Saturday, 30 July 2011

None so blind...

I started to write this item a couple of weeks ago – but didn’t finish it at the time. I was highly annoyed and wanted to think about things before posting.

We actually started work on the SAP rollout at our overseas site almost 2 years year ago. However, nothing was ever completed, even in the development system and eventually, the project stopped before being started again with a new set of consultants in the middle of last year. The plans then were to go live at the end of October last year – as the previous consultants had done quite a bit of work, it was felt that this was achievable.

However, it quickly became clear that most of what had been done was of no value. Essentially, it all had had to go right back to the very beginning. The SI were still trying to say that October 2010 was achievable, but most people on our project team could see that was never going to be possible. A date was set for the end of July this year – most people agreed that this could be done if all went well.

As it happens, in April of this year, the consultants still had not finished working on the development system. We also had issues because almost no testing had been carried out, and absolutely no training had been done either. Our project team then went to the site to do some work, and as I’ve indicated, we managed to get a great deal done. So much in fact that we all thought it might still be possible to go-live in July. There is still major work being done by the consultants, but from experience, it seems to be normal for them to still be working on items right up to (and even after) the go-live date.

But there has been a major problem with getting data loaded. First getting data out of their existing system has proven to considerably more difficult than anyone believed would be the case. Cleansing the data was also a bit of an issue – but the biggest problem has been getting it matched to the relevant data load into SAP. I had been told by the SI that about 80 - 85% of the data load had been completed into the development system – but during our test phase it became very clear that it was really closer to 10 - 15%.

Since then, some data has also been loaded into the test system – but no more than about 30 – 40%. Some training has been done with key users, but none with the rest of the end users. I had to go over there again a few weeks ago, and they were trying to do some testing work in the test system. Most of this failed simply because there was insufficient data to allow more than few test scripts to be used – it took them as much as half a day to complete a single test item.

As it became clear that July was slipping away, a decision was made to move the go-live date back yet again. This makes sense, and I have no issue with that other than once again, we are committing many resources to the project than was originally planned. But what has outraged me is a decision that has been made following further discussions with the SI.

The director of the SI suggested to our senior managers that we did not need to complete the data load process in either of the development or test systems, but simply go straight to the production system and load the data without having to do any testing. The argument was that we had done a few items so that proved the process worked and therefore any further data added to the same process would also work. He also argued that if we needed to have data in the other systems, it would be easier to do a system copy after the go-live.

I tried to point out that this view was completely wrong – we have seen repeatedly that people make mistakes when preparing the data. In some cases, the mistakes are so significant that the data will simply not go into the database without considerable manipulation by the people carrying out the work. To trust that all of the data will load correctly just because you have loaded a few lines is at best, wishful thinking in my opinion. As far as I am concerned, completing the data load process in the development and test systems should then save time and effort when loading into the production system.

But there we are – I can do no more than point out the issues and then do my best to see if I can fix the problems when they occur. But I must admit that I am now really beginning to feel that I can do no more on this project. I feel so frustrated all of the time – all I hear is that everything is going well, when it is abundantly clear that things are far from satisfactory.

I have no problem people trying to maintain a positive attitude – I can see that sometimes this is a necessity. When something requires such a major business transformation as SAP, it is essential to keep everyone motivated. But what worries me is that we are seeing “ostrich management” – burying their heads in the sand in the hope that if they don’t see the problem, it will just go away. I can’t believe that will be any good for us, now or in the future, on this project or any other.


  1. ...but an incredibly common problem

  2. I used to do projects of this nature for a decentralized manufacturing entity. I honestly can't see it working without core technology managers and implementers from the home location being on-site at least 1 week in 3 (I generally was on-site every other week for 5 full days), and a project driver with executive management authority on-site at least once every 6 weeks. Without executive and real technical management (not "project management") there is generally neither the incentive nor the knowledge at the division to get the job done.

    I've posted before about the use of outside consultants, but even if it is believed necessary to use them without strong knowledgeable internal management (again, not "project management") what you have described is exactly what will happen every time.

    I feel your pain, but I have to wonder what your organizations executive management team is thinking in making these decisions. I have twice been called into midsized organizations that were within a few weeks of collapsing with bankruptcy to follow due to failed business system implementations; one hopes your executives understand this.