Addressing a big issue in the auto-repair industry

RepairPal asked men and women to call auto repair shops in New York to ask for price quotes 50% of women received higher price quotes than men, to the tune of a 33-94% increase. After conducting this study, RepairPal's CEO (Art) set out to build an experience offering men and women an equal opportunity when receiving their auto-repair quotes, aiming to remove gender based discrimination from those seeking car care.

This work focused on improving the design experience and content architecture to improve our funnel completion and shop conversion rates. Customers had expressed to us that the tool as is, was too confusing and complicated to understand.

The outcome of the re-design resulted in a 27% increase in funnel completion rates, a 31% increase in conversion upon estimate received, and drove 62% more customer contacts for shop leads that V1.

Read more on the study here.

an image of the final repairpal design screen


The kickoff included existing research the company has collected over time. When we started this project, we didn't have specific personas in mind since car repair spans across all (or most) demographics. The market research also informed our approach with user empathy and average confidence within the space.

A study conducted by Northwestern University (2012-2013) where men and women called the same group of repair shops to gauge quotes on a Toyota Repair, estimated to cost $365, revealed:

The North Star for our company derives from the lack of transparency within the car repair community. The estimator tool is one of many features built to combat unfair pricing quotes, and in addition to that, bring confidence back to the consumer.

Our existing pain points showed low completion, with only 50% of users completing all steps of the estimator, with the largest drop-off at the ZIP (step 1) and service inputs. Customer feedback showed that only 33% of users were satisfied with the results the estimator provided. The biggest area of concern surrounded our conversation rates, with only 0.1% of users booking an appointment after receiving an estimate.


How can we create a trusted and accurate pricing model for users' car repair, when users themselves might not know what is wrong? Users who are seeking to understand more about their anticipated car repairs and those attributed costs currently have little trust in estimates received from car repair shops. Statistics show that over 70% of women receive incorrect (and hiked-up) price estimates when seeking a mechanic.


Our solution offered a transparent and intuitive process for users that inspires trusted confidence. Users need to navigate through the flow with as little friction as possible, helping them understand each lever involved when attempting to diagnose their car repair while also making it as easy as possible for users to maintain accuracy. Additionally, we cleaned up the existing pain points within our estimator tool and layered in innovative additions to assist in our customers' repair journey.

Part of improving the process is heavily rooted in providing as much detail as users need regarding their selections step by step. We assist them in diagnosing car problems as accurately as possible in order to bridge the gap with unfair price quotes for those who are not as well versed in car mechanics versus those who are.


The outcome of the design resulted in a 27% increase in funnel completion rates, a 31% increase in conversion upon estimate received, and drove 62% more customer contacts for shop leads that V1.

Jobs to be done

JTBD: Funnel completion via delight

Users need a quick way to pinpoint their car repair needs through our tool with as little guesswork as possible. Our solution entailed 3D car models users would see based on their make and model selection.

From there, users could interact with those 3D models pointing directly to where their issue stemmed from, in addition to displaying the most common problems associated with those areas. This would make it easier to diagnose their potential problem, increasing accuracy of estimate, and increasing the # of users completing the diagnosis funnel to reach our shop marketplace.

JTBD: Build trust to encourage action

At the end of the process, users are able to receive their estimate without needing to enter any details like email or phone number. Users can print those quotes and take them directly to our partnered shops, where they honor any estimate received through our tool.

We partner with repair shops all over the country (500+ and counting). We direct users to our extensively vetted shop partners, providing confidence in their physical shop experience.

Outcome: Accurate estimates

With an easy to navigate process that builds confidence in our users ability to diagnose their car repair, we can increase their confidence in the quote and pricing breakdown regarding cost repair estimates received. With increased accuracy in estimates, their experience can remain consistent as they choose a shop partner to work with (no change in pricing/big surprises)

Outcome: Shop partner leads

We hypothesized that the more users who reach the shop marketplace at the end of the funnel would increase shop leads/contacts. We will utilize the number of users submitting their information and estimate details to shops, compared to users who confirmed and completed appointments to inform any further success.

Success metric: Product engagement

Target: at least a 10% increase in our estimator completion rates. Currently, only 30% of users complete the end-to-end flow.

Success metric: Appointment conversion

Target: 15% increase in conversion rates from the initial 0.1%. Hypotheses for this measurement were rooted in UI clean-up for quick wins.



Initial discovery

With a core focus on empathy, we aimed to view product and market space pains from multiple angles. Research shows when car problems arise, the initial feelings are nervous, frustrated, and overwhelm. There is a specific focus on negative emotions attributed to cost fears. Research also shows the majority of issues that people experience with their cars are not covered by a warranty.



We interviewed users that had undergone a car repair in the last 3-6 months within the age demographic of 21-50. We were inquiring about the user's mindset (feelings and concerns) as well as their approach (talking to family, friends, or individual research) when seeking repair guidance. W were additionally looking for any considerations users made with online search results, driving past shops on their commute, or word-of-mouth recommendations.



We scoped our interview questions to focus our conversations on users walking us through their end-to-end journey, starting with when they realized there may be an issue with their car. Prior to speaking with users, we ran a query on the most common car repair searches and the standard process users follow when searching for mechanics. This formed a "skeleton" process for us to compare with what users were saying.


Usability tests

Initial testing had participants go through RepairPal's site, providing insight on how users felt about the current design, focusing on pain points, struggles, and areas of delight. We utilized results to compare against further testing with the new design proposals to weigh the pros/cons against all variations. Additional tests were scenario-based using an interactive prototype that contained our hypothesized features.


Thinking creatively

We learned that users did not respond as quickly with images for car make and model vs plain text lists due to their emotional connection with their vehicle. Users would spend time searching for a car that looked exactly like theirs instead of a basic stock image which caused them to overlook their intended selection. I built a skeleton model for vehicle type that users would have no emotional connection to visually but still represented base car models.



We asked users to apply their previous repair journey to RepairPals estimator flow to provide feedback on the current experience. The results of those tests guided the final designs that were implemented. Feedback scoped down our MVP and showed us the highest areas of value for customers as well as removing the highest friction areas they currently experienced within the flow.


We ran 3 separate tests with customers between moderated and unmoderated formats. The first was focused on our current design and flow to diagnose biggest areas of pain and friction amongst users. The second and third set of testing were scenario-based, using an interactive prototype of proposed design changes that users could compare to the existing designs and share feedback on improvements or continuing confusion. These tests were focused on prioritizing which features should be involved in MVP, as well as assessing their ranked value.

Customer feedback was able to pinpoint the exact areas within the flow that required immediate attention to avoid further dropoff and abandonment. The main takeaways were:

  • Customers abandoned due to confusion
  • Out competitors were outpacing us by available tools and help with diagnosing car repairs.
  • The existing design was overwhelming users, making it hard for them to focus on the task at hand.

8/20 participants used the estimator at some point during their exploration. Of those 8 participants, 5 had difficulty navigating the services portion of the estimator due to it seeming disorganized and gave up.

7/20 participants got a quote during their exploration of our competitor and all 7 responded positively to the search function, the 'Diagnostics & Symptoms' section, and the way the services were organized.

the previous design on repairpals website

Updating the flows design was crucial to success

The disorganization of RepairPal's page design seemed to discourage participants. Users felt their lack of car knowledge was being highlighted when they were unable to find what they were seeking. 1 participant said "I feel like the repair calculator on RepairPal is too specific”. We learned from research that our main goal needed to focus on inspiring trust, confidence, and clarity within the updated design to encourage continued engagement within the flow.

Screens used in testing

the design screen for choosing your vehicle
the design screen for selecting a service

The core concepts I focused on testing with users in the updated design proposals included:

Participants preferred our updated designs as compared to our legacy designs based on 3 factors:

In the new designs, we tested out car imagery for every make and model from our database to populate images when users were at the "select year, make, and model" step. We learned that users did not respond as quickly with images for car make and model vs plain text lists due to their emotional connection with their vehicle. Users would spend time searching for a car that looked exactly like theirs instead of a basic stock image which caused them to overlook their intended selection.

Users liked the idea of utilizing an image to assist in their diagnosis when selecting from the diagnosis dropdown but having too specific representations of car color and other details caused users to overlook it due to their emotional connection. Utilizing a skeleton model for vehicle type, users would have no emotional connection to the visual, but would easily be able to identify their vehicle's body type.

The majority of users did not know what was wrong with their car but knew where the problem or noise was stemming from. If a user is not technically proficient or familiar with mechanics, they can utilize an interactive vehicle image (example above) to select areas they might associate with their 5 senses such as touch, sight, or smell.

This approach layers into our idea of "humanity in repair" within the designs.If a user is technically proficient or familiar with mechanics, we still offer a quicker method of searching in a database or selecting from drop-downs with "top" or "popular" issues based on make or model. These two methods allowed users who became overwhelmed quickly to easily navigate their experience, and also offered the experienced users a version that did not feel "dumbed down" (as quoted by some experienced users).

Finalized UI

Use of signifiers and affordances to drive user engagementI structured the process with a progressive disclosure model to simplify the user's view when making selections. I wanted to signify to users the amount of time they should anticipate spending on the process, and additionally orient them on a clear path following completion of each step, including what follows after receiving the estimate. Research showed if users did not have visibility into the amount of involvement, they would leave the page.
Improving funnel IA and providing 2 options for viewing content selectionsI moved away from the drop-downs used in the legacy designs. I tested the idea of listing out all available make, years, and models in steps (steps 1, 2, and 3) to provide users a high-level view of options in order to make quicker decisions with fewer clicks. I also added the ability to “clear” input fields to start over or request other quotes without having to leave the current flow, whereas the original design would redirect you back to the start if you wanted to change input.

the estimator flow landingthe design screen for choosing your vehiclethe design screen for centering your detailsthe design screen for selecting a servicea screen depicting hover functionality for diagnosing your cara screen to select your service a screen showing the estimate



Funnel completion increase


Conversion increase


Increase in shop qualified leads


Of the rocketship built