How to provide relevant hotel filters?

Design Lead
Research
Validation
2020
problem
Existing hotel filters were outdated, complex, and often irrelevant. New hotel filters were a part of a larger experimental project to use machine-learning to show most relevant travel content to a potential holidaymaker.
  • The long list with many options was overwhelming.
  • Power users risked over-filtering, leaving them with too few hotels to choose from.
  • Static content led to mutually exclusive options, eg. a user might have selected "adult-only hotels", but still see children-related filters.
My Role
  • I collaborated with a researcher on user interviews to learn how people really use (and would like to use) filters.
  • I analysed existing quantitative data to gauge specific filters' popularity.
  • I coordinated with machine-learning team to verify concept feasibility and get correct data
  • I created sketches, wireframes, and prototypes for desktop and mobile versions.
  • I supported our Visual Designer working on the final UI.
investigation

Combining insights from user interviews with website analytics, I arrived at the following findings:

  • Most people preferred to use 3-4 filters, and expected the filter list to adjust to their preferences over time.
  • We had a chicken and egg problem - we weren't sure if the top filters are the most used ones, because they are most relevant, or if they are most used, because they're placed on top of the list
  • For filters with significant usage, there was little difference between travel groups (couples, families, single travellers). However, it was possible that they could not find certain relevant filters (eg. babysitter service for families) in the extensive list.
Basing on those findings, I decided to:
  • limit the number of visible options to what was most relevant to all travel groups according to quantitative data.
  • add a search feature, so users can look for additional filters. This would give us less biased usage data, and allow to display a list of filters tailored to a specific travel group in the future, and eventually to a specific user.

Final concept

Work on this feature ended at an advanced prototype stage. It was never implemented as the entire experimental project was closed. In 2025, it still disappoints me to see the old filter list on the HolidayCheck website.

Comparison of old and new filters design