Uncollected cases down 5% in four months.
Compliance risk gone in six.

Three businesses, three sets of data, one looming deadline.

Data across the three businesses was untrustworthy and siloed by business unit and service offering, which made accurate reporting and consistent decision-making impossible. On top of that sat serious compliance risk — data privacy, consent management and regulatory reporting gaps ran through all three businesses.

None of the teams had worked with machine-assisted decision-making before, and there was real scepticism about embedding predictive models into operational workflows. Investor pressure was mounting too, with the hold period approaching and a need to demonstrate measurable value without adding regulatory or reputational risk.

Three constraints were non-negotiable:

• Compliance risks around data privacy and regulatory reporting ran through all three businesses
• Data was untrustworthy and siloed by business unit, with no single source of truth
• Investor pressure was mounting as the hold period approached, leaving no room for missteps

A single source of truth, built in two months.

The team built a single enterprise data model and standardised workflows to sit at the core of the new case management platform, then transformed data from all three legacy business systems into that standard model — consolidating customer records, contact histories and case states into one trusted source of truth within two months.

The data model was privacy-first from the start: decision-support tools like propensity-to-pay scoring were firewalled from personal data, working from pseudonymised and aggregated attributes instead of raw identifiers. Alongside it, the team built consent and access controls, lineage tracking, approved transformation rules and an audit-ready reporting framework — remediating the identified compliance gaps within six months.

AI workstreams were deliberately reframed as ‘improving decision-making’, tied to concrete operational KPIs — propensity-to-pay accuracy, case prioritisation uplift, contact success rate, EBITDA contribution — with the technology kept under the hood so staff focused on outcomes, not tools. Clear board and investor communications mapped the roadmap to EBITDA and exit multiple targets at each stage, sustaining momentum through hold-period pressure.

5% down in four months, compliance clear in six.

Reframing AI as decision support, and keeping the technology invisible to frontline staff, accelerated adoption despite low initial AI maturity. The investor-facing roadmap tied every milestone to EBITDA and exit multiple, preserving momentum as the hold period neared its end — leaving the firm with a repeatable platform, reduced operational and compliance risk, and a clear line of sight from technical delivery to investor value.