Boxmate - exercise tracking for CrossFit

Improved exercise discovery of rep variations through heuristic-led review

PROBLEM FRAMING

Introducing Boxmate

Boxmate is a CrossFit exercise app for booking classes, logging and tracking exercises and connecting with your local community. This project focused on improving navigation and discoverability around exercise pages and rep-specific logging, based on real frustrations observed during day-to-day training.

This is a concept redesign focused on UI heuristics and competitor analysis. Further work could include usability testing, additional score types and edge cases.

Excerpt: Snapshots from the current Boxmate app, showing exercise list and landing page

The problem isn’t logging - it’s getting to the right exercise landing page

INFORMATION ARCHITECTURE

I focused first on the navigation and information architecture around exercise pages - specifically how users discover, access and log scores for rep variations.

Through day-to-day training, I noticed users (including myself) often struggled to find rep variations (e.g. 2 rep max) despite coaches assuming this “easy”. This mismatch highlighted a gap between they system’s structure and users’ mental models.

I mapped a local IA to explore how users currently land on exercise pages, and where rep-specific logging becomes hidden or fragmented across the journey.

Excerpt: Local information architecture showing navigation to exercise pages and rep-specific score states

USER FLOW

Interaction analysis revealed breakdowns in discoverability and hierarchy

Reviewing the user journey for logging a 2RM score showed that key actions and rep variations are not constantly discoverable. Users rely on expanding lists, secondary navigation, or indirect routes. This increases cognitive load and makes it harder to predict where scores live.

Screen 1 - Exercise List

Various icons and 3 tap targets across each exercise make affordance difficult and cognitively overwhelming.

Screen 2 - Expanded

Rep variations are revealed through the hamburger icon, relying on discoverability over visibility. Video interferes with primary action.

Screen 2 - 2RM landing page

Score data is clearly presented once reached, but with assumed interaction across Best / Latest scores.

Screen 2 - Related exercises

Rep variations are also grouped under “Related” which duplicates and doesn't match user expectations for logging.

Through further screen analysis, I discovered some other broader themes that could impact discoverability and decision making:

Inconsistent affordance and tap targets - several interactive elements lack clear affordances. Small touch targets and a reliance on a hamburger icon to reveal variations make it unclear what is interactive. As a result, users often rely on trial-and-error rather than confident action.

Language, labels and mental models mismatches - labels such as “related” (for rep variations) and “percentages” (for breakdowns) may not align with how users think about exercise data. This weak alignment between language and mental models make it harder to anticipate outcomes

INTERACTION ANALYSIS

Competitor analysis: patterns in discoverability and information hierarchy

Theme 1: Discoverability & landing affordances

Excerpt: Exercise list data hierarchy and affordance variations

Strong affordances at entry points reduce hesitation and unnecessary navigation. Clear chevrons (or the absence of icons entirely) signal a simple, predictable action, reducing perceived complexity. simplest navigation.

In Boxmate, mixed affordance across the exercise list (icons, menu, expandable rows) create uncertainty around where rep variations live and tap targets.

Theme 2: Information hierarchy on the landing page

Excerpt: Information hierarchy on exercise landing pages

Strong and Heavy use tabbed navigation on exercise pages to separate education, history and performance. This reduces cognitive load, allowing users to intentionally switch modes depending on their goal.

Boxmate, and Dreamwod, surfaces all information on a single page. This reflects retrospective logging model, where users arrive with a specific intent - logging and reviewing scores rather than education.

However, without stronger hierarchy, this density makes it harder to identify primary actions and relevant rep variations.

Theme 3: Rep variations and mental models

Excerpt: Rep variations as states over screens

Competitor app Dreamwod groups rep variations within a single exercise context to make switching between rep variations explicit and reversible. This reinforces the mental model that rep variations are attributes of the same exercise, not separate destinations.

I analysed how comparable products support discovery, decision-making and logging, focusing on affordances, hierarchy and how rep or content variations are surfaces. I explored two general workout apps (Strong, Hevy) and a CrossFit app (Dreamwod).

DESIGN DIRECTION

Prioritising clarity, accessibility and primary actions

Early ideation included low-fidelity sketches to explore navigation, rep variation placement and data hierarchy before moving into structured flows and high-fidelity design.

The redesign focuses on making rep variations visible within the exercise context, reducing navigation depth when logging scores and clarifying primary actions after a workout. Changes prioritise accessibility (slightly amending brand colours for WCAG AA), clear affordances and a more predictable mental model for finding and switching between reps.

I also prioritised ‘Log a new score’ a primary CTA to reinforce motivation through visibility and challenging the passive, retrospective logging aspect.

Excerpts: Final UI showed improved exercise discovery, clearer rep variation switching, and a simplified post-workout logging flow 

REFLECTION

Usability testing and iterations

This concept focused on improving discoverability and clarity around rep-specific logging using UI heuristics and competitor analysis. With more time, the next steps would focus on validation and edge cases.

Specifically:

  • Usability testing with CrossFit athletes to validate the revised rep navigation and logging flow

  • Testing mental models around rep variation to confirm users expect to switch reps within an exercise page rather than via separate pages

  • Refining edge cases such as failed lifts, partial reps, or logging scores during a workout vs. after

  • Exploring progression views, including trends across rep ranges (e.g. how 2RM relates to 1RM over time)

  • Accessibility testing, including contrast, touch targets and one-handed use post-workout.

This would help confirm whether the redesigned hierarchy reduces navigation friction and supports faster, more convenient score logging in real training scenarios.

Back to Home