Predictive search – First Step Towards A Great Site Search Experience

Predictive search – First Step Towards A Great Site Search Experience


This is a 4 part series on how to deliver a stellar a site search experience. In this post, we will share usability guidelines on designing a predictive site search – one that anticipates user intent and intelligently guides them towards the right product.

In our previous post we talked about the importance of relevancy for eCommerce site search, the 12 different query types that customers typically use to find a product, and how to address these. Let’s look at another pivotal building block for a great site search experience: Design. Pretty much like technology, usability plays an equally important role in making site search more engaging, easy to use and highly converting. It’s hard to imagine one without the other, especially when the site abandonment rate due to a poor site search experience is so high – almost to the tune of 80%! Therefore, to make your site search usable or human ready, design comes into play.

Designing predictive search:

                                                             ‘Anticipation denotes intelligence

This is a beautiful quote from the movie The Fifth Element, which seems highly relevant in a world that is moving towards smart systems with predictive capabilities. The operating word here is anticipation and it holds true for eCommerce site search as well i.e. to anticipate correctly what the customer expects and wants, and showcase relevant products accordingly. In the eCommerce context, predictive search refers to search autocomplete.

The primary benefit of a search autocomplete is to help users frame a better search query. In Baymard Institute’s research on eCommerce site search usability, we found that over 82% of the major eCommerce sites now support autocomplete. However, designing autocomplete can be tricky as customer behavior differs from site to site.

At Unbxd, we work with each of our clients to customize their autocomplete based on their business & how their customers prefer to interact with their site.

For instance, for one of our customers in the fashion space – Deborah Lippmann – it was important to take customers to the right subcategory just like a salesman would. If a customer searches for a pink nail color, it is important to prompt them to narrow their search to relevant sub-categories such as glitter or shimmer. Also considering most of their products are brands in themselves, showing best sellers in autocomplete led to higher click-throughs.

Predictive search – First Step Towards A Great Site Search Experience

For another customer that operated in the equipment and hardware space, we realized that a lot of searches are being made using part numbers and models. For this we customized the autocomplete to show search suggestions for both.

Predictive search – First Step Towards A Great Site Search Experience

Here’s to summarize a few essential design hacks for site search autocomplete:

  • Show best sellers to help customers get to the right product/product  category

  • Don’t show irrelevant or out of stock products

  • Don’t necessarily stick to a templatized autocomplete – customize your autocomplete as per your business needs and your shopper’s behaviour to let them search products faster and with ease.

Predictive search – First Step Towards A Great Site Search Experience

More To Explore

AI for Customer Loyalty
Machine Learning

How To Leverage AI for Customer Loyalty

Artificial intelligence seems like it’s everywhere these days. That’s for a good reason: artificial intelligence gives everyone access to a flexible set of tools that can