back to article Global pharma firm GSK opens Pandora's Box of its SAP system to find 28,000 variations on a process

The received wisdom in ERP is that businesses design their processes in the system for everyone to follow. Except they don't. Ten years after implementing SAP ECC worldwide, GSK considered its ERP unification something of a success. But a close analysis of how business processes were actually running revealed that SAP's …

  1. Aristotles slow and dimwitted horse

    It's always nice...

    It's always nice to read about a success story. Having led part of a SAP centralisation programme for a global insurer the other thing not to underestimate is backyard politics.

    1. I code for the bacon

      Re: It's always nice...

      Well, if they say they have 28,000 different variations for the process of running a sales order... surely they gave backyard politics their room. Hats off to anyone shouldering the responsibility of ERP unification on a global scale.

  2. Anonymous Coward
    Anonymous Coward

    In today's unprecedented news

    "a close analysis of how business processes were actually running revealed that SAP's system was not exactly, well, unified."

    That's not unprecedented, that's Standard Application Practice.

  3. Deft

    Nice extras

    Can't decide whether it's cool or scary, you have to develop software on top of the system to figure out how the hell everyone is using it!

    I suppose if you unleash the average person on Excel, god knows what crazy buttons they will press.

  4. Mike 137 Silver badge

    28,000 variations on a process

    Not remotely surprised. The degree to which management is generally out of touch with what actually happens in the front line is dazzling.

    On one assignment I found that what managers thought was "PCI DSS compliance" consisted of filling in the quarterly "progress towards compliance" report without making any investigation - i.e. making it up in a hurry the week before submission, and this had been going on for almost a decade without anyone challenging it (either locally or at the acquiring bank).

    On another assignment I found a heap of "policies" including multiple variants of notionally the same documents, some of which included reference to withdrawn standards. However they had all been "reviewed" to schedule - at least going by updates to the version listing on their front pages.

    Which is why, when I do BPA I always start in the front line - they know what actually happens.

    Incidentally, process mining is not a new rocket science it just BPA done right.

  5. Anonymous Coward
    Anonymous Coward

    Ah.......consultants again...........

    ......a while ago, it was common for organisations to hire consultants (you know, the usual suspects, Deloitte, PW, EY....) to charge your organisation a fortune to do the configuration and implementation of SAP (...or PeopleSoft...pick your ERP flavour).


    And here we are, much later, getting consultants (guess who) to analyse the implementation and find out just how incoherent the current implementation might be. Shock, horror!!! This needs another fortune to sort out.


    So here's my problem with this (recurring) horror story.......what were the senior managers doing while all this was going on?


    Could it possibly be that certain senior people were flitting from an ERP vendor to a consultancy, then to a client, then back to the ERP vendor......rinse and repeat.....with ever rising remuneration for "expertise"? Surely not........

  6. Norman Nescio Silver badge

    There's an xkcd for standards which applies to 'standard' processes as well, although multiplying up to 28,000 of the things is impressive. Anyone who looks at a standard tends to think of a variation they need (want) and following that, it is a short step to multiplication of 'solutions'. Saturday Morning Breakfast Cereal makes essentially the same point about 'definitions' as applied to Social Sciences, but it is, in my experience, a generally applicable point.

    As for BPA analysis, it suffers, like code from premature optimisation. As Donald Knuth famously said in his 1974 ACM Turing Lecture:

    The real problem is that programmers have spent far too much time worrying about efficiency in the wrong places and at the wrong times; premature optimization is the root of all evil (or at least most of it) in programming.”

    Donald Knuth, Computer Programming as an Art (Communications of the ACM December 1974 Volume 17 Number 12 1974)

    Removing flexibility from a process is a premature optimisation. The art of a good analysis is to leave the process flexible enough to cope with its anticipated use-cases (including the rare ones, and reasonable variations), while being standard enough to benefit from efficiencies of scale. If your process is too difficult to use, or can't be used by the users for all the things they need to do, it has failed. There's a great deal more to do than just automating the most frequently used steps and calling it a day. Sadly, that level of facile 'analysis' happens all too often.


    1. Anonymous Coward
      Anonymous Coward

      Process Models and Optimisation


      But a better question might be: Who's premature optimisation?


      Once upon a time, a long time ago, I got to assess three of the "standard process models" being hyped by three suppliers:

      - SAP

      - Baan

      - Deloitte

      The idea behind the SAP and Baan "standard process models" was that instead of configuring the (huge complicated) configuration databases underneath the package, the consultants would review and adjust the process model, and some spiffy software would run against the adjusted process model and spit out the package configuration tables to match.

      The Deloitte idea was different. Basically they thought it took too long (really? for consultants?) to build a client specific process model from scratch, so they would turn up and say something like: "Here's a best practice process to use as a starting point for our spiffy new reengineering effort".

      Now in all three cases, the process models used as a starting point were created by either:

      - software companies pushing their ERP package, or....

      - a consultancy promising spiffy "best practices" to the client

      In neither case does the client get to say where to start!!!! And the "premature optimisation" has already begun....done by self-proclaimed experts.

      Of course, a successful company can't possibly know much about its own successful processes, can it? SAP or Baan or Deloitte (obviously) have much greater "expertise".

      ......and the beat goes on...................

    2. This post has been deleted by its author

  7. Anonymous Coward

    Only 28000?

    Every layer of management wants their own configuration.

    Every layer of management has their own favorite consultants on contract.

    All the consultants go their separate ways to configure it.

    For a company as big as GSK, they're lucky some of the management layers decided to work with what they were given or the results would be even higher.

  8. sketharaman

    Every User ID Is A Variation?

    100,000 users, out of which only a subset will enter sales orders. I can believe the "28,000 variations in sales order" figure only if every sales order pathway having a different UserID (or SKU or CustomerID) is treated as a variation.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like