, 16 Oct 2026)
- Inside the $153bn Insights Industry (ESOMAR / Research World, 4 Nov 2025)