Clayco: Revolutionizing Cash Flow Forecasting

Clayco specializes in the art and science of building, providing fast-track, efficient solutions for commercial, institutional, industrial, and residential building projects. Always innovating and leading the industry, Clayco revamped its forecasting methods with data science to improve decision-making, enhance project efficiency, and deliver even greater value to clients.

Before their investments into data science, machine learning (ML), and AI, almost all forecasting work at Clayco was done manually and on an ad-hoc basis in spreadsheets.  After building a data science function in their organization, they began modernizing and optimizing this process with the goal of producing reliable, scalable forecasts that provide actionable insights for the company.

Their first order of business: predicting cash flow. As with most businesses, if Clayco’s cash on hand (COH) isn’t able to cover the upfront costs of project work and other short-term financial obligations, they have to draw on a line of credit. But if the company keeps too much cash in reserve, they forgo investment opportunities that could grow the business in the future.

Before streamlining their forecasting practices through ML and AI, Clayco estimated their COH through simple spreadsheet formulas. These previous attempts lacked multi-faceted insights, weren’t used across the entire organization, weren’t scalable, and didn’t have any validation or monitoring capability.

Clayco’s data science team had two goals:

1. Deliver an agile, scalable, and reliable model to estimate COH.
2. Set up a data science practice and the necessary infrastructure for future success, starting with a team of just two data scientists and an intern.

Changing Established Practices With Dataiku

The Clayco team quickly got to work, understanding that to go from no data science to a full production model that delivers insights, they needed a platform that managed back-end infrastructure, supported flexible modeling approaches, and integrated with their existing systems and platforms.

We were a small team tasked with pioneering a data science practice at a 40-year old company. Dalston Ward Senior Data Scientist, Clayco

They found an ideal partner in Dataiku. Not only did the Dataiku platform fit what Clayco needed to be successful, but it allowed Clayco to focus on growing their data science practice at the same time.

Remarkably, the company was able to implement their unique use case in less than four months. They credit their success to Dataiku’s built-in ML features.

Leveraging the no-code options available in the Dataiku platform, they designed an XGBoost model to replace the spreadsheet-based predictions they used to analyze how long it took to collect open invoices. They also took advantage of the full-code options in Dataiku’s Python recipes, deploying a custom survival model that tagged atypical cases for review. After shifting to production, they automated scoring, retraining, and monitoring with Dataiku’s scenarios feature, freeing up time and resources to solve for the next challenge at Clayco.

Streamlining their workflow was made much simpler because they were able to integrate Dataiku with their existing tech stack.

Connecting Dataiku and Snowflake promoted collaboration between data scientists, data engineers, and business intelligence analysts. Harry Qiu Data Scientist, Clayco

Dataiku’s built-in git support and the ability to link to GitHub repositories contributed to their workflow, as the Clayco team was able to simultaneously progress multiple tasks, catch bugs early, and implement rigorous peer review. Indeed, Dataiku made deploying models safe and reliable. With its design, pre-production automation, and automation nodes, and its local deployer, Dataiku naturally facilitated a development-test-production approach and easily connected with dedicated testing and production spaces in Snowflake. This made it simple for teams to deploy changes with confidence, with low risk of downtime or degraded performance.

This architecture fostered a team culture that valued growth, accountability, and collaboration.

The Data Science team also relied on Dataiku for sharing the results of their work with other parts of the business. They used Dataiku to preprocess model scores into useful inputs before they went to a dashboard, making collaboration with their business intelligence teams more useful, and allowed their accounting teams to easily interpret and act on insights that their model generated.

Collaboration was also improved by Dataiku’s low- and no-code tools that allowed junior data scientists and interns to make meaningful contributions early and often. Users then upskilled easily with Dataiku’s tutorials which served as a gateway to more advanced data science work.

A New Data Science Status Quo

The initial investment Clayco has made into deepening and reinforcing their data science foundation is paying dividends for teams throughout the company. Thanks to models in Dataiku, accountants spend less time assembling forecasts from spreadsheets and more time engaging with insights and predictions in a custom dashboard. They’re able to act on more current, accurate information, which has improved Clayco’s cash flow management.

Previously, it would take an accountant a month to produce one report. The ML versions cut the production time so that they can produce them weekly. Their accountants also combine the model insights with their domain expertise to identify and resolve problem cases.

As more Clayco teams have become aware of the work that the small but mighty data science team has done for the accounting team, other departments are developing their own forecasting use cases and are eager to partner with them to solve challenges using ML and AI.

This initial forecasting project has had significant impacts to their journey towards data-driven decision-making. Their business has quadrupled the frequency they can produce cash flow forecasts.

The data science team at Clayco is also looking ahead. Dataiku has made it easy for them to reuse processes and code across different projects. They no longer worry about the “how” in building and deploying ML models — they focus on the “what” in the value they’ll create for the business.

Throughout their data science journey, Clayco’s employees have developed new skills and an understanding of how to build and produce ML models. They’ve formulated a template for future projects and a way to identify reusable components, all of which have served to sharpen Clayco’s edge in modern analytics.

Clayco remains committed to revolutionizing their data strategy, and they’ve found that data science is key to achieving a strong data culture that improves decision-making and reduces manual work. Dataiku has brought pivotal, recognizable value that helped the data science team at Clayco augment their success, foster their growth, and allow them to lead in their ongoing data journey.