Search – Translating intent into finding the right content  

There’s a core philosophy behind Search at The Filter, and that’s respecting the user’s intent. Whether they come to a service for something they already know they want to watch or are looking to interact with the service to find something new, we believe in providing solutions that deliver users what they’re looking for in the fewest possible clicks. A user who is resorting to placing actual text into a search box has probably done so because they didn’t get what they want presented to them already. It’s critical therefore that search gives users the best possible opportunity to find the content they want. Iyour search function does not deliver the right resultsthen users are left to seek that content elsewhere, increasing the potential risk for users to churn. 

Presenting a user with what they want to watch is a problem that starts before a user even attempts to search. Simple things like immediately providing a returning user a “Continue Watching” rail, allows a user to go back to the series they are binge watching. In addition, personalised recommendations that home in on content they are likely to watch, is an easy win that keeps users engaged. However, if a user still doesn’t find what they want from a platform, then their likely next step is to use search.  

Search becomes about understanding intent. What is a user looking to do? Do they know what to search for? Or are they using search to discover contentgather ideasexplore your catalogue? 

Effectively catering for user intent is not simple task and needs to be tackled in two distinct areasFirst and foremost, the user’s intent must be captured and understood. Secondlythe catalogue must contain the information needed that allows you to match content to that user’s intentBoth are deceptively complex problems to solve, and to misalign or focus on one area too heavily when designing solutions can disappoint a user who would otherwise have easily found exactly what they were looking for. 

Understanding user intent 

The first challenge with intent is to understand the distinction between what a user intends and what the user provides in a search query. In simple cases like a title search, a prefix-based match or a full-term match is likely to find what a user wantin only a few simple moves (example A below). Acting correctly on a more complex search term takes us past simple term matching and into the realm of needing to systematically translate the provided set of terms or phrase into intent (example B below). 

This is where the relationship between understanding what is input and having the data to support search terms comes into play. When a user approaches search with a term that just happens to line up with how their desired content is described in the catalogue the outcome is simple. But the reality is that all users are a little bit different.   

If we take a look at a word cloud for all searches used for EPIX’s new series of Godfather of Harlem as an example, you can see users have a whole variety of ways in which they will use search in terms of the input item. Evidenced by the variety of search terms used,  a search solution must be able to present options to a user when the desired intent is not as easy to establish.  

Matching Intent to quality metadata 

It’s easy for a user to find what you want if you know exactly how a service represents it, but how does a potential viewer looking to get their space combat fix know if a service has what they are looking for labelled as “Sci-Fi” or “Science Fiction”? A user searching for something “funny” should not miss out being presented content that was instead labelled just as “comedy” or mis-categorised entirely. From the perspective of the user, this should not matter. Their intention is only acted upon when we can apply both understanding of their query to an understanding of the available content, producing an outcome where they find what they want to watch.

To solve this dilemma at The Filter, we build our search experience from a combination of data sources to supply the richest possible experience for the end user. Ensuring that we are maximising the possibility that a user’s input can lead them to the content they’re looking for. Enriched and sanitised metadata from the original catalogue source is combined with our platform’s built-in entity system to further enrich that data, with a mixture of automated inference processes and manual, “common sense” assignments, linking thoriginal catalogue to a rich network of known and understood entities. By joining together these datasets, we’re able to take sparse or limited metadata catalogues and build feature-rich search around them. 

Empty space is a wasted opportunity to fulfil a user’s intent 

The benefit of having this approach to understanding intent and matching intent to feature rich metadata is that our technology will return results for most search terms. This is crucial to deliver improved user experiences, as empty space is a wasted opportunity to fulfil user’s intentLikewise, badding features like mood, locations, people, franchisesand popularityour search solution ensures that users always get the best chance to find what they are looking for, or more importantly, have a search function that allows them to explore, discover and unearth the content they want to watch.  

To find out more about search and content discovery, lookout for our next blog which will explore how The Filter are championing a change in how VOD services design their search page layouts to facilitate greater content discovery and remove ‘empty space’. 


About The Filter

The Filter offers personalisation of key customer touch points based on proven data science. We put the right content in front of a customer at the right time. More views mean more loyalty, lower churn, and more revenue. 

The Filter prides itself on working closely with clients both during set-up and roll-out to make sure that the client’s customers are always getting optimal recommendations. We are an agency not a SaaS provider. We have a genuinely deep TV understanding which means our data science solutions work exceptionally well for your viewers. We work with you to tailor our AI to your customer experience, your audiences, and your content. All integrated effortlessly into your data ecosystem allowing you to benefit from the power of personalisation easily and quickly, developing your platform at speed.  

For more information please email