Creating a Problem-Driven Social Shopping Experience

Taking Mira's Community from 0-50k MAU

MIRA BEAUTY is the world's first universal store and collaborative library for makeup and skincare. It's made up of over 100,000 products and counting-- that can be sorted and compared based on what you look like and what you're looking for.

At MIRA BEAUTY, I had the opportunity to grow our community from a small group of 12 beta users to 50k MAU. Our community was not just a place to chat about beauty, it was a place of learning and information exchange. Mira's community played a crucial part of our product and search ecology.

This is a high level walkthrough that strives to capture the major stages of growth within Mira’s community over the span of 2-3 years. In this brief we’ll cover:

To see an example of a project that takes a closer look at my design process, check out my Low-Intent Community Exploration design sprint here.


UI/UX Lead


September 2017 - May 2020


2 designers, 1 product manager


Design Research
User Interviewing
UI/UX Design
Information Architecture
Visual Design
Prototyping & Testing


I was employee #1 and first designer at MIRA BEAUTY. During my term, my role focused on the overall community building efforts. I led most design initiatives for the community and grew the product from a data-driven search tool to a holistic beauty community and platform.

Taking a step back: why we launched community

At first, Mira was a search engine for beauty products. We quickly realized community engagement is a necessary part of Mira because product hunting happens across multiple touchpoints and beauty research at its core, is a highly social activity. 

To better evaluate Mira’s community and product ecology, we created journey maps that follow beauty users' experience from product exploration to purchase and beyond.

Mira's Pre-Community Journey Map; primarily a search tool
Before we redesigned our community, Mira was mostly used as a search and evaluation tool. People would mostly research their products online, in-store or through friends and then use Mira to validate their findings. With Mira's community and search functionality, we wanted to encompass the entire user journey starting from exploration, all the way to product advocacy/expression.

V1: Groups

We first started by listening and learning about our community with two How Might We questions: 

We started with a  co-creation session with 16 participants.  In this session, we guided our participants through different sacrificial concepts that illustrated different online and offline community concepts (i.e. guided shopping, quizzes, AMA Q&A sessions, etc.) and topics of conversation. 

From our discussion, we generated five topic areas to organize the community around, which were Beauty Advice, Best of the Best(products), Beauty&identity, Looks, and Beauty Deals). In my first design iteration, I made each topic a "group" within the app where people could join to talk about the assigned topic.

High engagement posts from the Beauty Advice group often include photos, personal details, specific questions about products, brands, and specific issues they were dealign with i.e. eczema, dry skin, etc.

In our pilot, the most active conversation channel by far is Beauty Advice where people talked and asked about products, their skin, how products interact with skin. Of 200+ posts generated by our early users, over 70% of them were posted in Beauty Advice.

The most engaged posts were ones that were product focused and problem driven.

The most engaged posts were ones that were product focused and problem driven. This provided validation for our design research and we started to focus on making it super easy for people to share, find and answer questions.

V2: Building for high intent social situations with Questions for You (QFY)

In our next version, we focused on surfacing product focused and problem driven posts. We started by asking
Example feed
Our community started with a chronological feed. This format didn't foster much engagement as much of the content was repetitive, out of context and most importantly irrelevant to the user.
Early version of Questions for You
We began testing a number of features / optimizations to bolster engagement. One of the features we tested was Questions for You-- a module at the top of the page that surfaced random posts with a "pass" and "answer" button (gamifying posts). We wanted to know if people would be more inclined to answer questions when they were more visible and highlighted differently.
QFY feature data & stats

In reality, there was no special functionality or algorithm that surfaced personalized posts. These questions were just the most recent posts--nothing changed on the back-end-- but engagement skyrocketed!

We knew that people responded well to personalization and this quickly became a grounding mechanism in Mira's mobile app experience.

Product tagging microinteraction

At the same time, another seemingly insignificant change that let to big gains was product tagging in comments and posts. Initially we used icons like "@" above the keyboard to tag products, brands, or users. However when we eperimented with a large dedicated product tagging button, we saw a significant increase in product tagging. This not only helped the poster and commenter reference products more easily, it also changed the way clickers/lurkers interacted with product pages and posts. Through features like QFY we realized how powerful and consequential microinteractions can be.

A few insights

People in high intent mode who asked product focused, problem driven questions actively engaged people who were in low-intent mode.
There was a significant group of community members in high intent mode looking for answers and a similar cohort of people looking to give answers. This not only created a true value exchange but also crystallized our value proposition.

It should be blatantly clear on every page what we want the users to do. There should be no question as to what the most important jobs are on a page.

V3: Assimilating Community and Search content via personalization

After the success of QFY, Mira was able to help community members address each others' problems in high intent mode by prompting people to ask problem-driven, product-specific questions and creating affordances that allowed them to further explore Mira’s data catalogue.

Mira's Integrated Community Journey Map; a holistic beauty platform

Referring back to our user journey map, we achieved our goal for Mira to encompass the entire end-to-end user experience. However, the overall experience was still relatively disjointed; our powerful search functionality and community could have been two standalone experiences. We realized there were a number of opportunities to marry the two experiences (Community and Search).


IA and Navigation

Mira lacked the proper mechanics to provide seamless exploration between the product catalogue and community space. There were many “dead ends” in Mira’s flows. We called this "branch-like" exploration. There was no trace of community content within our product catalogues and barely any traces of product pages within our community space. Most posts funnel under “Beauty advice” making topics obsolete as a personalization tool (an important aspect of surfacing the most relevant questions and answers). An opportunity here would be to link all of our product content and UGC content in a way that felt more "web-like" - in a graph-like structure similar to that of Pinterest.


As we found from QFY about personalization, community members expressed how comfortable they felt posting and answering questions because they didn't have to be "credible" or "influential" to add value, they just had to relate to the question asker in some way through shared traits or experiences. This was a core community value that was foundational for question asking and answering (QFY) on Mira. There is an opportunity to connect people more intimately by bringing forward similar traits, interests or experiences by surfacing questions or conversations where they can actionably contribute.

Content Merging

In a big A-ha! Moment, we realized that people are essentially looking for the same thing whether they were query searching or asking questions to the community. Engaging with the community can be thought of as the "multiplayer" mode where you ask more open ended questions, receive follow up feedback, and collect personalized and anecdotal data. Query searching is then the "single player" mode where one might receive many more results and collect more quantitative data.

  • "multiplayer mode"
  • personalized experience
  • follow ups
  • "One player mode"
  • dozens of results
  • product focused, data forward

Either way, the user is looking for more context. What started as a design exploration led to production as we merged the two verticals (community + search) together by rolling out Topics-- thus tackling both opportunities in IA and personalization.

Adding Topic Nodes

In the past, we only had nodes in Mira which were essentially just brand or product pages. To organize both our catalogue and discussions, we created Topic Pages, which were a set of 30 topical tags that were used to organize and further enhance some of our existing taxonomy (i.e. brands and skin attributes). Topics ranged from skin type (oily, dry, etc.), skin concerns (large pores, eczema, rosacea, etc.) to subcategories(foundation, serums, etc) to attributes (drugstore, cruelty-free, etc).

Topical tags such as #combinationskin help us surface more relevant posts and contextualizes content to the reader

Topic tags allowed UGC content to self-organize in unstructured spaces. People are able to follow tags, find content that they are interested in, and create posts with topic tags so that it’s easily discovered by others who follow those tags. These tags became the backbone to our IA and our personalization features.

Clicking on a "Combination Skin" would lead them to this page. People could also further "combine" tags by continuing to click tags at the top of the page to refine their search-- the same mechanism used in our product search.

This allowed UGC content to surface in the right contexts no matter what you were searching whether it was for Maybelline foundation, eczema or date night makeup inspiration.

We continued to experiment with the flow of the page and the different types of content near the top of the page whether that was category search, a personalization quiz, or product carousels.

Along with topics being a primary method in cataloguing and repopulating content within Mira, each topic has its own page with its own content-- meaning you could explore Mira’s community AND search in a multitude of ways. This not only made Mira’s search and product browsing experience much more fluid and rich, it was also key to the success of Question for You.


Beyond the initial design, we took our learnings in the process and continued to experiment with new and different nodes, topics and content types on the mobile app. This is especially relevant as explorations on the discovery page were instrumental to user acquisition and retention. As our ecosystem grew, topics eventually laid the foundation to a lot of SEO content that drove traffic to our app and website.
Discovery tab experimentation with different content types


  • It's important to follow sound testing frameworks with some degree of consistency but you can't always test your way into something that works. It's worthwhile to have workflows that explore new product concepts early on, especially if you haven't reached PMF.
  • A/B test early and often. Move content around, even just to value test and gain context.
  • Follow your hunch :) 

A few fun stats

  • After we rolled out the new updates, 42% of 250+ users that we surveyed purchased a product recommended by a Mira community member, which is way more than we had hoped for!
  • 70% of users that had at least one product recommended to them returned to their post and clicked through to further evaluate the product.
  • The median number of comments on a question posted to the community was 7 and 45% of all comments included at least one tagged, clickable product and 65%+ of comments included a product recommendation.
    In other words, a typical user can expect to post a question on Mira and within a day or so receive over 7 comments and recommendations and at least 3 or 4 (if not more) personalized product recommendations with tagged products!
  • Additional Materials and Future Explorations