A case study on Outlook refiners
Enhancing Outlook Search
Lead designer — 2023
1. CONTEXT
Millions of users perform searches in Outlook every single day, seeking emails, files, or messages. Any friction in the search experience directly impacts millions of users.
292 M
Monthly Active Users
Across all endpoints
171 M
Daily Searches
Across all endpoints
55 M
Monthly Active Users
OWA and Desktop
Outlook Search is a critical user habit, the #2 search experience in Microsoft, only behind Bing.
*Data is from December 2024
2. KEY CHALLENGE
The "Too many results" problem
Under pressure or due to low recall, users typically type a single keyword to initiate their search. They often don't remember the sender, date, or other details needed to formulate a precise query.

This short query leads to an overwhelming volume of results. When faced with this information overload, users tend to resort to the inefficient, tiring task of endlessly scrolling through the search results page.
3. GOALS
Reduce the time-to-find an email.
Support users' habit of typing short queries.
Decrease negative signals (e.g.: re-queries).
Visibility vs. Seamlessness
The successful integration of filters depended on striking a careful balance. If the filters were too subtle, they would be missed—failing to solve the problem of result overload. If they were too visually intrusive, they would interrupt the user's primary goal: scanning and clicking the search results.

Our challenge was to design a filter system that was obvious to the user, yet seamlessly integrated into the search workflow.
4. STRATEGY
Meeting Users Where They Are
To maximize the impact of the filters, we identified two key moments—or stages of search. Our strategy was to explore placements and behaviors that align with these distinct moments.
PRE-SEARCH / QUERY FORMULATION
POST-SEARCH / RESULTS PAGE
We hypothesize that providing filters directly within the search bar flyout will lead to more precise searches. By allowing users to apply filters before executing the search, we can guide them toward a successful outcome immediately.
We hypothesize that placing persistent and visible filters on the Search Results Page will act as an effective relief valve against information overload. This positioning allows users to immediately and easily narrow down a large set of results.
5. THE REFINER
Anatomy of a Filter
Once we determined where the refiners would live, our focus shifted to the fundamental building block: the refiner itself.
Designing the visual language for the refiner was one of the main steps of the process. It needed to clearly communicate its status (selected/unselected) and its function (what it filters) without disrupting the visual hierarchy of the Outlook interface.
To ensure the final component was both intuitive and space-efficient, we undertook an iterative process, conducting three rounds of design and user testing until we landed on a final, unambiguous design.
ITERATION 1
The filter transforms into a primary blue button with a dismiss 'x' icon upon selection.
ITERATION 2
-
Added a function-specific icon for quick recognition.

-
Usability testing showed confusion: users thought the 'x' icon would permanently delete the filter, not simply toggle it off. Also, having two icons consumed excessive horizontal space.
ITERATION 3
The final design simplified the interaction by displaying only a function icon in the rest state. When selected, the icon is replaced by a checkmark.
The checkmark clearly signifies "applied", avoiding the negative connotation of "deletion" from the 'x' icon.
6. THE EXPERIMENT
Implementation and Learnings
To accurately isolate the impact of our different refiner placements, we executed a phased rollout to internal users, utilizing two independent flights. This allowed us to specifically analyze the performance of the pre-search refiners separate from the post-search refiners.
PRE-SEARCH / QUERY FORMULATION
Why they were introduced
-
Refiners in QF were meant to help users narrow down results quickly.
-
They aimed to provide parity with SERP refiners and improve discoverability compared to Advanced Search or Ribbon filters.
What we learned
-
Users rarely interacted with refiners or related entry points. Experiments showed extremely low click-through rates (often <0.01).
-
Adding refiners increased pane length and visual clutter without delivering meaningful improvements in success rates.
POST-SEARCH / RESULTS PAGE
Why they were introduced
-
Mitigate Overload: The refiners act as an immediate "relief valve" for users overwhelmed by the huge volume results from their typical short queries.
-
Enable Immediate Action: The refiners allow users to immediately narrow down results preventing search abandonment (quick-backs and re-queries).
What we learned
-
Improvement in Search Success Rate (SSR): The SSR for users interacting with the refiners saw a +0.5% uplift, moving from ≈0.08% to ≈0.58%. While this percentage may appear modest, applying a half-percent improvement to a user base of over 100 million MAU translates to hundreds of thousands of additional successful search sessions every month.
-
Increased Engagement on the Search Results Page: The Post-Click Rate (PCR) saw an improvement from ≈47.27% to ≈50%. This slight shift indicates that the refiners effectively reduced the initial feeling of overwhelm, making the results page usable.
7. CONCLUSION
Strategic Outcome
We made the decision to keep the Search Results Page (post-search) refiners and deprecate the Query Formulation (pre-search) refiners based on a clear analysis of user value versus cost and complexity.
The QF refiners showed extremely low engagement. This indicated that users simply don't want to scope their search before they've even seen the results. Maintaining this feature would have resulted in unnecessary visual clutter without providing a meaningful improvement in success rates, representing a poor investment of engineering and design resources. The cost far outweighed the benefit.
The SERP refiners proved to be the necessary. The increase in Search Success Rate and Post-Click Rate confirmed that these filters directly solves the problem of information overload for millions of users.
By making this decision, we successfully channeled our efforts into the area of maximum impact. We upheld our commitment to the user's natural short-query habit while ensuring that the system provided an elegant and effective solution to the resulting complexity. The outcome is a demonstrable success: the refiners are actively enabling hundreds of thousands of successful search sessions every month, directly fulfilling our project goal to boost search efficiency across the massive Outlook user base.