Improving data-driven decision making for a large environmental services company with dynamic dashboard reporting
Customer Background and Context
In the environmental services industry where hundreds of thousands of households are serviced each year, degradation in data integrity can have a real impact on operational efficiency and the bottom line. Our waste and environmental services client spent an inordinate amount of time manually standardizing and curating data in about 50 reports per year on this issue. They found that unbilled servicing plus complex garbage collection scenarios result in an estimated annual loss of $18 Million.
Solution/Tech Stack Overview
Apex Systems partnered with this client on several projects and work streams of which the Parcel Data Analytics (PDA) solution was a part (leveraging a partial scrum team delivery model). The Apex team worked with the client’s Product Owner to understand how each deficiency could be measured and what key features this analytical solution should contain. The tech stack associated with this project included:
- AWS
- Matillion
- Power BI
- Snowflake
Business/Functional/Technical Requirements
Because this client is currently migrating its enterprise data platform to AWS and Snowflake, our team was able to bring keen analysis and engineering skills to solve the challenge. The Apex team – comprised of data modelers and engineers – designed and implemented several data pipelines using Matillion, which curate data specifically for a custom Power BI dashboard, to advise business users where locations were potentially being serviced without being appropriately billed.
Service Areas/Capabilities Deployed
These pipelines integrate data from several source applications, including both traditional billing systems and from connected-vehicle systems which supply video, image, and location data of the collection vehicles. Artificial Intelligence (AI) services in AWS were leveraged to classify conditions such as “over-fill” and to confirm that the truck was indeed servicing a “suspect” location. Other deterministic algorithms were used to associate truck position (measured by latitude and longitude) to the customer’s location.
Partnering Model
We leveraged a partial scrum team delivery model with a team of data modelers and engineers to deliver on this Parcel Data Analytics (PDA) solution.
Commercial Construct and Performance Management
The commercial construct for this project was time and material billing. Our dedicated Engagement Manager and Technical Solution Lead provided ongoing performance management throughout the build, communicating stakeholder feedback and guiding consultants through to product delivery.
Impact
This complex technical solution was implemented in a performant and intuitive way for business users (accessing this information through the same BI tooling they were accustomed to). Almost immediately, $2 Million of savings were realized through the insights provided by the data and dashboard, and the client saw a ten-fold increase in the number of reports it could generate and analyze.
Lessons Learned
- Communication is important - emphasizing communication and collaboration with the product owner and business stakeholders is mandatory for success.
- Always establish success criteria first - establishing minimum viable product (MVP) success criteria to make sure key features are defined upfront and a plan to prove their feasibility was quickly validated.
- Be proactive - proactively identifying dependencies and risks by allowing the technical team to consider solutions while the stakeholders and product owner align options to business outcomes. Top-notch technical skills around data architecture, engineering, and business intelligence are table stakes with Apex teams which enabled us to rapidly address data consistency issues with one-off solutions that enhanced the overall PDA solution. It was the confluence of technical acumen with a proactive partnership that drove the solution to real success.
