TAGS: AI CX Merchandising Online Shopping Retail
Take onsite search challenges head on with our expert tips.
Imagine you have a quiet Friday afternoon, and you would like to optimise your site to offer your customers a better shopping experience. Where should you start? Unless you have a well-defined optimisation plan and a dedicated team using the platform, it is hard to know the best starting point.
Customers won’t buy what they can’t find. So it is key to make their visit to your site as successful as possible. Reviewing and optimising your search strategy and configuration should be part of a regular review of the platform (for example, quarterly), but you could start with a few easy steps to improve the findability of your products.
A good start would be to take your top (for example, 50) most frequent zero-result searches from your analytics system. How you treat zero results determines greatly the customer experience. A zero result without alternatives will very likely make the customer think you don’t sell that product, and it can potentially make him leave your website.
First question to ask is, ‘why is it a zero result in the first place? Do I even sell this product?’ If the answer is yes, this could indicate an issue in your search configuration. But more often, customers are searching for a product differently than what it is actually called. One common example is customers searching for ‘tv’ and expecting to find televisions, but it is very likely that ‘tv’ is never mentioned in the data, making it very difficult for the search engine to find.
Setting up a synonym would help you to overcome this problem. Defining ‘tv = television’ will allow you to return televisions for both keywords. A few things to keep in mind, though, when creating synonyms. If configured correctly, the search engine ignores capital letters and special characters, applies stemming, and tries to correct for spelling mistakes. No need to define synonyms for all of those cases.
Next thing to keep in mind is how specific each of the keywords are, and what type of synonym to use. A synonym consisting of a very general term and a very specific term should be treated carefully. Extending our example to ‘television=tv=hdtv’ clearly helps to show high definition televisions (let’s assume it is only called ‘hdtv’ in the data and there is no reference to ‘tv’ or ‘television’ whatsoever). However, a user specifically searching for a high definition television would find also regular televisions or 4k televisions, which is not what he is looking for. Setting a unidirectional synonym would make more sense in such a case, e.g. ‘television>hdtv’. Search results on ‘television’ will include ‘hdtv’, but not the other way around.
The examples above describe a situation in which you do have the product, but the customer is used to calling it differently than you do in your data. It is often the case that you don’t sell the brand the client is looking for, but you do have similar products. A quick-and-dirty solution could be to set up synonyms to take those searches into account, but it will only confuse the user. If the customer searches for a specific brand of television, showing all televisions will not inspire confidence. A neater alternative would be to create a redirection to a landing page of the corresponding category. Ideally, this would be combined with feedback from the front-end, mentioning that the current search didn’t return any results, but a relevant alternative has been proposed.
Last words on zero results, it is impossible to cover all the zero result cases beforehand, so think about a general page/campaign to show to the customer. A well-designed navigation page, for example, could allow the user to jump directly to the category he is interested in.
Once you are done with creating rules, and it’s almost ‘beer o’clock’, one last thing I would definitely recommend is some clean-up of the old rules you are not using anymore. Large amounts of legacy rules can make you quickly lose the overview. Rules can be placed in the recycle bin if not needed anymore, and otherwise can be stored in the archive for later use. Just like in Windows, consider also emptying the recycle bin once in a while, as well. This makes the platform neater, more workable and ready for the next week’s use. It also contributes to its general responsiveness.