There Are Still Silos in the ERP Prairies
ERP integrated systems are designed to destroy silos.
The integration of all data in a single database, with a similar approach to access it, goes a long way to eliminate the incompatibilities between department systems that were commonplace in the past. The batch transfer of information – sometimes even manual transcription – is no longer needed, with the consequent reduction of errors and the advantage of real time visibility of the whole business.
This is what the brochure says; but we are not there yet, probably never will.
Problems remain, just deeper into the network where they are less visible. In addition, the idea that information silos are based exclusively on software compatibility is misleading; as people, processes and policies also contribute to create them.
No standard way to link data: The information may be in the same system but there may not be a standard way link, sort, filter, calculate and present it the way the user wants. As it is impractical to cater for all possible information needs; sophisticated ERP systems offer flexible tools to generate operational reports (e.g. SAP ABAP queries). These tools allow for unique data crunching that may only make sense in the context of a distinctive company template. In my experience, these tailored reports can become addictive – the user wanting more and more – due to the exhilarated feeling of increased productivity they provide. Unfortunately, for reasons of performance and authorization some companies have an absolute ban on user generated operational queries.
Excel is the new silo: If the report required to do the job is not available; the user will quickly realize that Excel (or similar) is the solution. Data will be copy-pasted, processed, filtered, sorted and visualized there; bringing back all the usual information silos problems: translation errors, no-real-time availability and a general pain to get data across. Don’t get me wrong, Excel is an excellent tool to analyse and graph information – even SAP’s BEx Analyzer uses it – and there are ways to integrate it better; but it is still a silo.
Performance issues: Data processing is expensive in terms of performance and in a shared system not only the user waits; it also slows down everybody else a bit. That is why some companies steer away from non-standard solutions. But performance cannot be seen only in machine terms. A few more seconds processing a customized report can make the difference between solving a real world problem in minutes rather than hours/days. Why make the user work harder than it needs to? There is no perfect solution for this performance trade-off between system and job performance. My preferred approach is to coach the user on those flexible tools to explore possibilities. Then, after a stable solution is found, have the IT team audit it to improve its performance and consistency.
One popular solution to improve performance is to move pre-digested data into a standalone system for flexible analysis there. This is the case of SAP’s Business Warehouse/Intelligence (BW/BI) solutions. But the reporting system becomes a silo in itself, with typical silo problems; like having to wait for the nightly extraction run to have the most recent information available.
Not as seamless as advertised: Some integrated systems can be a combination of detached systems where the transcription of the data still happens but is less visible than before. The data transfer routines and problems are no longer managed by the users but by an invisible – usually remote – IT team. That is the case of the data extractions for the Business Intelligence solution or the interface between SAP’s ECC and APO – different systems for historical reasons – infamously known as “the CIF”.
Need to know basis: The fact that integrated systems allow you to see everything doesn’t mean that the bosses want everyone to see all. It is normal that authorization profiles, which determine what you see or not, are used to restrict access to classified or sensitive information. But knowledge is power and some people are reluctant to share power even if it benefits the business. They can use policies – reflected in system authorization checks – to recreate silos and the status they provide.
Restrictions can also be excessive in their initial version. In the heat of the project is better to err on the conservative side, right? This shouldn’t be a big problem, if it not for the common reluctance and red tape to review authorizations after the system gets going.
Knowing how to access the information: The information you need may be a couple of clicks away, but it doesn’t help if you don’t know where to look. People may be reluctant to play with a system where their actions have a companywide impact. To break this wall between useful data and its consumer the user has to learn to handle the system in a flexible and confident way. Time restrictions during the implementation favour a step-by-step-recipe training style, based in lineal scenarios. It is better to train for flexible use after the system is live.
Knowing what to do with the information: From a more philosophical point of view, having access to the data, and the tools to analyse it, is of no use if you cannot understand the result, or if the result is not relevant to what you do, or if you are not allowed to act on the acquired knowledge. The value of the potential access to information is hard to predict, in order to decide what to do. A parallel extreme case can be seen on the new boom of “Big Data” initiatives. Expensive implementations are needed before knowing if you can do something useful with the results.
Silos will always be with us, but we have some demolition charges to get as close as possible to the levelled ERP landscape advertised in the brochure.