The Customer & Project
We had just completed our second round of integration testing for a large ERP program go-live and were approximately 9 weeks from cutover. The optical division of a retail business was implementing SAP, new planning and forecasting tools and a substantial upgrade to a custom Point-of-Sale (POS) solution that affected over 800 stores crossing 3 optical store brands. The year prior, the flagship Optical brand had gone live with the new platform and had a very rough cutover. For our deployment, there was a 16 hour window on a Saturday night/Sunday morning to upgrade all the retail fronts with a new POS system. We were in the steering committee meeting reviewing the program dashboard, which was a series of green and yellow lights. Overall consensus that we were on target for a successful cutover, while assuming some manageable risks.
One indicator was still red though. Earlier that week during performance testing for the POS system we had realized that the very dated hardware for almost 35% of the Sears Optical brand stores was not going to be able to run the new version of software. We had to deploy a third party vendor to upgrade the memory on those boxes and run performance diagnostics ahead of the cutover. The good news was that those stores were in the most remote parts of the country for the smallest brand being implemented- so if there was disruption, it would not be at a flagship or high volume location. A quality issue would be relegated to a handful of stores without significant financial impact. The bad news was that getting those boxes upgraded on-time would be nothing short of miraculous because of the remoteness of the locations and the fact that the “store within a store” model limited access for the contractors. We had just planned the hardware deployment and found that we had no room for error.
The year prior when the flagship optical brand had cutover to the new platform; the weakest link in the program had been the performance of the upgraded POS system. For several weeks after go-live, many of the stores were augmenting with pen and paper “system down” kits because, while sales transactions would go through, less used but equally crucial transactions such as returns and exchanges would cause the POS terminal to freeze up.
Prior to the project, the POS software was over 15 years old and had been customized and patched substantially; there was virtually no way to get the solution to integrate with a tier-one ERP solution without refactoring much of the code. The business decision was to continue with the existing POS solution rather than replace it. The reasoning was that the organization could handle only so much change and we wanted to minimize the change impacting customer facing associates. Replacing the POS would be a future project. With that direction, the development team took on the heroic effort to transform the data model and re-design the integration, working around the clock.
It was soul-crushing to the team when the go-live was so problematic. A lot of excellent resources left for greener pastures because, after all that hard work, they bore the brunt of the blame post go-live. When preparing for this project, all of those challenges were re-hashed and flushed out in an effort to minimize the likelihood of repeating the mistakes.
With the memory of the difficult go-live still fresh in their minds and significant pressure to make up for the associated lost sales, the leadership team was understandably gun-shy to pull the trigger on the cutover. The business teams were restless and keeping everyone focused and believing in our success was crucial. We ran through the active risk mitigation strategies and what we would do if the hardware replacement was running behind. Mary Anne, GM of Sears Optical Brand, spoke up.
“That weekend is our friends and family event. The day of go-live represents our largest sales day of the year. Is there any way that our brand can delay to the following week? Our sales this year have been lower than expected and we cannot afford lost sales due to delays at the point-of-sale.”
The room went quiet. Friends and Family weekend timing was driven by Sears. The Optical store front within the store was owned by Luxottica. The go-live date had been set for the better part of a year and making changes to that degree so late in the game could jeopardize all 800 stores across the other two larger brands as well as Sears Optical. Everyone in the room was hyper-aware of the weak sales position of the brand and the importance of strong numbers to it’s future viability. The timing of the friends and family event was not communicated previously, most likely because despite the importance to that brand, it was the smallest sales volume and number of stores to convert for the project – so the risk would likely be assumed for the good of the Optical division as a whole.
The President of the division, Mark, spoke up. “Mary Anne, we probably cannot change the date, but we understand how crucial that sale is to the Sears Optical team. Shannon – figure out how we can guarantee that sales at those stores will not be affected.”
My heart was in my throat as the dictate to “guarantee” a solution was difficult to swallow. “We will review our options and initiate a change request today,” I replied. Mark nodded in my direction and the meeting adjourned.
A change request was not the only way to respond to that request, but it did several things:
• Provided time for our team to huddle and come up with options
• Ensured it was clear to everyone that a shift in priority was being requested. While there were no acceptable levels clearly defined of “system down” at go-live, the guarantee of 100% up-time was also not clearly articulated or expected.
• It would take more money to manage this risk. We could not compromise on time or quality, so money was the only variable we had to work with.
• His nod indicated that if accepted, the increased dollars associated with the change request would be added to the budget so that our metric of on-time, on-budget would not be negatively affected.
In the end, we hired additional vendor resources to deploy the hardware more rapidly and worked with each of the store managers to facilitate the install during business hours. We re-directed a project manager to focus solely on managing the vendor and store coordination for the next 5 weeks. We increased the hardware budget and replaced the boxes entirely for a few key stores. On go-live day of that year, to everyone’s surprise, Sears sales volume exceeded that of the year prior. The implementation had no impact what-so-ever on sales.
The key take-away of that experience was not only about identifying and managing risk through change requests.
The other underlying message is that if you really want to manage risk effectively, blame cannot be part of the equation. Accountability was crucial, but nobody was blamed for the risk existing or the change request being required. In fact, the team was applauded for thorough performance testing and risk identification in the first place. If we had not been so cognizant of the lesson learned in the first go-live, we likely would not have done as thorough of modeling during our performance testing and we may have gotten caught in a vortex of blame once again. Instead we all focused on the path to improving the probability of our success.
This shift in mindset is what takes a project across the finish line to successful solution delivery. At critical junctures, the team must know where we stand, yet be focused almost entirely on moving forward rather than why we are at this point to begin with. The time for reflection is at lessons learned. In this case, we applied our lessons learned from the first implementation to delivery success for the next implementation.