A taste test measures the first spoonful. Most food is bought for the twentieth. A formula that wins on day one can wear thin by week three, and a formula that seems a little plain at first can grow on people. An in-home use test is how you find out which is which before the product reaches the shelf.
The method
Households who already buy the category receive the products at home, blind and marked only with a code. They use them the way they normally would, at breakfast, in a lunch bag, in a smoothie, for several weeks. A short rating after each use, a weekly check-in and a count of what was actually eaten turn ordinary life into data.
All of it runs through one integrated study app on the participant's phone: the ballot after each use, the weekly check-in, a photo count of what is left in the fridge and short video diaries. Every answer is time-stamped and lands in one place as it happens, so nothing depends on paper forms or memory, and we can see in the first days if a household falls behind.
The question it answers
Which version should we launch, and what should we fix first? A taste test answers half of that. The in-home test adds the half that decides repeat purchase: whether liking holds, whether usage holds, and why it changes when it does.
How we run it
We screen for real category buyers and set quotas so current customers and competitor customers are both in the room. A typical design is sequential monadic: each household tastes every product, one at a time in a rotated order, and rates each on the same ballot. Then each household lives with one product for the rest of the test. The ballot uses the standard 9-point liking scale plus just-about-right questions for the attributes a developer can actually change, such as sweetness or thickness. Instead of a written diary, participants record a short video in the app whenever something goes right or wrong. People rarely write down the morning they skipped the product. They will show you the cup pushed to the back of the fridge.
What you get
A readout that puts first-taste results next to week-by-week results, a penalty analysis showing how much liking each miss costs, coded video clips from the app that explain every change in the numbers, and a plain recommendation: launch this one, fix these things, in this order.
Where it goes wrong
Tests that stop after one or two weeks miss the drop that usually arrives in week three. Penalty analysis gets misread when every small miss is treated as a fix; only misses that affect a meaningful share of people and cost real liking deserve attention. Cells that are too small cannot be split by subgroup, and teams try anyway. And participants know they are in a test, so we weigh what they ate more heavily than what they say they would buy.
The sample below follows a fictional plant-based yogurt brand testing two new formulas against its current product. Click the tabs to move from the design to the ballot, the week-by-week results and the readout. Every brand, household and number is invented.
Illustrative only. All brands, products, households and figures are fictional. Open the sample full screen.