In this article
What does consumer research for food actually measure?
Where sensory analysis characterizes the product — how sweet, how crunchy, how creamy — consumer research characterizes the response: does the target audience like it, prefer it over alternatives, feel something about it, and intend to buy it? For a food brand, that response is what ultimately decides a launch.
The stakes are well documented: peer-reviewed estimates put new food product failure at 50–75% within two years. Much of that failure traces back to decisions made on what consumers said rather than what they do — the say-do gap. Good consumer research is designed to close that distance.
The main methods, and what each is for
Qualitative: focus groups and interviews (why do people react this way?)
Small groups or one-on-one conversations that surface language, associations, barriers and usage rituals. Strong early in development, when you are still shaping the concept. Weak as a launch predictor: small samples, group dynamics, and everything is stated rather than observed.
Quantitative: surveys and hedonic testing (how much do they like it?)
Larger samples score liking (typically on the 9-point hedonic scale), preference and purchase intent. This produces comparable numbers across recipes and concepts. The limitation is that scores are still self-reports — they capture a considered opinion, not behavior.
Central location tests, or CLTs (which sample wins under controlled conditions?)
Consumers come to a facility and taste coded samples under identical conditions. CLTs excel at clean product comparisons — same temperature, same portion, no branding. What they cannot tell you is what happens when the product meets a real kitchen and a real household.
Video interviews (fast qualitative signal at quantitative scale)
Short recorded interviews — Eatpol's Vox format is about 15 minutes — where consumers respond to a concept or product on camera. AI analysis of language and expression turns hundreds of these into structured insight in days rather than weeks.
In-home use tests (IHUT): testing where buying decisions are made
An in-home use test (IHUT, also called a home use test) ships the product to consumers, who use it at home the way they would after buying it: cooked on a weeknight, served to the family, eaten from their own bowl. It is the only method where your product competes with real life — the household's taste, the schedule, the usual brand already in the pantry.
| Aspect | Central location test (CLT) | In-home use test (IHUT) |
|---|---|---|
| Context | Controlled facility, coded samples | Real kitchen, real occasions |
| Best at | Isolating pure product differences | Predicting purchase and repeat usage |
| Data | Scores under identical conditions | Usage behavior over one or more occasions |
| Logistics | Facility, staff, travel | Shipping to the consumer's door |
| With video + AI | — | Observed bites, emotions, spontaneous comments |
Which method at which NPD stage?
| Stage | Question | Fitting methods |
|---|---|---|
| Ideation | Which concepts resonate? | Video interviews, qualitative interviews, trend analysis |
| Recipe development | Which formulation wins? | Discrimination and descriptive sensory analysis, CLT |
| Validation | Will people buy and re-use it? | In-home use test (IHUT), hedonic testing with target consumers |
| Pre-launch | Does the full proposition land? | IHUT with packaging and price exposure, video interviews |
For a step-by-step walkthrough of this whole sequence, see how to test a new food product.
Costs and timelines
Focus groups are priced per session; CLTs carry facility, staffing and incentive costs; IHUTs are priced mostly per participant including shipping. As a rule, methods that use the consumer's own device and home have lower fixed costs than facility-based ones — there is no venue, staff or travel to pay for.
Timelines follow the same logic. Facility studies typically take 4 to 8 weeks including recruitment and reporting. Platforms that recruit from an existing consumer community and automate analysis deliver in about one week.
The AI video approach: what Eatpol adds
Eatpol, a spin-off of Wageningen University & Research, runs video-first consumer research: Vox video interviews for fast concept feedback, and Domus in-home tests where consumers cook and eat your product on camera in their own kitchen. Computer vision analyzes bites, chewing, facial expressions, eating speed and spontaneous comments — behavioral signals that correlate with purchase intent, not just stated scores.
Recruitment comes from Eatpol's own tester community, results arrive within one week on average, and all footage is pseudonymised — faces are automatically masked by AI and data is handled under GDPR.
Curious what your next product could be in the first place? Eatpol Nova mines award-winning launches and cross-country trends for validated ideas.