Many customers buy more specifically in the online shop than in the stationary store. Digitally, they want to find the right thing as quickly as possible, regardless of time and location – and that works best with an intelligent search. But what exactly do you mean by that? What must a search be able to do to be “intelligent”? In this post, I’ll give you a checklist.

An intelligent search compensates for input errors

Ideally, you don’t give your visitors a reason to leave your shop early. Even if error tolerance is now an eCommerce standard: In many online shops it is still the case that visitors end up on the zero hit page if they already make small input errors. Be it because they don’t know exactly how a word is spelled, because they are hectic making mistakes, because their smartphone keyboard is too small, etc.

An intelligent search is able to compensate for spelling mistakes and show the correct products. It also automatically recognizes plural and singular forms. And in the case of compound words separated by a hyphen (“hiking shoes”), it also searches the index for the variant written together (“hiking shoes”).

All of this makes an intelligent search without having to rely on “stemming”: This refers to a process in which different variations of a word are traced back to its common root – so that the search can theoretically infer other related words.

But if your search is based on a good similarity algorithm, you don’t need stemming. Instead, all entries in your product data that are similar to the search term are found and classified according to relevance. There is no such thing as “nothing found”, at most “nothing found that would be relevant enough according to the user settings”.

It is important to note that what is “similar” is not only determined by the letters, but also, in particular, by the sound of the word. The classic example of why a search should also identify phonetic similarities is the term “Eau de Toilette”. If you are looking for this term for the first time and have no knowledge of French, there is a high probability that you will incorporate spelling mistakes. In the extreme case, you enter something like: “oh day teulet”. A search term that hardly matches the desired product type, but at least sounds similar. If your search can handle such discrepancies, you are offering your customers a real service – like for example Douglas.

An intelligent search brings customers into the long tail of your range

If customers already have a precise idea of ​​what properties the product they are looking for should have, they enter very specific terms and word combinations. Such complex search queries are often aimed at products in the long tail of your range.

The term “long-tail” was originally coined by Chris Anderson, who used it to describe the sum of niche products as a success factor for online shops. These products are seldom sought, but due to their size they have an almost inexhaustible sales potential.

If a user makes the specific search query “red men’s running shoes from Nike”, they probably know exactly which product they want. This makes it very likely that he will become a buyer – in contrast to a shop visitor who is only looking for “shoes”. General search queries are canceled more often because the purchase decision has not yet been made in most cases. Specific search queries are less common, but they sell significantly more. And in many cases, they bring customers to high margin niche products.

The more complex the search query, the greater the intention to buy.

As a result, you as a retailer benefit twice when your customers get relevant results for their multi-word search queries. However, this is exactly where the problem usually lies: It is difficult for online shops to recognize all possible word combinations and all imaginable incorrect spellings – without resorting to manually created word lists.

You can teach your search engine 10 or even 1000 long-tail keywords. But such an effort is impossible to do with 100,000 keywords. To take advantage of the sales potential of long-tail searches, you should therefore use a search function that can deal with complex search queries automatically and intelligently.

Let’s take the product “amethyst ring”. If a shop visitor uses the phrase “ring with amethyst stone”, many search functions are already overwhelmed. But an intelligent search, the algorithm of which was modeled on the human perception of similarity, can recognize word components in all conceivable combinations. Even without long, manually created word lists. And even if the word “ring” is in the product name, “amethyst stone” is somewhere in the description field.

Intelligent search uses AI to optimize search results

Who does not know it: You search for “smartphone”, but the search first of all shows all possible smartphone cases. This is because product texts usually contain the same terms as the texts for accessory articles. So it often happens that the search cannot distinguish this – which is of course annoying for the customer and anything but goal-oriented.

Another example is when a customer searches for “blue pants” at a fashion retailer. This description applies exactly to jeans products, but he hardly has that in mind, because otherwise he would have simply looked for “jeans”. Most likely the chinos that are clicked on, which appear further down in the results.

With a patented AI called Semantic enhancer a function is available that recognizes the connections between search queries and subsequent customer transactions. The algorithm learns which product is required when: With every click on product detail pages, every shopping cart content and every completed sale, the quality of the hits increases and with it the satisfaction of the shop visitors.

By the way: The Semantic Enhancer has been integrated into our FACT-Finder search technology as a standard component for several years and is one of the main reasons why FACT-Finder was named “Best Site Search Solution”.

An intelligent search provides a variety of input assistance

Anyone looking for something online often trusts that the shop knows better search terms than you do yourself. An intelligent search therefore always includes a suggest function that provides suitable suggestions after entering the first letters in the search field.

You can also show other data such as prices, customer ratings and brand logos in the suggestion list. This also supports entry into your range – provided that you don’t overdo it. Because an overload of text and images only confuses the user.

So which information is relevant and which is not? Basically, you should set priorities: Depending on the industry, product range, target group and business strategy, you should decide which elements you want to use in addition to the actual search suggestions. As a tip: Most users will help category suggestions, product suggestions and prices. And if you have an assortment that requires explanation, advice content could also be very useful – even a shopping cart function in B2B.

intelligent search suggest

Intelligent search on B2B buyers immediately receive relevant product suggestions and can fill their shopping cart within seconds.

You should also think carefully about which suggestions you want to show at the top. The ones with the highest search volume? Those who are currently doing a doctorate? Or is it the one with the highest margin? The answer comes from your sales strategy and the image you want your shop to convey. Those who advertise the quality of their offer would want to show products with the best customer ratings, for example. Your smart search can then choose which products best fit your rules.

An intelligent search takes into account the person behind the search query

Too many online shops orientate themselves solely on the typed search word – not on the person who enters it. So everyone gets the same results, in the same order. If you can’t find the product you’re looking for straight away, you either have to set a filter or enter a more specific search query – in both cases the conversion is still a few clicks away.

By personalizing your search, you can greatly simplify the shopping of your customers and stay with them in a positive way. Provided that you don’t personalize it broadly based on target groups or personas, but rather individually.

By tracking user behavior through click, shopping cart and purchase events, your search can learn what buying patterns and preferences your customers have. Such preferences can be, for example, certain brands, categories or colors. On the basis of this, the shop visitor can be presented with precisely those products that match both his request and his basic preferences.

Of course, everything has to be data protection compliant. Your intelligent search should not process any personal data such as name, address etc. – only tracking data and IDs. These are either cookie-based session IDs or user IDs that are assigned when you log in.

Smart search personalization

The better the intelligent search understands the customer, the higher the conversion rate and customer satisfaction.

Are you a retailer with a stationary branch network? Then you can even go one step further: In industries such as DIY or electronics, customers like to deal with product selection at home and be inspired by suggestions and worlds of experience in the online shop. If your search uses geo-localization, it can display the results in the shop based on the customer location: those products that are available in the nearest branch can then be displayed in a targeted manner. A typical omnichannel concept to strengthen your stationary branches. You can find a specific application of the geo search in the user report of our customer OBI …

Smart search improves your entire shop

Finally, one more criterion that is underestimated by many: the learnings that your search collects should also improve other shop functions such as navigation, recommendations, etc. For example, your recommendation engine can learn something about the interests of your customers from the intelligent search. And your navigation can improve the product results on category pages based on the filter behavior – and vice versa. The more your shop learns in different places and uses this knowledge, the more appealing the buying experience for your customers and the higher your conversion rate.

Checklist: all criteria at a glance

An intelligent search …

  • compensates for input errors. It automatically recognizes plural and singular forms, composite words and phonetic similarities.
  • brings customers to the long tail of your range. It understands even complex search queries that consist of several words and reliably shows the right (niche) products.
  • uses AI to optimize search results. It learns independently what expectations are behind your customers’ search queries and improves continuously.
  • gives a variety of input assistance. It already turns the suggestion menu into a separate sales area with products, prices, category suggestions and more.
  • takes into account the person behind the search query. She personalizes the search hits based on the preferences of the individual customer …

By admin

Leave a Reply

Your email address will not be published. Required fields are marked *