Compology

Streamlining the Rightsizing Workflow

Timeline

User Research
Product Design
Visual Design

Role

6 months

Tools

Figma
Userinterviews.com
UserZoom

Collaborators

Product Management
Engineering
Customer Success

Impact

Average client savings of $1800 per dumpster per year or 30% reduction in total waste costs when rightsizing recommendations are implemented.

What is Compology?

Compology is a hardware plus software solution that uses machine learning to help Fortune 100 companies, waste brokers and governments meter their waste. The outcome is greater insight into an organization's waste data; allowing customers to eliminate inefficiencies, drive toward their sustainability goals, all while reducing waste costs. 

What is Rightsizing?

Rightsizing has historically been a manual process of auditing the amount of waste produced at a site to determine if the current pick-up schedule is adequate or not. Compology has been able to automate this process by running historical data collected from in-dumpster sensors through proprietary machine learning algorithms. These algorithms can then recommend optimized service schedules to businesses and waste brokers, ensuring they are always paying the lowest amount possible for their waste collection needs. 

Business Problem

Since this was a greenfield product with no competitors or previous customer feedback to guide us, we had very little insight into how these recommendations would be used as part of our customer’s day to day workflow. Because of this we decided to launch an initial version of rightsizing from within the Compology application that would require minimal engineering and product effort, but give us valuable feedback into what information users needed to trust and accept the recommendations.

This initial version was added into the detail view of a dumpster, which could be accessed through the core containers page. A user could filter to see all of the containers that had a rightsizing recommendation and then click into the container to view the details of the recommendation, the data that informed the recommendation, and accept or dismiss it.

We tracked the percentage of recommendations that were both accepted and dismissed within the web application over the course of several months. We found that users were only interacting with ~5% of the recommendations. We were also continuing to receive requests for custom reports with rightsizing recommendations, which led us to believe that customers viewed the recommendations as valuable, but found them difficult to access or use within the current UI. This posed a business problem because:

  • The more savings customers realize using Compology, the more likely they are to renew and expand their contracts.

  • If we can show high realized savings across our customer base, potential customers are more likely to trust us and buy the product.

Core containers page where users could filter to a list of all containers with a rightsizing recommendation

View when clicking into a container, which shows the rightsizing recommendation and the data that informed it

Rightsizing Version 1

User Problem

To hone in on the issues users were having, I knew we needed to dig into when and how our customers were currently rightsizing. From previous research we already knew a bit about who was doing rightsizing.

Who Does Rightsizing?
The primary group already engaged in rightsizing were our waste broker customers. Waste brokers act as a middleman between a waste producer and a waste hauler.

Large waste producers who have many locations across a large geographic area may have contracts with multiple waste haulers. A waste broker can act as a manager of these contracts, getting waste producers better rates, and finding other means of making their waste programs more efficient and cost effective. 

Within a waste broker organization, an account manager is responsible for managing the waste programs of their set of customers. They are the users who are primarily responsible for enacting any rightsizing service changes.  

User Interviews
Given this, I knew it was imperative to talk with account managers to understand what their daily responsibilities were, how right-sizing fit into that, and if/how they were currently using Compology rightsizing recommendations. I conducted a series of interviews with both account managers within waste broker organizations as well as our internal customer success managers who interface regularly with account managers.  

I came away with the following findings which I used to develop a journey map and a light weight persona that could act as a guide throughout the design process.

  • Account managers' highest priority goal is to maintain a good relationship with their customers. 

  • Because of this, they will prioritize urgent customer requests and needs over operational optimizations and finding cost savings. This means rightsizing falls as a lower priority task that gets done in short bursts during slower times of the day or week.  

  • Account managers often receive bonuses in the form of savings sharing. They will receive a percentage of the money they are able to save their customers.

  • Their limited time combined with their bonus incentives means account managers want to focus on rightsizing recommendations that have the highest opportunity for savings.

  • Account managers need to get approval for any schedule changes from both the hauler and the customer before they can implement them.

Pain Points
Through these interviews I was able to glean the following pain points in the current UI:

  • The volume of recommendations made it daunting to act on all of them or tell which ones fit the account managers criteria for high priority.

  • Because the details of a recommendation were located within the container details page it was not possible to quickly know the savings that would be generated from accepting a recommendation. 

  • Getting stakeholder approval meant it took a lot of time to switch schedules for just one container

  • No way to prioritize, group or compare recommendations. Some of the ways we heard that account manager wanted to do this was:

    • By hauler - so they could focus on quicker stakeholder approval

    • By savings - so they could focus on higher savings first

    • By customer - so they could focus on larger/more important customers

Goals

After presenting these findings to the team; the product manager, the Chief Product Officer and I collaborated to narrow in on the following goals for our second iteration of this feature. We wanted to:

  • Help users determine the customer/hauler/locations with the highest savings.

  • Allow users to see all recommendations for the group they care about in one view.

  • Help users get approval for recommendations from stakeholders more quickly.

  • Increase the speed in which users were able to accept and dismiss recommendations.   

To measure the success of the feature we knew we wanted to see the following:

  •  An increased acceptance rate for recommendations in-app.

  • A decrease in the monthly number of requests for custom recommendation reporting. 

Wireframes

Next, I came up with several high-level ideas for how we might accomplish these goals. 

 Solution 1
Keep rightsizing recommendations on the containers page, but add granular filtering, more recommendation details on the container card and the ability to accept/dismiss recommendations.

Solution 2
Add rightsizing recommendations to the customers page. Show a summary of the rightsizing recommendations for each customer and allow filtering to narrow to high priority customers. Each customer can be opened to show a detailed list of recommendations and  the ability to accept/dismiss them.

Solution 3
Add a new page to top the level navigation. This page would initially include just right-sizing recommendations, but would expand over time to include recommendations of other kinds that we already have planned in the roadmap. The recommendations would be in a table format to align with users' current mental models around data. 

Narrowing on a Solution

These solutions were then presented to product management and engineering. We collaborated to determine the effort for each solution and then weighed that against the impact that each solution would have towards our stated goals. 

We were able to narrow in on solution 3 as the best path forward and I went through several rounds of iteration and feedback before settling on a solution we all felt confident in.

Version 1

Version 2

Version 3

User Testing

To give us additional feedback before moving to development I completed a round of usability testing with 5 users that were in charge of running the rightsizing program for their organization. We chose this method to make sure that the design we proposed actually met their needs and was easy to use. From this we learned the following:

  • There were additional data points users needed in order to make decisions that were not initially included in the UI, for example: when the recommendation was created or, why a container did not have a recommendation.

  • Users wanted to curate which columns of data they share with their customers or haulers, for instance they do not want to share potential savings with either their customer or hauler.

  • Users needed a clear affordance for accepting/dismissing a recommendation that was separate from the status of the recommendation.

  • Certain data points were often considered together when making a decision and therefore needed to be close together in the UI for easy visual scanning.

  • Overall, participants conveyed a preference for this new in-app experience because:

    • It was easy to filter and drill down by customer or hauler.

    • The layout made it easier to scan the recommendations and associated data. 

    • They were able to mark the status of a recommendation.

Final Solution

Taking the information from the usability testing into account, I revised the UI a final time and then worked closely with engineering throughout the development process to provide any necessary context and design updates, so we were all in alignment about what we were building and why. 

Launch

Then we launched! Tracking our key metrics over the following weeks we began to see a steady increase in the number of recommendations that were accepted through the app. 

Learnings

It’s important to take a step back and look at the bigger picture of what your user is doing. Cool, the flow is intuitive and easy to complete, but how many times are they going to have to do it? How does it fit into their overall workflow? Is doing this flow a priority for them compared to everything ELSE they need to do?