How to make your store search learn from what shoppers click and buy
AlFinder's search learns from what shoppers do after they search. When several different shoppers keep clicking a product and adding it to their cart for the same search, that product moves up the results for that search. It works on its own and updates every day, with nothing for you to set up.
- 0:47 video
Why matching words is not enough
Many searches are broad. A search for "bag" matches handbags, laptop bags and travel bags, and the words alone cannot tell which one your shoppers usually want. Ordering by text gives a reasonable list, but not always the list your shoppers would choose.
Your shoppers answer that question every day through what they click and add to cart. Self-learning search turns those choices into better ordering, so the next shopper who types the same search sees the popular choice sooner instead of scrolling past products that rarely sell for it.
How self-learning works
After each search, AlFinder notes which results shoppers click and which they add to cart. An add to cart counts for more than a click, because it shows stronger intent. Each search term learns on its own, so what shoppers pick for "abaya" does not change the results for "black abaya".
A product only moves up when several different shoppers choose it for that search, so one shopper clicking the same item again and again does not change anything. The learning looks at the last three months of shopper activity and refreshes every day, so as tastes and stock change, older patterns fade out on their own.
What it looks like in real stores
In the video, a search for "bag" first shows handbags. Shoppers on that store kept choosing one laptop bag for that search, so the laptop bag now comes first. The same thing happens in any category where one search covers many products.
- Perfume: shoppers searching "oud" keep picking one دهن عود, so it rises above sprays and incense that also mention oud.
- Electronics: for "charger", the fast charger most shoppers add to cart moves ahead of older cables.
- Groceries: for "rice", the bag size your regular shoppers buy shows up before the sizes they skip.
Working alongside your own choices
Self-learning reflects what shoppers prefer, which is not always what you want to promote. When you need a specific product at the top, such as a new launch or a clearance item, set a rule in Merchandising to decide the order yourself.
It also needs shoppers to click something before it can learn. Searches where nobody clicks a result show up as negative searches in your analytics. Those are the ones to fix by hand, with a synonym, a clearer product title or a product you don't stock yet.
Common questions
Do I need to turn self-learning on?
No. It works automatically, with no switch or setting to change. It starts learning as soon as shoppers search, click and add products to their cart.
How quickly will I see results change?
Learning updates once a day. A search needs activity from several different shoppers before its order changes, so your most popular searches improve first, while rarely used searches may keep their usual order.
Can one shopper push a product up by clicking it many times?
No. A product moves up only when several different shoppers choose it for the same search, so repeated clicks from one person are not enough.
What if I want a different product first for a search?
Set a rule in Merchandising. Merchandising rules let you choose the order yourself, while self-learning keeps improving the searches you leave to it.
