Digital and AI
7 common mistakes to avoid in collections implementations
Matt Riddall, SVP, Client Solutions · · 4 minute read
UK
Implementing a new collections system is a major step for any creditor, whether replacing legacy infrastructure or using cloud native solutions to digitise manual processes.
Matt RiddallSVP, Client Solutions, arum
What you need to know
- The seven most common implementation failures are all avoidable, and all predictable.
- Programme governance is the one that most often decides whether the others happen.
- Complexity is not the problem. Unmanaged complexity is.
When done well, it enables more efficient operations, improved customer outcomes, and faster recovery of outstanding balances. But too often, critical mistakes derail delivery, waste investment, and introduce risk.
At arum, we've supported over 500 system implementations globally so we know (and have seen!) the most common mistakes, and more importantly, how you can avoid them:
- 1. Don't treat your implementation solely as a technology project (it's also a business transformation)
- Mistake: Focusing solely on system features and technical delivery, without aligning business processes, people, and strategic goals.
- Example: A financial services firm implemented a collections system but failed to redesign its collections strategies. As a result, the system automated outdated manual processes and delivered limited benefit.
- How to do it right: Treat implementation as a transformation programme. Engage operations, compliance, IT, and customer experience teams from the outset, and ensure the system enables your future-state vision and not just today's processes built in legacy technology.
- 2. Don't underestimate the complexity of your data migration
- Mistake: Assuming historical data can be easily lifted and shifted without cleansing or restructuring.
- Example: A UK utility ran into serious delays when legacy account data didn't align with the new system's schema. Key fields such as customer segmentation, account status, and transaction history required major transformation.
- How to do it right: Conduct a full data mapping and quality assessment early. Build time into your plan for data cleansing, transformation, and validation, especially for live balances and regulatory-relevant data.
- 3. Ensure you test end-to-end journeys adequately
- Mistake: Testing only components (e.g., strategy logic or interface screens) without simulating real-world customer journeys and operational handoffs.
- Example: A collections team went live without testing a full 'missed payment to resolution' journey. The result: broken workflows between automated letters and call centre agents, with cases lost in the process.
- How to do it right: Develop a comprehensive test pack that includes full journey simulation (inbound file > strategy > activity generation > outbound contact > resolution > reporting). Include time travel testing and edge case scenarios.
- 4. Don't over-customise out-of-the-box functionality
- Mistake: Immediately modifying core platform features rather than adapting business processes or using configuration.
- Example: A bank spent months building bespoke case workflows when their system had configurable worklists that could have achieved 90% of requirements with minor tweaks.
- How to do it right: Use out-of-the-box capabilities where possible. Use a "configure, don't customise" principle unless there's a compelling case and consider the long-term cost of support and upgrades.
- 5. Engage and train users as early as possible
- Mistake: Delivering a technical solution with minimal focus on user training, communication, and role readiness.
- Example: Agents at a collections outsourcer were handed a new interface with little guidance, resulting in resistance, errors, and lost productivity.
- How to do it right: Start training early and tailor it to different roles (agents, managers, analysts). Use super-users, walkthroughs, and feedback loops to drive adoption. Document processes and decision paths clearly.
- 6. Define success metrics before launch
- Mistake: Launching without clarity on how performance will be measured or how to track benefits realisation.
- Example: A debt purchaser implemented a new platform but couldn't quantify whether it improved liquidation rates, reduced DPA breaches, or lowered cost-to-collect.
- How to do it right: Define measurable success criteria (e.g., contact rate uplift, cost savings, improved customer NPS) and build reporting dashboards from day one. Track KPIs monthly to identify early wins or course-correct where needed.
- 7. Don't ignore compliance and audit requirements
- Mistake: Addressing compliance requirements (e.g. FCA, GDPR, internal audit) as an afterthought rather than a design principle.
- Example: A collections system went live without full audit trails on strategy changes, creating regulatory risk and extra effort retrofitting controls.
- How to do it right: Embed compliance into your design authority and delivery checkpoints. Ensure systems include traceability, access controls, and audit logs from the outset.
Final thoughts
Collections system implementations are complex, but common pitfalls are avoidable with the right approach. The most successful programmes have three things in common:
Strong programme governance
At arum, we support clients across the full system lifecycle, from strategy and selection to implementation and optimisation. If you're planning a change or already on the journey, let's talk about how to de-risk and accelerate your success.
From the same author
More from Matt Riddall.
Digital and AIWhat three weeks at MIT taught us about building agentic AI
Ella and Matt took MIT's agentic AI course to pressure-test the AI-assisted system arum was already building. Here is what held up and what did not.
Matt Riddall and Ella Grice · 17 June 2026
Digital and AIIs your AI agent actually ready? 11 questions to ask first
Eleven questions to answer before an AI agent handles a customer in arrears, from a team that has been building them since 2019.
Matt Riddall · 2 April 2026
Digital and AI5 ways to build a strong collections system implementation partnership
Choosing the right collections system implementation partner is one of the most important decisions your business can make.
Matt Riddall · 22 July 2025
Keep reading
More from arum.
Digital and AIBuy vs. build for debt collection platforms
In today's economic and regulatory environment, transforming collections capability is high on organisations' strategic imperatives.
Reuben Gates · 13 May 2026
Digital and AIPreparing for AI in collections and recoveries
Artificial Intelligence is rapidly moving from concept to practical application across financial services.
Forid Meah · 16 March 2026
Digital and AIShared ETL code libraries: improving data quality and delivery
Shared ETL code libraries stop the same logic being rebuilt in every pipeline, and lift data quality and delivery speed in collections operations.
Stephen Wright · 19 February 2026
Talk to arum
