Clutch is Canada's largest online automotive e-commerce platform, with whom customers can buy a used car completely online, hassle and haggle free, and have it delivered to their home. The business prides itself on being able to provide a seamless and transparent, no-pressure car buying experience to Canadian drivers across the country.
Growing pains and a shift in the winds
Given a less desirable change in the macro-economic landscape in early 2022, the company devised a new plan that was geared towards better unit economics rather than solely focusing on volume. The reason for this was twofold:
Our explosive growth during COVID was not something we could take for granted or advantage of forever.
Clutch had grown to the point where our initial systems weren't appropriate to our scale anymore.
As such, our product team accordingly identified a set of front end and back end initiatives to move us in a new direction:
One of the biggest items on our 'to tackle' list was reconditioning costs. Reconditioning in this context referred to the work done to make a newly acquired vehicle clean, functional and safe as mandated by regulations and internal standards.
As evident from the diagram above it was a high impact, medium effort project that would yield great results if done properly. We therefore put together a dedicated pod of product manager, designer and a team of engineers to get this done early in Q3 of 2022, with a projected launch by Q4.
The current state of things
To get a sense of what needed to be done, we first had to understand why our current process wasn't working. Alongside our product manager, I sat down with our Ontario facility managers, mechanics, and even our accountant to get a better picture of what the whole process looked like.
Simplified illustration highlighting the main roles within the reconditioning process in Ontario
We spent a week unravelling how an inspection report is done, what tools were used and how many touchpoints happen for a standard vehicle reconditioning. At the end, we were able to understand the full extent of our reconditioning troubles.
Wait... what was it again?
By relying on analogue methods for our inspection reports, our teams did not have a centralized point for storing and updating information regarding repairs. Given how many times a vehicle changes hands, our staff spent more time verifying information with each other than doing the required work.
Estimating estimates
As a byproduct of the poor record keeping process, our cost estimates were also wildly inaccurate compared to our final bill. It also didn't help that sometimes our vendors would recommend more work to be done, leading to hastily authorized purchases that accounting has to resolve without written records on our end.
Non-standard standards
Clutch prides itself on the quality of vehicle sold to a customer and the reconditioning stage is where it all happens. While a standard manual existed, adherence to it was non-uniform which often meant more work and higher costs for Clutch. If we were to bring down our reconditioning costs, better guidelines and stricter controls were needed.
The game plan
Being able to diagnose our problem was a big step towards fixing it. Our guiding question was now:
How might we help our reconditioning teams process vehicles efficiently while maintaining our high internal quality standards?
Collaborating with my PM, we devised two solutions. One of which dealt with the quality control portion of our problem, the second became this project, a reconditioning cost tracking (RCT) tool — it would serve to centralize and organize information relating to reconditioning work costs, allow for better estimates and payment reconcillation, and allow team leads to approve, review or reject the work proposed based on said estimates.
By doing so we aimed to inroduce more accountability, consistency and efficiency into the once chaotic workflow.
How we're measuring success
Now that we had a sense of the scope, it was time to set goals for what this software should accomplish. To that end, we set up three major metrics by which we would measure the success of this project.
Implement new standards
Achieve 90% adherence to our reconditioning standards
Improve estimate accuracy
Accuracy of cost estimates equal or greater than 95%
Reduce unit costs
Get average reconditioning costs down by 40% per vehicle
Establishing our targets early was critical not just for evaluating the end product, but to guide our MVP as well.
Early proofs of concept
Our MVP was put together in a matter of weeks; a week for design and two for the engineering. Naturally everything here was pretty rough from the get-go, but we had to consciously make an effort to not let perfect be the enemy of good. We just wanted something out so we can start testing and gathering data.
In our MVP we included the most basic of features that we knew would form the backbone of our new system:
Creating a reconditioning job from a standardized reconditioning list,
Inputting cost estimates for said jobs and;
Adding and associating invoices for these jobs once completed.
Once we had the MVP out, we immediately put the system in the hands of our Ontario reconditioning team. Given that having something was better than nothing at all, the teams immediately took to the MVP and gave us fantastic feedback over the course of a two week trial run.
Lessons from our MVP
By all metrics, the MVP proved that we were on the right track; initial numbers were better than what we hoped for and gave us a much needed confidence boost. Despite that, there were also a few areas we had to work on:
Allow for multi-invoice uploads
We needed to allow more flexibility when assigning invoices, as work is usually done in batches.
Account for more edge cases
Returning parts and cancelling labour is very common and always affects cost calculations.
Have better vendor integration
Cost tracking with vendors was still highly inconsistent due to loopholes we didn't anticipate.
All of these improvements were taken into consideration for our V2 that we rolled out shortly thereafter.
Reconditioning 2.0
Incorporating all of the feedback, RCT 2.0 would become more fleshed out than before.
Our first order of business was to tackle how invoices were being uploaded. Our MVP called for one upload invoice per job, but because vendors almost always collate invoices and send them out in batches, this quickly led to one invoice being uploaded repeatedly. We obviously didn't want our mechanics to be uploading invoices all day - especially not the same ones - so we came up with a system to dramatically cut down upload times: allow for multiple jobs to be selected when uploaded, then the software would handle the rest by making copies for every single selection.
Unused parts and labour that has been cancelled before commencement was a common occurance, and our MVP had no ability to account for or rectify that, preventing our cost estimates to achieve the best possible degree of accuracy. I subsequently designed a system that allowed returns to be intiated from and attached to an existing order, then using a modified version of our new multi-invoice upload system, single out the unused part or labour and appropriately assign receipts for that one aspect of the job only.
While we cleaned up our interal cost inconsistencies with the aforementioned two major fixes, we still had a problem with our vendors adding on unauthorized work on site last minute. To tackle this, we took a different approach. We negotiated with our vendors to adhere to our Purchase Orders (P.O.) system, an automatically generated document that was sent prior to any work being done. Since only authorized work would be on the order, anything that fell outside of it required another P.O. to be generated. This gave us a perfect paper trail that would also help accounting with finances after.
Of course, a new version also required visual polishes and quality of life improvements. Among many minor tweaks, one of the biggest changes here was the move to condense information and make editable cells instead of the formerly space heavy input boxes.
With all of these changes, we knew we were close to the final product and were confident in making our Q4 release date.
Putting it all together: RCT 3.0
The results
Once we had given our mechanical team the full rundown of the latest changes, it took them almost no time to acclimate and continue work as usual. As we monitored the data that was coming in we found that not only did we hit our targets, but in some instances, exceeded them as well. What's most significant is our cost savings; We were able to bring average reconditioning costs down by 45% per vehicle, which represented hundreds of thousands of dollars saved.
With RCT we were able to be more efficient and consistent with our resources while still giving customers quality cars with amazing value. Thats a win-win for everyone involved.
Within a month of launching in Ontario, our software was adopted by all of Clutch's other markets with equal success.
It was extremely rewarding to see our reconditioning teams process vehicles more efficiently, with significantly reduced frustration and mental fatigue. Our accounting team was also equally pleased with the greater degree of transparency and accountability, not to mention the massive time and wage savings for the business in general.
What's Next
Summary
Building an ecosystem
Our cost tracking tool was just the beginning. Riding on this success, we released three additional pieces of internal software to optimize and scale operations, namely a vehicle inspection and onboarding suite of software that served as a precursor step to reconditioning, as well as delivery tracking and management system for the vehicle further along it's reconditioning lifecycle.
Lessons learned
What is also important here was what we learned as a team and business. The success of this project proved to the organization that good process and informed design decision making was not only beneficial, but essential in helping our business scale going forward. Whereas formerly we'd only emphasize design in customer facing products, this project paved the way for design to be involved in the decision making process and add value for our internal tools as well.
Want to learn more?
There's a lot more detail and thought that went into this project than is practical to show in this format. If you'd like to get a full walkthrough, please contact me.
.jpg)
.svg)
.svg)
