Where was the problem?
The initial recommendation system on the e‑commerce platform was prepared in a very basic way. Persoo added extra connections between products: what product is selling with which item, recommendations based on personal preferences (color, brands, type of clothes) but also both long‑term and short‑term trends.
How did it end up?
Customers' behavior has changed - they now spend more time on the website, browse more pages, but what’s more important - they return more often now. We’ve monitored an increase in visits from returning users by 27 %. Turnover increased by 24 %. We achieved it by improving the conversion rate by 7.8 %, increasing the number of visits per visitor, and increasing the average order by 1.7 %. Assisted conversions, in which Persoo affects half of all sales, were very successful as well and also increased the average order value for these assisted conversions.
News on the homepage was set manually. Customers were often confused by the huge amount of different brands.
We added personal recommendations based on previous behavior i.e. what the customer browsed, what types of clothes (swimwear), what colors and which brands. CTR increased five times to 16.5 %.
There were no recommendations in categories. Customers had to search, filter or browse through multiple pages.
We sorted filters according to personal preferences and put them in the first line of product feed. We took preferred color, brand, what are similar visitors browsing, what the customer has not seen before, and labeled products and special offers into account. Best‑selling and most‑browsed products are being slightly preferred. Most men see any difference between multiple swimwears but trust us - to pick the right color is not an easy task! CTR increased to 9.8 %.
There were no recommendations in product detail pages. Customers did not get the idea of buying second, colormatching upper part for the swimwear.
We placed two feeds below product details. One with recently viewed products, and the other with recommended products (according to customer preferences) from the same category. CTR of recommendations increased to 47.1 %. CTR of the recently browsed products was 10.2 %.
There were no recommendations in the pre‑checkout. What a pity.
We added a feed of products which are somehow related to what the customer put into the shopping cart to the checkout. The relation between these products was set according to what other customers bought together with the main product. CTR in pre‑checkout increased to 5.2 %.