︎
Home    About    Resume



Notion

Providing structure and relevancy to help users find content quickly



DURATION
4 months
(Feb-May’22)
ROLE

Product Designer
RESPONSIBILITY

UX Research, Strategy, End-to-end Design





CONTEXT

Notion is a B2B and B2C productivity application used for note-taking, wikis, and project management


Notion aims to maximize your productivity so that you can complete your tasks most efficiently.

As a passion project, I redesigned Notion’s search tool with the goal of making content retrieval relevant and intuitive.




PROBLEM

Users struggle to use the Search tool to find relevant content quickly 


Notion users are teams or individuals who use productivity apps to store, organize, and retrieve information needed to complete their tasks.

Workplace Notion users who do not publish content are unable to use manual navigation to retrieve information, because they don’t know where the content they need lives. So, they resort to using Notion’s search engine.

However, when using Search, users are confused by how to interpret their search results and feel that their search results are irrelevant. Results don’t clearly show the 3 aspects of relevancy: recency, frequency, and similarity. This blocks users from getting the information they need and they cannot continue with the rest of their workflows. This is a retention issue that causes Notion to lose long-term users.


What is it like to use Search today?




FINAL SOLUTION

Increased relevancy and structure 





After iterating, users quickly understood the hierarchy in search results and were able to feel fully supported by the new search logic concepts.

Simple visual changes helped users see the tools at their disposal, and therefore felt more confident about using Search in the way that works best for them (which aligns with Notions’ goal of creating a customizable workspace tailored to each individual’s needs).






See how I arrived at this solution ⤵




RESEARCH

Competitive analysis: How might we make Search feel like it can read your mind?


It’s no secret that search engines feel like black boxes - you don’t fully understand how they work but you can guess what inputs they need to spit out the information you want.

Notion’s Search is no different.



The retention issue is a symptom of users not feeling supported by Search’s complicated search logic and functionalities. Before I jumped into ideation, I faced the challenge of cracking the code to search engines.

With a quick Google search, I found that the search logic behind each engine is determined by the context in which the engine is being used. For example, Google Search sorts results by categories based on user relevancy. If a user looks up “pizza”, Google shows pizza restaurants before pizza recipes because restaurants are the most relevant to users for food-related searches.

To determine relevancy for productivity app users, I conducted competitive analyses of productivity and content storage apps like Coda, Asana, and Google Drive.



Search logic testing + identifying frequently appearing features


I discovered that relevancy was determined by: recency, frequency, and similarity - in no strict hierarchy as each individual retrieves information differently.

Since people have different mental models with retrieving information, I ideated ways to create structure in the search results and give users tools to help them determine relevancy themselves.



User interviews


I kicked off research by reading through software review forums for general customer feedback. To gain further context, I interviewed 5 Notion users - 4 workplace users and 1 personal workspace user.

Note: I originally planned to investigate why Notion felt unintuitive and cumbersome to use in general, but through generative research, I noticed that users were consistently blocked in their workflows when retrieving content, so I rescoped my research plan to focus on Search.

With affinity mapping, I analyzed the trends from the interviews and identified the top 3 pain points.

Note: Search was named “Quick Find” at the time of my research.



Pain Points


People struggle to retrieve content in Notion due to:

  1. Lack of relevant search results. Users wasted time testing out different keywords to get relevant keyword matches.
  2. Lack of organization/information architecture. Users found the list of search results overwhelming.
  3. Lack of context. Users found it hard to parse through search results.




DESIGN GOAL

Make content retrieval via Search relevant and intuitive


To achieve this vision, I broke it out into supporting design goals:

  1. Provide relevant search results - that are frequently and recently used
  2. Organize search results in a way that users understand
  3. Reduce the friction that users experience when sifting through search results

User impact : Increase ease and efficiency

Instead of wasting time being blocked in their workflows, users can spend more time crossing off their to-do’s and moving workflows along.

Business impact : Increased product satisifaction!

Users noted that as their small startups grew, the larger & more complex the data in their company Notions became. These startups eventually left Notion for apps that were more robust for productivity at a large scale.

Creating a relevant, intuitive and optimized Search means faster content retrieval - which is key for productivity. Less time wasted on content retrieval should increase user satisfaction of Notion as a product. Higher user engagement decreases the likelihood of churn.




IDEATION

Exploring ways to provide context and strengthen user’s understanding of Search


I moved forward with concepts around visual standardization Nielsen’s heuristic of flexibility and efficiency of use.

Invisible design explorations: Most frequented sorting + increased search engine flexibility

To provide relevant content, I:

  • Created a new Sorting prioritization based on most frequented content and
  • Increased flexibility to the search engine logic by adding keyword flexibility and
  • Added recently renamed/deleted content to the search results
    • Note: Notion released big Search updates during this case study, and one of the features was including recently deleted content in Search! 😊




Visual explorations: Last Edited information, standardized iconography, and creating visual structure in search results

Initially, I explored using a preview panel to give users more context, but through feedback from other designers, I decided to scope down my ideations to the following:

  • Displaying Last edited information. This information is currently hidden under the Sort tool and timestamps are not displayed after sorting.
  • Referencing the most commonly used iconography, I created document type icons to replace the blank/filled out Page icons assigned to all types of search results.
    • I also played around with a feature that would limit the visibility of icons set on documents, so that custom icons could only be seen by the user that set them. This was to eliminate the confusion users had when they saw random icons set by others in their search results. This new feature would be communicated via a tooltip.





Scoping

I explored a breadth of search result layouts using varying degrees of structure based on categorization of document types. After receiving feedback from other designers, I decided to pass on pre-categorizing results because it added visual bulk to an otherwise minimalistic search modal and didn’t help users more than the other redesigns.


The combined concept for an intuitive Search experience


The top-voted wireframes were those that:
  • Anticipated and supported work users’ most frequent search scenarios
  • Provided a balanced amount of context without overwhelming the user
  • Matched the mental models of other popular search engines





VALIDATION

Testing the relevancy and intuitiveness of Search


I conducted usability testing on 5 Notion users - 3 workplace users and 2 personal users.

First, I tested the intuitiveness of my new search logic concepts and standardized iconography in crafted search scenarios, then evaluated the helpfulness of them. Afterwards, I tested the limited-view iconography tooltip and evaluated user comprehension.

🏆 Wins

  • 5/5 users were delighted to see Recently Renamed content appear in their search results.
  • 3/3 users expected keyword matching flexibility and felt supported to see the functionality.
  • 4/5 users felt more confident in their ability to find relevant content quickly using the redesigned Search.

🛠 Areas of improvement

  • 4/5 users did not notice the Document Type filter.
  • 4/5 users were not sure they’d use the limited-view iconography feature.
  • 3/5 users did not notice the Last Edited information.




ITERATION

Although users enjoyed the new search logic concepts, they were confused by the information hierarchy of the search results. Users also didn’t notice existing and redesigned Search features that could’ve helped them sift through search results better.

In my iterations, I focused on:

  • Creating a clear hierarchy based on keyword matches in Titles + frequency/recency of each document.
  • Making redesigned features more visually distinct.



To prioritize the most impactful redesigns, I tabled the following explorations:

  • Limited iconography views. Further research is needed to measure the number of users that actually use icons in the workplace - to determine if it’s worth pursuing this exploration. Additionally, limited views may not be needed if document type iconography successfully provides users enough context to determine relevancy.
  • The presentation of search results filtered by Document Type. The document type filter was proven helpful to users, but 2/5 users said it’s not visually clear that the filtered search results shows the document types chosen. More testing is needed to determine if this a problem a majority of users would face. However, it is rare to see search results with only custom icons attached, so this would be an edge case that can be investigated later.




MVP: PLAN OF ACTION

For the MVP, I would prioritize (in this order):

  1. Sorting search results by most frequented. Although it is a high engineering lift, frequently used documents is the #1 thing users look for and is the most relevant.
  2. Adding Last Edited information. This is a medium engineering lift as last edited information is already recorded. Users use recency to determine relevancy (as reflected in competitive apps).
  3. Creating a clear search result hierarchy based on keyword matches in Titles and frequented documents. Information architecture is the first thing users look at to determine relevancy.

Nice to haves:

  1. Adding Recently Renamed content and increasing keyword flexibility. Although these require an engineering lift, these 2 features were the most delightful to users and made them feel supported, which would encourage people to continue using Notion over other productivity apps.
  2. Standardizing iconography by document type. Giving users visual context reduces the cognitive load of sifting through search results, especially when the current default iconography does not accurately reflect the document type of each search result.

How would I measure success

Click through rates upon opening Search to landing on a document page. Because it is difficult to correlate retention rates to the usage of a specific feature, we can gauge success by measuring time saved through improving Search. Comparing a heat map of users’ interactions against the number of clicks it took a user to stay on a landing page, can help us determine the amount of time saved.




VISION OF THE FUTURE

A search engine that can read your mind 🧠


Creating a delightful Search experience, that feels anticipatory to a user’s specific needs, is not only aligned with Notion’s “fully customizable” brand, but would also make Notion standout in the B2B marketplace.



From my research, users perceive productivity apps as extensions of their brains for storing and accessing information. Search engines are still blackboxes that users don’t comprehend, which makes users feel disconnected from their “extended brains”. To have a tailored and intuitive Search experience would make users feel more confident in their ability to use their “extended brains”, and therefore themselves, to get work done.

Empowered users means better brand perception as a go-to SaaS tool and would help Notion capture more of the B2B marketshare in the productivity space.




SELF REFLECTIONS 📓

  • Let the research and design process guide you. I came into this case study with personal bias because I’ve used Notion for years, and would jump straight to solution-ing. I learned that in order to create a truly impactful solution, I have to lean into the research and feedback from other designers to guide me to clearer user/business impacts, design goals, and iterated solutions.
  • During Ideation, I was doubtful that this would be a good product design case study because Search issues felt like an engineering problem, not a design one. After chatting with my mentor, Jennifer Wong, I realized that Search is actually where UX shines the most because it is a complex function on the backend that needs to be easy to use on the frontend. My takeaway is: just because my most impactful explorations are not visible designs, doesn’t mean they won’t make an impact.

  ︎  ︎  ︎