Credit decisioning and scoring | Credit Decisions | Process management

What is the key to agile lending in a competitive market?

Published October 2024

| Credit decisioning and scoring | Credit Decisions | Process management

Written by Jonathan Taylor

As a Senior Consulting Partner within Experian’s dedicated Banking Team, Jonathan is responsible for helping lending organisations implement data and analytical change programmes across the credit environment.

Mastering ModelOps

As lending organisations are looking to become more agile, the ability to react to market opportunities and pivot into new lending sectors or niche segments will be driven by their ability to ingest data and model outcomes faster than their competition. But the reality is that over 65% of UK organisations report that it’s still taking too long to develop and deploy new credit decisioning models.

Skip to section...

With over 70% of lenders recognising that ModelOps will shape the industry’s future over the next five years, finally the importance of data ingestion, rapid iterative model development and deployment will become a key market distinguisher is an ever more competitive environment.

The importance of your data infrastructure

Nearly 1/5 (18%) of UK lenders said that improving their data infrastructure was one of their biggest priorities around modelling.

Having the right infrastructure available to evaluate data, develop new models and fine-tune existing ones has often been a disparate time-consuming process. In an environment whereby data evaluation and model deployment has to keep pace with the needs of new digital decisioning platforms, a new phase of Data and Model development (ModelOps) processing will to come to the fore to improve and further accelerate credit risk decisioning.

Industry challenges

When asked what key barriers organisations face in developing new models or updating existing models in lending, constraints with existing technology and integrating models into the lending process were two of the largest challenges.

The need for more data and faster analytical updates is challenged by legacy data infrastructures and internal processes., The time taken for lenders to evaluate and deploy new data assets can be slow and more than often reactive and dominated by a stream of regulatory changes.

To ensure our ongoing solution and data strategy is continually focussed and aligned to key industry challenges, we undertook detailed analysis with a number of tier 1 banks to understand their primary issues in more detail.

Head of QA

“Although we want to get to a place whereby we can be more dynamic in our data evaluation and quicker in our model techniques and development, we are someway off.”

Credit Risk Director – Retail

“Unless any data programme adds significant new uplift or value rarely seen in bureau scores these days, funding for such large-scale programmes is unlikely to get the priority.”

VP Risk – Unsecured lending

“Our biggest challenge isn’t the modelling, it’s reliance upon other areas such as IT and critically ensuring all new data/scores are updated in our DataMart for MI and monitoring, which has huge lead times.”

CRA Strategy Manager

“Today there are circa 25+ models across Acquisition & ECM in total – not all will use external data but a good proportion will, therefore if you consider this applies to only 1 product then the overall total impact is considerable.”

29% of UK lenders admitted that time spent on the overall data assessment and model development process was more than 24 months. And 21% referred to data gathering and data execution within model operations as the key themes that consume the most organisation’s time and resources.

Worryingly, data used within models/decisioning (e.g. policy rules, scorecards) etc. is often restricted by lenders ability to consume and ingest it for analysis and implementation, often meaning new predictive data is not always considered when new models are being developed, potentially having a detrimental effect to both lenders and consumers.

Jonathan Taylor, Senior Consultant at Experian

Addressing the legacy challenges

To gain that competitive edge, lenders and data providers are now finally addressing the barriers and focussing on the long-term benefit of accelerated data and model development processing.

This primary challenge and the market’s need for data and model acceleration to remain competitive means that active, proactive and innovative lenders are looking to invest and address the following operational and analytical data blockers.

Data lineage and time-consuming data preparation

Data lineage/consistency and accuracy across an organisations heritage data estate still requires constant data re-engineering to create an effective data assessment and thereafter a rapid modelling process.

Elapsed time for clients to physically deploy new scoring models is actually extending

Due to the accessibility of the data within the development (analytical) and live operational environments, plus the disparate manual implementation (IT/technical) phases.

Legacy data and analytical environments negate challenger approaches

Restricted ability for clients to parallel run new data and models, new data/scores are not proactively incorporated into analytical environments and data warehouses ahead of any reactive score or model decommission.

How can we help?

To compete in today’s lending environment where margins squeezed by the cost of funds, pricing and challenger competition, faster data optimisation and the acceleration of model development will be a key differentiator. Recognition and focus, previously concentrated towards wholescale banking platforms or risk and technology infrastructures, are now being channelled to the data and model development environments, making the faster adoption of data and modelling roll-out a key strategic investment area.

Our suite of services within the Ascend Technology Platform answer and deliver against the latest industry challenge of accelerated ModelOps development processing, offering comprehensive, compliant data, actionable insight and an advanced analytical sandbox.