Personas in
Adidas Running
Seeing runners as people, not just numbers.
Main goals
Run the first-ever foundational user research on the Adidas Running app to better understand existing and potential users.
Create and launch personas, facilitating autonomous cross-functional teams to make decisions and create experiences based on a shared understanding of users.
Main contributions
Led end-to-end research project, handling budget, tooling, user recruitment, research planning, execution, and insights sharing.
Collaborated with leaders across Adidas to unite research efforts.
Paved the way for hiring a research team and establishing stronger research practices and outcomes.
The Adidas Running app helps you track your runs and other workouts with GPS, set goals, and stay motivated through training plans and a global community.
In 2019, the Adidas Running app team had many questions about who were our existing and potential users. While we did the occasional usability test or conducted interviews to evaluate and learn how to improve features, we lacked generative, exploratory research as a foundation to our strategic decision making.
Despite the limitations that any tool provides, personas stood out as an effective tool for us as they were flexible enough to reflect real user diversity, and concrete enough to support conversations across different teams.
To hit the ground running I negotiated with management the support of different teams (CRM, Data Science, Legal…) and secured budget for tooling and incentives. I found allies that helped push this effort on multiple fronts, which later led to the creation of an OKR for the entire fitness app area dedicated to Personas and later expanded the research to the Adidas Training app. To ensure progress that would prove the value of the project and land us bigger investment, we had to stay scrappy and work iteratively.
Key result
Create and validate 3–5 user personas representing at least 80% of the Adidas Running app’s active user base, and onboard all cross-functional teams to use them in decision-making.
My research approach was informed by a knowledge exchange session we organized with a designer at Spotify, shortly after they released their personas case study, and was also inspired by the Year in Sport Trend Report by Strava, one of our main competitors.
Drawing from both, I followed a mixed-methods approach to ensure the personas were grounded in real usage patterns while also capturing user's underlying motivations and context.
I distributed an internal survey to understand what different teams wanted to learn about users and how they would be used in their daily work. In a kick-off meeting with stakeholders I then shared the common priorities identified and explained what could and could not be covered by this kind of research and artifact, getting everyone on the same page and securing support. We agreed to tackle the following themes:
Demographics
Basic user information like age, gender, and location to understand who our users are.
App behavior
How users interact with different apps, frequency of use, and usage patterns.
Fitness behavior
Users’ workout habits, sport types, goals, and frequency of physical activity.
Social behavior
How users connect and interact with others digitally and in-person in the context of fitness activities.
Content behavior
What kind of content users engage with, such as articles, tips, or training plans.
Psychographics
Deeper insights into users’ motivations, values, lifestyle, and attitudes toward fitness and technology.
To ensure useful and reliable insights, we went through the following process:
Set user participation criteria: filtered for active users with consistent behavior.
Defined features to be included: defined how to measure aspects we were interested in, such as what metrics qualified someone as “socially active”.
Ran hierarchical algorithm: identified the optimal number of segments.
Applied K-Means algorithm: assigned each user to a segment, making the segmentation scalable and repeatable.
Ran variance analysis and feature importance scores: identified which behaviors and attributes drove the segmentation.
This resulted in visualizations such as:
In total, 8,117,170 users were included, and we ran the segmentation separately for active, dormant, and newly registered users, to avoid the segmentation simply reflecting these different engagement states.
For each segment, we were able to get insights into almost 30 different measurements, such as gender and age distribution, avg. number of activities/week, max/median running distances, top sport types tracked, % with connected wearables, % with friends, % who joined a challenge, % who visited blog posts, top blog article categories, and more.
This started to give shape to the types of users that would be further researched and become our personas. Below is a simplified sample results:
Our goal with the interviews was to uncover user motivations and explore the context behind how different types of users engaged with running within and outside the app.
Planning and running the interviews required intensive cross-team collaboration, and led to a reusable recruitment process that was then used by several teams in future researches. Here’s it was done:
Participation incentives: secured budget with the Finance team.
GDPR compliance: worked with Data and Security teams to track interview invitations across user segments while ensuring user data stayed private.
Interview invitations: planned with CRM the newsletter send-out.
Screening survey: selected 30 diverse participants from around the world.
Interview script: designed a guide to explore users’ lives beyond the app, their fitness habits, and the app’s role in that journey.
Interview & note-taking: moderated remote sessions via Google Meet with cross-functional colleagues as notetakers, building empathy and enriching the analysis.
Analysis & synthesis: used Airtable for affinity mapping, clustering quotes, behaviors, and observations by theme to identify patterns.
Some of the interviewees in our video calls
Interview findings enriched the segmentation and guided future initiatives. Here are some key takeaways:
Challenging assumptions from segmentation
Interviews exposed gaps in data interpretation. A user labeled as socially active due to high Newsfeed usage in reality was using it in unexpected ways to check their own activities. Another user labeled as highly retained had actually moved on to other apps, however smartwatch data that kept being synced and app openings triggered by engaging with our newsletter articles led to a different assumption.
Understanding behavior beyond the app
The conversations uncovered rich offline habits. Some users from “not social” segments, were actually training with friends or active in communities on platforms like Facebook or Meetup, context that wasn’t captured by app data but had a big impact on their fitness journeys.
Connecting with users on a human level
Many shared emotional and personal stories behind their fitness goals, such as running for charity, for postpartum recovery, or managing chronic health issues. These narratives gave depth to each persona and helped teams connect more meaningfully with users, as well as feel proud of the work we were doing and impact we were having on people's lives.
Running a user survey could help us gather broad, quantifiable insights about user behaviors, needs, and motivations to validate and enrich persona development, making the bridge between the segmentation and the interviews. This was our first time running such a large-scale survey, and to ensure we collected significant and diverse responses we followed the process below:
Raffle planning: partnered with Finance and Legal to raffle an iPhone and Amazon vouchers as large-scale participation incentives.
Survey guide: built a 19-question SurveyHero form with multiple-choice and conditional logic for easier, scalable analysis.
Survey translation: worked with Localization to translate it into our 8 core languages, to reach a wider, more diverse audience.
Launch & tracking: organized with CRM the newsletter send-out and tracking to be able to link responses back to user segments for richer insights.
Responses collection: despite concerns about length, over 20,000 users responded within the first days, far exceeding expectations and validating findings.
Analysis & synthesis: Cleaned and standardized data, ran cross-tabulations to explore correlations between self-reported motivations and previously identified behavior segments, and checked response rates by segment, gender, and country.
Sample of survey results
Here are some of the key takeaways:
Validated our segments with user-reported motivations
The survey confirmed many of the behavioral patterns we saw in the segmentation and helped us connect them to users' motivations we learned from interviews. This triangulation of qual, quant, and behavioral data gave us strong confidence that the personas we were creating were grounded in reality and usable across teams.
Put numbers behind the stories
While interviews were rich in depth, some stakeholders unfamiliar with qual research were skeptical about the sample size. The survey helped bridge that gap by putting numbers behind the narratives and surfacing the same themes across a much larger population. This made the personas easier to socialize and champion across the org.
Uncovered blind spots
While much of the data aligned with earlier findings, the survey also surfaced patterns we hadn’t fully captured through segmentation or interviews, such as a sizable group of users highly motivated by stress relief and mental well-being, which hadn’t been as prominent in our earlier discussions. These findings helped us fine-tune our personas before moving into asset creation.
To secure buy-in and build credibility and shared ownership, I brought key stakeholders from product, marketing and engineering into the room for a collaborative workshop. We started by reviewing and discussing findings from the segmentation, interviews, and survey. From there, we organized the most relevant insights into an initial persona template, using prioritization exercises and open discussion to align on what mattered most. To start making the personas feel more human, we brainstormed names and visual elements that gave each user type a relatable identity.
Workshop to review insights, outline personas, and brainstorm template
For the layout, we followed a similar style to personas from shopping teams at Adidas. We used the RISE framework, which covers Rational, Identity, Social, and Emotional aspects. We then produced both longer and shorter formats of the personas, one for deep learning and onboarding, and the another for quick referencing. We then printed persona posters and placed them around the office to keep users visible in the day-to-day and build empathy through exposure.
We ended up with 3 Adidas Running personas that encapsulated the main characteristics identified through the research. Here is a summarized version:
Short version of the 3 Adidas Running persona posters
To make sure the research made its way into day-to-day decisions, I partnered with the Principal UX Designer in the Adidas Training app (who was then also finalizing a first version of personas) to make the personas visible and facilitate their adoption through the following actions:
Presentation to fitness apps area
I shared the personas, explained findings, and exemplified how they could be incorporated into daily worked. This marked the delivery of the OKR and helped raise awareness and engagement.
Q&A & research Slack channel
Bookable weekly sessions were offered to provide support to whoever needed. A Slack channel was created for updates and questions. This open dialogue fostered ongoing collaboration.
Team trial-run workshops
In these sessions, teams could practice using personas for story writing, feature ideation, and test scenario creation. This drove hands-on adoption and surfaced valuable feedback.
What began as a simple question of “Who are we really designing for?”, grew into a large-scale, cross-functional effort that connected data with human stories, challenged long-held assumptions, and opened doors for collaboration across adidas.
Our Product Manager Chris, holding the Chris persona from the Training app
The entire research included:
Here are some of the ways this project impacted the organization:
Clarity and empathy for runners
By combining quant and qual insights, we painted a nuanced picture of our user base. Bringing colleagues into interviews built empathy and ownership, and user quotes from those sessions were being brought up in meetings long after the project ended.
Push toward strategic decisions
Employee engagement surveys showed a need for clearer direction. This project helped us answer who we’re serving and who we could serve next, while highlighting the tough choices we have to make with limited resources.
Improved data quality & awareness
While tracking issues weren’t new, this project exposed just how much we were missing, sparking more accountability and urgency to fix gaps. It also gave teams the confidence to push back on rushed releases without proper tracking.
Foundation for better research practices
Fostering a culture of curiosity turned answers into new questions, making a strong case for hiring our first dedicated UX researcher. The impact and delivery of 100% of the OKR boosted confidence in the practice and helped secure budget for a diary study to deepen insights.
Cross-functional collaboration & culture change
The project brought teams together across product, marketing, and engineering to co-create and adopt personas. They were actively used in story writing, test scenarios, marketing briefs, and other daily workflows, informing better decisions.
Creating personas is one thing. Setting up a research practice, gaining stakeholder buy-in, bridging team silos, and driving real change was something else entirely. Here are some of the ways I grew throughout this project:
Seing things from different teams' perspectives
Collaboration wasn’t always easy, but working closely with teams like CRM and Data Science taught me to speak their language and understand their constraints, leading to better briefs, smoother timelines, and more mutual wins. This also gave them better visibility into what UX could bring to the table.
Getting comfortable with quantitative research
Before this project, I was primarily focused on qualitative methods. By learning how to set up and run a user segmentation, create and analyze a large-scale survey, and to triangulate insights across data sources, I became a more versatile researcher.
Taking ownership of research ops & full research lifecycle
I led the entire research process: from shaping questions to delivering strategic insights. I handled everything from budgeting, tooling, GDPR compliance, incentives, and built repeatable workflows. Being a UX research team of one gave me insight into the important work of setting up and maintaning a research practice.
Through this process, I not only created personas to guide the team forward but deepened my own ability to listen, synthesize, and to continue designing with people at the core.
