CCM/DataVue

Where they were

CCM is a US credit marketing company that depended on an external vendor for their credit scoring model. For all practical purposes, this model was a black box they couldn’t audit, improve, or explain to their own clients. With more than 300 million consumer records flowing through Equifax and other providers, they needed to understand their own data; but they had no data team, no AI, no infrastructure, and no way to build scoring models in-house.

What we found

The surface problem was “we need a datalake.” The real problem was deeper: CCM’s entire business logic (who to target, with what offer, at what cost) was outsourced to a vendor they couldn’t question. The scoring models were opaque, the data pipeline was manual, and there was no way to iterate on model performance. CCM didn’t just need infrastructure: they needed ownership of their most important intellectual property.

What we built

A first machine learning model became an entire product, separate from the initial business unit. Over the years we achieved many things: encrypted data ingestion with AWS Transfer Family + PGP + Step Functions, built a datalake on S3 and SingleStore with more than 4500 attributes per record, and created over 30 ML scoring models (response and conversion) with MLflow on SageMaker, automated training and scoring workflows, a data visualization layer, and monitoring and retraining pipelines. And as a consequence of all this, CCM was able to design DataVue, a new product, which adds an APRA compliance framework for US data privacy.

What they can do now

CCM operates DataVue independently. Response rates doubled. Conversions up 130%. Scoring error down 30%. They went from an opaque vendor to over 30 models that are theirs and that they understand. The platform now serves 3 times more clients than when it started.

Why they stayed

26 months of continuous engagement. What started as a datalake project evolved into a complete ML platform, a new product line (DataVue), and a compliance framework. CCM kept expanding scope because they trust the diagnosis and the execution.