To buy new products, a lot of consumers do not go to the shop anymore but buy products on the internet. To sell more products, companies offer different approaches on recommendation techniques, because in general, product recommendation leads to more sales.

The first approach is Rule based non compensatory, which can be based on Elimination by aspects (EBA)and lexicographic. Rules based non compensatory on Elimination by aspects is showed through the red lines in the picture of bol.com.

Consumers first make a choice between different products and then make a choice in the product category. In this example consumers first choose books (instead of DVD’s, toys, games etc.) and then which languages, like English, Spanish or French. Rule based non compensatory on lexicographic is showed through the red lines in the other picture. It is the same as elimination by aspects but in this case people can fill in what they think is most important. So if they think a low price is important they get camera’s with low prices.

Drawbacks for non-compensatory technique are that it makes it complex for more options, because consumers has eliminated a lot things to come where the want to be. It’s easy to imitate for other companies and it passively provides information, which mean that inexperienced consumers don’t know what to with it. As example the pixels from the camera, what are 6.5 pixels?

The second technique offered by companies is Rule based compensatory is based on weighted added evaluation and equal weights. Consumers can fill in which aspects are important and exactly how important each aspect is. In the first picture the consumer chooses only for one aspect (body &size)  and in the second picture this aspect is combined with price. The benefit from the technique is, that it is more objectively, witch gain a high utility. Drawbacks are that it’s complex for the beginner who does not know the product yet, it’s a higher effort for customer to adjust, ant the extended algorithm for the company.

The third technique companies use is Collaborative filtering. This is the process of collecting information of preferences about interest of many users to make a prediction about the interest for one person automatically. Collaborative filtering is shown in the following picture:

In the table on the right (named ‘andere bekeken ook’), consumers ,who are searching for the book of Steve Jobs, see the preferences of other books from other visitors, who were also searching for the book of Steve Jobs. Benefits of collaborative filtering are that it’s free date for conjoint analysis, it increases potential consumption and companies can provide information actively. Drawbacks are annoyed customers because of the tracked data and Business defamation issue from competitor.

So we gave some benefits for companies of the recommendation technique, like that companies can sell more products due to the opportunity to show consumers products that are related to the product they are interested in. But for consumers there are several benefits as well. It can Help overcome search limitations for consumers. It sorts alternatives in terms of attractiveness. It reduces cost of search and it facilitates comparison between alternatives. And at last does it increase decision quality.

Freek, Shrikesh & Kevin