Swinging Away with GolF AI

Roles & Responsibilities

UX research (discovery + validation)

Participant recruitment


UI design (iOS)



Pen and paper


Zoom (remote usability testing)


Timeline: 14 weeks

Timeline: 14 weeks

Discovery user research (3 weeks)

Expert review (2 weeks)

Paper prototyping and testing (2 weeks)

Hi-fidelity designing (3 weeks)

Remote usability testing (2 weeks)

Client presentation (2 weeks)

Setting the stage

Last October, I was fortunate enough to have the opportunity to join Golf AI, a startup using machine learning and computer visioning technology to provide golfers with instant swing analysis. I had worked with the rest of the design team on creating an experience that allowed golfers to record their golf swings and get feedback.

Users would be able to see the angles of their different body movements (ex. hip rotation or shoulder turn) at the three main parts of their golf swing.  As designers, we had really focused on taking that feedback and making it more digestible and intuitive in our preliminary design efforts.

The existing design we had created.

However a low number of app users, and correspondingly, low app sales had caused us to take a step back and ask ourselves if the feedback that we were giving to users was actually valuable in the first place. This case study is the in-depth story of how I led our product thinking and design efforts to look past our app as just a swing analyzer, and instead create a personalized coaching experience.

The Need for product positioning

Part of the problem with the design process we had used was that we designed around a “golfer”, but we hadn’t considered what kind of golfer would actually use our app, and why. 

When looking at competitors, we found that there wasn’t as much focused on trying to help golfers, especially newer ones, actually improve their stroke without having to pay for a coach. Since this seemed like a promising audience to target, we now came up with a new mission of creating a cheaper alternative to coaching within a golf app.

Understanding how to hit our target audience better

We needed to understand the characteristics of a wide variety of golfers, and especially the less experienced ones who wouldn’t necessarily want to pay for coaching. This way, we could be much more intentional in how to create an experience tailored to a more specific audience


To do this, I worked with our UX researcher, Darin, to lead the creation of a survey with the two other designers. We were able to get feedback from 89 golfers and 16 former golfers, with responses from golfers of all ages, income levels, experience levels. Some of the key insights that stood out:

Problems with the current app

Looking at the persona, we could then see how the current app experience didn’t give value to our target user. The experience hadn’t been built around helping the golfer improve one area of their swing at a time, but rather was more focused on providing a holistic breakdown of the golfer’s swing.

Although we had a toggle button that allowed golfers to switch between that showed their 3 “areas of improvement” and 3 “best areas”, these areas were just clumped together with all the other parts of swing.

The top 3 areas to improve were listed. but weren't given any special attention to in terms of how to improve them (compared to the other areas).

We were trying to build value in the holistic data we could provide about all the angles the golfer had for all the different parts of their swing, compared to the angles that pros had. However, these angles were simply displayed as metrics without any explanation of what the angle was referring to, which made them confusing to our target user.

The tips were also very limited, with just one generic tip for each of the different body movements. The drills that were recommended didn’t have any additional context or explanation as to how they would help address the problem within the golf swing. 

When clicking on a card for a certain body movement, the golfer would see a tip and recommended drills to help target that area.

Based off our research as well as analysis of the current app, we came up with the following problem statement to design around:

How might we combat the high costs of golf coaching by creating an experience that guides golfers to improve one area of their swing at time?

In doing this, we wanted to remove the “stagnancy” of just a technical swing analyzer to make feedback more actionable, dynamic, and tailored to the user. From a business perspective, we wanted users to spend more time interacting with their feedback to build up continuous usage in the app.

Design Principles and Branding

With a focus around these golfers who had felt stuck in their path of improvement, we came up with the following design principles to be more intentional about the “feel” we wanted to create in the our app:

  1. Welcoming 

  2. Supportive

  3. Personalized

  4. Simple and intuitive

From Idea to Design

When brainstorming a new feedback experience, our main goal was to create a more immersive step-by-step experience to guide golfers to work on one thing at a time. In doing so, we wanted to avoid the choice paralysis present in the current experience, as well as make the help we provided much more tangible.

We converged onto this high-level user flow we wanted to create our experience around:

In this flow, users will now spend most of their time targeting one specific area of their swing, before deciding to analyze another swing or work on another area.

In building this experience from the ground up, we explored a variety of different concepts and went through 4 different iterations. We got feedback on each iteration through remote usability testing with 9 golfers to validate our assumptions and ensure that our design was best fitting the needs of users, until we came up with a final design we were confident in handing off to our developers. 

Our explorations focused on:

  • How much choice we should give users in seeing multiple areas of their golf swing

  • Creating better interaction with recommended tips and drills

  • Different ways to encourage users to record another swing

  • Balancing the use of copy vs visuals (ex. using different shades of color) vs components (call-to-action buttons) to guide user behavior

The evolution of our designs, from sketches wireframes to final design.

After each user test, we organized the data in an affinity diagram. The main takeaways from our research:

Although golfers want to work on thing at a time in their golf swing, they still want to have the flexibility of choice to be able to see more than just one area.

Golfers like seeing videos of themselves and of pros to understand what they are doing wrong and right, and to visualize how can improve. 

When golfers take another swing through the app, they want to be able to immediately compare their recording to the previous one.  

What Part of my Golf Game should I work On?

In order for golfers to spend time to working on one thing in detail, the first part of the new experience would needed to focus on guiding users to choose that one thing to work on. A lot of our iterations and efforts creating this guided structure.

With a goal of avoiding choice paralysis, our first iterations defaulted to only showing the area of the swing that needed the most work, with users having the option to go the "Action Plan" where they could see two other areas of their swing to work on. However, we quickly learned that users liked being able to see more than just one area of their swing, even if they were most likely to only work on the one that needed the most work.

So, we decided to build our experience around the Action Plan. It became the centralized hub that provided a high-level coaching breakdown, focused on leading the user to work on one thing at a time. 

After the user has recorded their golf swing, they will be taken to a landing page that highlights their Swing Score. We added a clear explanation of the Swing Score concept since it wasn't cleary understood by some users. The "View Action Plan" call-to-action button takes users to their action plan.

Once in the action plan, users are presented with the list of the 3 main priorities they need to target to improve their golf swing.

We used a card layout in which we used our orange accent color to highlight the priorities, and the darker orange to highlight which area was the highest priority to work on. This ended up being the most visually cohesive of all the other visual design explorations we tried, and this association was also clear to users.

If users scroll down in their action plan, they can see their best area. Users found it valuable to show which area of their swing was good, but that it was something that they’d want to see after already seeing their areas to work on.

Users liked being able to see a GIF of their best area compared to that of the pro, so that they could see why it was good

"I like seeing my best area, so I'll try know to mess that up" - quote from a participant in user testing.

The action plan would also play a role in a key business decision - building impulse to get users to start their free trial. Our app had a model where users would get a 2 week free trial before they had to start their $8.99/month subscription, and we felt it would make sense to show the free trial pay wall after the user sees their action plan.

Since users felt excitement about their action plan and were willing to click the first area to see how they could target it, popping up the premium model at this “peak of impulse” could be key to close the deal and get people to commit to the trial.

Personalized tips and drills

Working with Content Creators

Based on our survey insights, the things golfers looked for most in an ideal plan were tips, drills, and visualizations specific to them and it was important they learn what to “feel” in order to fix their body positions in non-technical terms. Content was king, but in our current app the content had not been prioritized.

As a designer who didn’t golf, it was key to build this empathy with golfers to understand what they really needed. We then worked with Jacob and Vedanth, experienced golfers and coaches in our "golf" team. We pass on our research insights to create a strategy on what kinds of videos and copy would be needed, and in what situations that content should be used. This content was put together in a JSON file, which was passed on to the development team later.

Using a Tab Menu to navigate through Content

Along with these personalized tips and drills, Golf AI’s technology offered a side-by-side video comparison highlighting a golfer’s swing to the ideal pro. Golfers valued this comparison because it allowed them to visualize what they were doing differently from the pro.

With these 3 “layers” of content, we decided to use a tab menu to give the user flexibility in how they navigated between the overview (a breakdown of the swing problem and the video comparison), tips, and drills.

After the user has recorded their golf swing, they will be taken to a landing page that highlights their Swing Score. We added a clear explanation of the Swing Score concept since it wasn't cleary understood by some users. The "View Action Plan" call-to-action button takes users to their action plan.

Creating Engaging Tips

When researching examples of golf instruction online, we realized that a lot of the help videos focused on showing golfers what they might be doing wrong, and then comparing that against the correct way to do things and address those errors. To follow this coaching model, we brainstormed an approach where we would have a "good" and "bad" video for each of the tips in the Tips tab.

After a few different explorations, we ended up using dropdown containers to hold the content for each tip. This decision resulted in the most engagement and intuitiveness, as we observed users exploring with the different tips and testing.

Presenting Golfers with Drills They Care About

As mentioned earlier, we provided users with drills in our current app experience, but they wasn't any context or information provided as to why they were recommended or how they would help. Through our research, we learned what information was important to golfers before choosing a drill, including how long the drill took, what equipment was needed to do the drill, and how the drill address the particular swing issue.

The card format of the drills we used enticed users, as almost every user in user testing was interested and clicked on one of the drills. They liked that they were able to gain some context into what the drill was, before they decided to do it.

Handing off to engineers 

In the previous work I had done at Golf AI, we would just have developers try to use our Figma file to bring our designs into life. But this had led to screens looking very different than the designs we had worked on. I had used Zeplin before at my internship, so I decided to give it a try - it ruined out to completely change the way we would do design-developer handoffs.

Zeplin greatly helped reduce the discrepancies between design and development.

I led the handoff and introduced the tool to the developers. In doing so, we built more empathy with developers and better understood their workflows, which helped ensure that coding our screens would be as easy and accurate as possible.

Handing off to engineers 

In the middle of August 2020, the screens were coded and finally implemented into the app! At this time, we also got new marketing interns that made ad campaigns using the new designs we had made, and highlighting how they built value for golfers. With all this progress made, we saw tangible changes in both active users in the app, and sales.

Screenshots of our newly designed screens in the App Store. Check us out at https://apps.apple.com/us/app/golf-ai/id1442971815!

We had 8 sales from July 20 - August 19th. But after August 20th (when the new screens were implemented), we had 12 sales in just just a little over 2 weeks!

Encouraging Golfers to Swing Again

The key measure of seeing user retention was if users came back to the app to take more swings. In our research, we immediately learned that users want to have a way to see how a new swing they recorded compared to the most previous one. And they wanted this comparison to be visual, so that they could immediately see how their form changed and how the app had helped them get closer to the ideal swing.

After the user has recorded their golf swing, they will be taken to a landing page that highlights their Swing Score. We added a clear explanation of the Swing Score concept since it wasn't cleary understood by some users. The "View Action Plan" call-to-action button takes users to their action plan.

If the user improved their score, this improvement would immediately be highlighted on the score landing page.

2020 Anmol Arora - "Can't wait for it to be 2021"

  • linkedin-icon_red
  • email-icon_red