Putting out fires
On bonfire night an explosive article in The Times blamed a faulty integration between Oracle and Blue Yonder (formerly known as JDA) for shelf gaps that had lowered sales by 3.1%, damaged customer loyalty and threatened the crucial Christmas retail period.
In July, The Guardian reported that Next employees had been underpaid by up to £200 a month after Next’s IT team struggled to integrate in-house software with Oracle.
Eighteen years ago, Computer Weekly reported that Sainsbury’s had written off a £260 million supply chain IT project after poor stock availability damaged its profits.
What went wrong?
One of the major benefits delivered by successful integration is that data is smoothly synchronized from one system to another. As experienced ERP integration consultants, the Times report suggested to us that synchronization had failed and stock movement updates were not reaching Waitrose’s central ERP in time. This would have the effect of a system showing sufficient items in store, when in fact they were out-of-stock. The lack of real-time data would impact delivery scheduling: resulting in shelf gaps, disappointed customers and lost sales.
Why ERP integration is business critical
In the fast-moving consumer goods industry, the reliability of supply chain infrastructure is business critical. Systems need to integrate seamlessly and update each other at speed and at huge scale. If something goes wrong, it’s immediately apparent to consumers and has a measurable impact on revenues. The Grocer reported that product reliability had fallen from 93% to 91% at Waitrose, while Kantar Worldpanel figures show that sales fell by 3.1% in the twelve weeks to 2nd October 2022.
Those numbers demonstrate why it is crucial for software engineers to undertake rigorous non-functional testing of ERP integrations before systems are made operational. This testing must anticipate and prove multiple failure modes so that everyone is clear on what to do when issues arise. When things do go wrong, the operations team are under intense time pressure to fix them. Every second counts.
Why it’s crucial to plan for failure
Enterprise integration involves connecting multiple systems, each of which could fail.
Integration can also involve carefully dismantling core systems before consolidating, simplifying and integrating with re-engineered systems. This is a high risk process as the data that flows around mission-critical applications needs to be redirected. Old enterprise platforms often have hundreds of applications running on them. Software engineers have to have an intimate understanding of application and data interdependencies so that they can meticulously plan how to move off old platforms one chunk at a time.
Clever coding aside, 80% of the work to provide a successfully integrated system involves planning what happens when things go wrong. Software engineers are usually managing a complex set of failure modes. Successful integration depends on how software handles these failure modes. To make bullet-proof solutions, software engineers need to plan how to keep everything running even when things break.
Why visibility is vital
Black boxes belong on aeroplanes not enterprise systems. Another crucial aspect of successful integration is the creation of visualisation tools that provide real-time visibility of the health of data repositories and applications, so that if something is not working as it should, the fault can be diagnosed and resolved much more quickly, before it impacts customer experience.
Enterprise integration projects demand a strong focus on site reliability engineering (SRE) that includes robust operational monitoring, alerting and diagnostics. That level of testing relies on having a support organisation that really understands how the system works. It’s important that this organisation is brought in at the very start of the project.
Don’t forget about the humans at the heart of ERP
While the Times article wrote this off as a ‘bungled integration’, this project had already been running for six years. This suggests to us that may have been an issue with the handover from the ERP development team to the ERP support team. It could be that the support team don’t know how to resolve issues when they arise, including delivery issues. It could also be the case that the operational side of the solution was not specified and that Waitrose concentrated solely on functional delivery. Success is built in at the planning stage. If the focus was on simply getting the new ERP system working, the potential operational glitches may not have been foreseen.
As the reported product availability and sales figures demonstrate, this type of ERP integration is mission critical, and so reliable operations need to be planned into the project right from the start. Reliability can’t easily be bolted on later as an afterthought.
Why simplification supports successful outcomes
Enterprises that are working with multiple IT suppliers may suffer organisational complexity that makes it harder to achieve the simplification, productivity and speed that the ERP integration was designed to achieve. When Sainsbury’s suffered stock control integration issues in 2004, it was discovered that the majority of the problems stemmed from four automated service depots provided by different companies. These all fell outside of the scope of the supermarket’s main IT outsourcing contract with Accenture. At the time, Sainsbury’s commented that the contract was being renegotiated to increase Accenture’s involvement in the selection and implementation of IT solutions and systems.
As the experiences of Waitrose, Sainsbury’s and Next have shown, successful integration depends not just on software engineering, but on planning, people, processes and pre-empting failures, so that systems will work reliably and be performant in the real world.