Introduction
Admiral Group is a Cardiff (UK), based insurer with 9,000 employees.. As digital demand grew, hand-coded pipelines and waterfall releases slowed delivery—deployments lagged, documentation & lineage were hard to maintain and impact analysis was risky. To move at the pace of the business, Admiral adopted an automation-first approach with WhereScape 3D and RED, enabling rapid prototyping with stakeholders, governed change and platform agility across Teradata and Microsoft SQL Server.
Pre-WhereScape Challenges
Admiral’s digital ambitions, combined with the increasingly disparate and complex nature of its data ecosystem, had outgrown waterfall methodologies. While the C-level needed fast, accurate data to inform and shape strategy, a reliance on hand coding meant requests took an unacceptable time to complete.
Aside from hand-coding restraints, Admiral also had the following challenges:
- Building and executing database deployments were slow and inaccurate
- Technical and business documentation creation delayed and out of date
- No clear way to track data lineage
- No database change impact assessments
- Problems in scheduling and managing component dependencies
The Solution
This clear gap in technical capability led the Admiral team to realize they could only achieve the agility they required with automation. They sought a tool to aggregate their data infrastructure into an efficient data warehouse, then enable them to model that data quickly to fuel custom BI reports.
While in the past this was done by developers writing custom lines of code by hand for each request, automation software enables the same team to move data sets around to rapidly create new structures within a wizard-driven, drag-and-drop GUI. It then automatically creates the code needed to action these requests using industry best practices, without human error. This allows IT to react to end-user requests in a fraction of the time, significantly reducing time to value.
Proof of Concept
Admiral’s IT team invited a software provider to their offices in Cardiff, Wales, to see if the products could meet requirements around time to production, but also future-proofing their technology investment and strategy. The team knew that with constant disruption from new technologies, any change made must also be adaptable in the future.
During the POC, the software demonstrated:
🚀 Time to Production
- Code can now be deployed into production in under a day, including dev to test and test to deployment.
🚀 Performance Testing
- The time taken to refresh a new dashboard (showing three months’ worth of data) takes 50 seconds compared to multiple hours previously.
- Data can be loaded daily rather than weekly, and this ingestion process takes just nine minutes. Delivering requirements around a web statistics project took a single two-week sprint, which was half of the time Admiral expected.
- Key features requested were a flexible design and self-service capabilities, both of which were fully enabled.
🚀 Futureproofing
- The architecture was confirmed to be adaptable with the ability to manage migration from Teradata to SQL Server. Platform-agnostic capability gives Admiral the agility to switch technology quickly and easily.
“The turnaround of our bugs has been incredible. It takes two hours now where it used to take us two weeks. The autonomy the developers have over deployments means a quicker turnaround for performing re-tests.”
– James Gardiner, Data Warehouse Technical Lead – Admiral Group
The Journey Begins
After excellent feedback from Gardiner’s team, the POC was deemed successful and Admiral adopted an agile philosophy. This started by arranging all data sources, from various silos, into a cohesive data warehouse structure. Rather than relying on a warehouse built by hand-coding and prone to human error, IT and business stakeholders agreed on the ideal architecture and modeled it using automation software, then built it by automatically creating faultless code that follows industry best practices.
Admiral chose a third normal form enterprise data warehouse on Teradata. The switch to this architecture was attempted previously but changes, bug fixes, and deployments took far too long to implement, which didn’t suit agile delivery mechanisms. By using automation to speed up these processes, including complex change management, agile development on Teradata was enabled.
Gardiner said:
“The software is very rapid to develop, very easy to learn. Even junior members and graduates with little experience can learn how to use it. Quick turnaround of deployments is very important. Running through all the checks and generating a deployment package is really straightforward and quick – we’ve been able to run multiple deployments and reconcile environments in hours rather than days. The turnaround of our bugs has been incredible. It takes two hours now where it used to take us two weeks. The autonomy the developers have over deployments means a quicker turnaround for performing re-tests. We have up-to-date technical documentation and mapping that testers use as a baseline, and lineage out of the box so it’s very easy to maintain and debug because you know exactly where everything is derived from.”
Where Admiral Is Now
Admiral was fully self-sufficient shortly after a successful five-week pilot scheme. “We very rarely hit an issue where we need support. We can usually work things out as a team if there is something we are struggling with,” said Gardiner.
The service IT can offer the business has been transformed – Admiral can deliver new projects at the speed demanded by today’s digital marketplace. The IT team can now enable agility and reduce time to value, moving from short consultations with business users to rapid prototyping and into production in just a few weeks using Collaborative BEAM (Business Event Analysis and Modeling) meetings.
Now the business can communicate exactly what they want, directly to the developers. In the past, closed silos and communication through a third party sometimes led to demands being misinterpreted and hours of work wasted. This cost valuable time and money, which is now minimized.
Using automation software, the Admiral team can now:
- Prototype data models to check requirements have been understood and can be actioned technically
- Convert agreed prototypes to working data models
- Deliver working functionality to end users at a fraction of the previous time
- Have vastly improved faith in the numbers, as automation removes the potential for human error
- Automatically produce perfect documentation and full lineage to accompany every action committed
Summary
In the six months since starting to use automation, Admiral has delivered four full projects to production in four different subject areas, and leveraged the flexibility of scheduling tools to prioritize EDW loads depending on business need.
What the Future Holds
Admiral is now expanding its data platforms by implementing SQL Server and Cloud technologies. The idea is to run the whole infrastructure under a common framework. Rather than specialized knowledge being held by just one team member, the whole team is now trained.
The full data lineage created automatically enables Admiral to be prepared for GDPR well before the regulations hit EU law.
“There were huge benefits gained from the automatic documentation generated. We use the track back and forward functionality a lot for accurate impact analysis. We know exactly what we have to change in the model and where it is because we can track it all. It’s all dependency tracked too.”
– James Gardiner, Data Warehouse Technical Lead – Admiral Group
