Predicting food purchases with purchase intention

Elina Kytö, Senior Research Scientist, specialising in sensory assessment and research at Valio, defended her PhD thesis at the University of Helsinki’s Faculty of Agriculture and Forestry's Food Chain and Health doctoral programme.

In her dissertation, Elina Kytö studies the ways to predict food purchases using purchase interest and other measurements. Her research theme (Ostoaikomuksesta ostokäyttäytymiseen: ostokiinnostus sekä muut mittarit ruokaostosten ennustajina – From purchase intention to purchase behavior: Purchase intention and other measurements as predictors of food purchases) she discovered over the course of her work.

“My work is in running studies on products that are still in the product development stage, where we think a lot about what kind of results in product testing are good enough to predict commercial success. I wanted more information on the best measurements that could be used as predictors for purchase behaviour,” says Elina Kytö.

Elina’s thesis work was supervised by Research Manager Sari Mustonen, whose expertise covers sensory assessment methods for food products, having been a Valio employee at the start of the work:

“It was incredibly valuable to have Sari and her long Valio background to help. She knew the background so well. Even after leaving Valio, she stayed on as my thesis supervisor. I took a lot from my job into the dissertation but wrote most of it outside of office hours. Sari and I started going on thesis cruises once or twice per year, to better focus on a specific topic together,” says Elina.

Which measurements are best for predicting purchase behaviour?

The goal of the study was to see how purchase intention as well as other direct and indirect measurements work to predict food purchases. The focus products were unflavoured yoghurts and flavoured protein quarks.

In the study, subjects were given different amounts of information on the product, they were asked a set of different, direct questions, and then kept a purchase log for one month. Some participants were additionally subject to so-called indirect, physiological response measurements: on the day following product testing, their reactions to the products and packages were observed. In these tests, we observed the subjects’ pupil movement, brain activity and reaction speed.

The study showed that purchase intention and recommendation were the best measurements for predicting purchase behaviour. As a slight surprise, indirect measurements were somewhat poor predictors for purchase behaviour.

How much should we say about a product?

The study clearly showed that the more information on a product is available, the more likely it is for a purchase intention to be realised as a purchase decision in the store: the combination of the package and tasting produces better results than just the package or blind tasting.

“The reason for this is that people create expectations based on information on, for example, the product’s taste and other sensory properties. For yoghurts and quarks, the package creates expectations that the product then either meets, or it ends up being a disappointment. It’s important for a product to have overall appeal. The package is what entices a consumer’s purchase, and naturally, product quality must match for the product to find its place in the shopping basket again,” explains Elina.

Based on this study, the best predictability is gained through introducing the package to a tasting session. The more real the test situation feels, the closer we get to true purchase behaviour.

“Blind tasting does, however, still have a purpose, such as in studying product quality and the ways samples differ in their sensory properties,” says Elina. Product testing is usually carried out in the stage that the feedback can be used to make changes to the product or the package design.

“If significant changes are made to the product or package after testing, and a new test is not carried out, it's good to remember that the study was done on a different thing entirely. That study will, then, not be able to predict purchase behaviour,” reminds Elina.