What your current collections data isn’t telling you about your customers

Most teams have a reasonable level of confidence in their segmentation strategy. With CRA data, internal records, and historic engagement to work from, it often feels like the key boxes are ticked.

But confidence can be misleading when the underlying data isn’t telling you what’s changed.

Customer profiles change quickly: employment status, income, financial pressure points. And when those changes go unseen, segmentation models start to drift. And when those shifts go unnoticed, it becomes harder to tailor support, prioritise effectively, or meet the expectations set out in Consumer Duty.

In this blog, we’ll look at the signals standard datasets tend to miss, how those gaps play out in real-world decisions, and what a more complete segmentation approach could enable, from operational improvements to fairer customer outcomes.

 

One-size-fits-all strategies are hard to justify

Many firms still treat collections segments in broad strokes, based on ageing arrears balance bands, payment history, or static bureau scores. It’s a familiar approach, but increasingly difficult to defend.

The reality is that two customers with similar arrears balances may be in completely different financial situations. One may have a stable income and a short-term disruption. The other may have recently lost their job, taken on additional borrowing for essentials, and now be at high risk of harm.

Without accurate, up-to-date indicators of income, employment status, and financial strain, segmentation models can’t differentiate between the two. And that’s where the challenge starts.

Consumer Duty expects firms to design processes that support good outcomes. That includes identifying customers who need a different type of support. And not once they’ve complained, but from the outset. If segmentation doesn’t reflect real-life change, it’s much harder to meet that standard.

 

The less obvious signals your current data can’t see

It’s easy to think segmentation gaps are just about missing phone numbers or outdated credit files. But there’s a deeper issue: most datasets weren’t built to reflect how quickly people’s circumstances shift, or what those changes really mean.

Here are five signals that tend to go unnoticed but can tell you a lot about someone’s ability to pay, engage, or recover:

#1. Housing instability

A move from private rental to living with family isn’t always captured, but it often points to financial stress. Without that context, someone might be misclassified as low risk.

#2. Recent high-cost loan applications

These don’t always show up in CRA data, especially if the loan wasn’t approved. But even an application can be a useful sign of income volatility or mounting financial pressure.

#3. Contact channel recency

Last-seen data shows when a contact point was last active and how often it’s been used, helping your team avoid chasing leads that went cold months ago.

#4. Credit purpose patterns

If someone’s last application was to cover rent or energy bills, that tells you more than a credit score ever could. It’s a direct signal of essential expense pressure.

#5. Linked changes across data points

A change in employment, combined with a new address and a fresh I & E submission, suggests a major life event. When you can join those dots, you can respond faster, and fairer.

These signals don’t replace core data, but without them, you’re often missing the real reason a customer has fallen behind.

 

Why traditional data sources don’t give you the segmentation insight you need

CRA data shows how people have managed credit over time. It’s well suited to tracking payment history or balances, but it doesn’t pick up what’s changing in someone’s situation. Employment shifts, income volatility, short-term borrowing for essentials. These don’t tend to show up until much later, if at all.

Internal systems tend to reflect outcomes: missed payments, broken arrangements, and formal arrears. By the time those events appear, the underlying changes have usually been in motion for a while.

And even where contact data is in place, few teams can see if it’s still active, recently used, or tied to new digital behaviour. Without that context, segmentation becomes guesswork.

This is where FinTrace changes what’s possible.

The data is based on live consumer behaviour, including application-level insight from high-cost credit, real-time contact updates, and income signals that don’t rely on customer consent journeys. That means you can see all the less obvious signals we’ve just mentioned above (and more).

This level of detail gives you segmentation inputs that are predictive and treatment strategies are built around what’s true now.

 

What better segmentation makes possible

With live data feeding into segmentation, it becomes a tool for action, guiding how teams prioritise, engage, and escalate.

Here’s what that looks like in practice:

» Cleaner handoffs: Fewer accounts passed between teams. The right type of case lands with the right handler, first time.

» Faster decisions: Less time spent validating outdated details or requesting the same information twice. More clarity on next steps.

» Smarter use of resource: Teams focus on accounts where contact is possible and recovery is likely. Low-potential cases don’t absorb high-cost effort.

» Better treatment outcomes: Segmentation reflects what’s changed rather than what’s overdue. That makes it easier to align support with real need, without over-engineering the process.

» Stronger compliance position: When segmentation inputs are auditable, up-to-date and explainable, it’s much easier to evidence fairness and proportionality.

To sum up, better segmentation gives teams clearer priorities, quicker next steps, and fewer unnecessary handoffs.

 

Is your segmentation still fit for purpose?

Most segmentation frameworks look fine on paper. The challenge is knowing whether they still work when customer circumstances are moving faster than the data behind them.

These questions can help test whether your current model is keeping up:

• Are segmentation inputs refreshed in real-time or still based on credit application data and historic arrears behaviour?

If you’re relying on stale data, it’s likely you’re missing key changes in employment, income, or contactability.

• When circumstances shift, how quickly does your model reflect it?

A good framework can surface meaningful change before it becomes a complaint or a missed payment.

• Are vulnerability risks visible in your data or only raised when the customer tells you?

If self-disclosure is your main mechanism, the model is reactive by design.

• Do contact strategies fail without a clear explanation?

Bounce rates, goneaways, and failed RPC attempts often point to a data quality issue. If customers aren’t picking up, there’s a good chance the details are out of date or never verified in the first place.

• Can you show how each segmentation decision was made and what data it was based on?

Under Consumer Duty, it’s no longer enough to show that a strategy delivered the right result. You also need to show how the decision was made and whether the data behind it reflected the customer’s situation at the time.

 

If any of these questions raise doubts, it’s worth testing it. That’s exactly what our Proof of Value projects are for. We’ll run your data through FinTrace and show you what changes, what improves, and where the blind spots were.

Book a Proof of Value