A common strategy we’ll often see from Architecture leaders is one that stipulates the use of best-of-breed software to meet the business goals. Selecting such an approach is often an easy choice to make in the early days of a project where the functional requirements are at the forefront of the mind, the budgets haven’t been fully defined and snazzy demonstrations are ten-a-penny.
SaaS products especially offer easy to digest pricing models, remove burden from internal teams and appeal to business leaders who’d much rather focus on the business issue than the technical challenges. It’s often these people who have the last word in the direction the project takes.
And, to an extent, I have no issue with this approach. Fixing the business problem should guide much of our thinking – gone are the days when we select the product first and figure out the benefit of using it later. Offering business benefit and being able to attribute a quantifiable cost is no bad thing and reducing friction by aligning it to a particular thread of discussion in the boardroom certainly helps.
The problem is that the real cost occurs after the licence has been purchased. You’ve gone best of breed and, as such, the next dozen or so months will be spent planning the integration effort.
The reality, as is so often the case is that the happy, warm, comfortable world that existed in the product demo where all the data is together, third party applications can be integrated using out of the box templates and such bears absolutely no resemblance to the client’s infrastructure. With disparate systems galore, each with varying levels of integration capabilities and each holding a piece of the puzzle the chances are your life is about to get immensely complex.
Carefully considering where your data is located and understanding the processes that wrap it are critical. As tempting as it may be, a weekly load of data from your current systems into your fancy new platform is probably an accident waiting to happen.
It’s not just the integration capabilities that can slow progress, contractual obligations for support and maintenance arrangements with your third party for that legacy HR system will get more complex, perhaps triggering the need for that painful and long overdue upgrade. Each element of remediation needing deep-dive analysis, support from the in-house teams and approval from the change advisory boards. Multiply this out by even half a dozen systems and that freshly purchased licence is officially sitting there, doing nothing.
A fool’s errand would be run to an integration product, the claims for which will often give the impression that this is all completely achievable and, to an point that’s true – but it’s more accurate to say that you’re probably just purchasing a toolbox to facilitate the integrations, rather than a one-size-fits-all solution and dealing with the old purchasing platform with an ancient, unstable and feature limited SOAP interface hasn’t gotten any easier.
Detailed analysis of the existing systems, especially those which appear on your longer-term roadmap is often a better place to start. It isn’t as sexy as plucking companies out of the magic quadrant and arranging fancy off-site lunches but it’s honest work.