Powering up findability with Silverstripe Search

As Product Engineering Lead at Silverstripe and the lead developer of Silverstripe Search, I spend a lot of my time thinking about how our customers and their end-users use their search tools to find content using the search function of websites.

Our clients normally have a very good understanding of their content but their users, well, they vary considerably in how they engage. Most organisations use specific and intentional language to describe what they do but a website visitor doesn't arrive at their site after an employee induction. We can see this in Silverstripe Search's analytics that shows queries users make that don't get any results. Perhaps someone is looking for what they call a "disclosure" form but officially it's called a "notification" form.

Tools to improve search results

We've got a few tools up our sleeve to help in these scenarios. The first is, perhaps unsurprisingly, called 'synonyms'. This allows you to tell your search engine to treat words as if they were the same. So searching for "disclosure" will also search for "notification". This is often helpful for things like expanding common acronyms.

Synonyms apply globally and sometimes you need more of a surgical intervention; that's where the curation features come into its own. This allows you to change what results you see for a specific query. Say someone searches for "cat pictures" and they would normally get a list of your finest feline prints. One week you run a promotion for picture frames and you'd like to promote it when people search for related items. With a curation, you can tweak the results, so that the product page for your frames appears first above the usual prints.

Supercharging search with AI 

So far I've only discussed traditional keyword searching techniques which are great because they are fast and cost effective. But there's a new world that AI technology has supercharged in recent years. Techniques like vectorisation have become mainstream now and they have had a big impact on the search landscape. Processes such as embedding takes site content or queries and maps it against categories that large language models (LLMs) have been trained on. This means it can access what the intention or meaning of the content is even if their wording isn't what we expect. 

Silverstripe Search currently uses this in its semantic search preview feature in combination with keyword searching. The natural next step, which is on our roadmap, is to use LLMs to understand visitors' natural language queries better, and generate high quality results that use the unique knowledge of our users' content. AI models provide a whole new way of interacting with machines. Silverstripe Search is building that in to meet the evolving needs of digital content creators.

We offer consultancy around Silverstripe Search to help improve your site's findability. We can work with you to ensure your visitors easily find exactly what they need. Book a chat with one of our search experts to get started.

Because your users should spend their time using your website in a valuable way, not playing hide and seek with it. Read more about our findability-related services.