The Myth of Turning a Knob

July 4, 2008 by roland.

I’ve been exposed to a new marketing pitch for ERP packages lately. Now many claim you can change your business “simply by turning a knob” in their software settings. I’ve watched management saddled with an older system get a gleam in their eye thinking about all of the programmers they wouldn’t need anymore. In short, I’ve watched a pink elephant dance on the conference room table and been the only one in the room who pointed out the elephant was pink.

This marketing scheme (which seems to be used by most of the big packages today) fits in well with the kaka Gartner has been spewing recently. In case you haven’t seen the spew, the gist of it is “For companies to get ahead they must marginalize IT”. How can you possibly marginalize IT? Quite simply, you need a package which lets you change your business direction simply by turning a knob. It all fits so well together, it couldn’t possibly be a coordinated marketing effort could it?

Let’s first take a look at the knob myth. Let us assume for the sake of this discussion that each knob has to be turned in units of a full degree, not partial degrees like an old fashioned radio tuner. That means each knob has 360 possible positions. Software vendors have learned that most of upper management is the same group of people who walk into an electronics shop to buy a stereo with “big knobs and lots of lights”, so they have lots of colors on the screens and an array of big knobs to turn.

Just how well do you think this has been tested? Assuming the math is combinatorial instead of a permutation, each knob doesn’t add another 360 possibilities, it adds a set of 360 instances of all preceding possibilities. That means 360 * 360 * 360 … until you have factored in all of the knobs. There is no way in Hell they ever run a full regression test for that. It doesn’t take too many knobs before you it takes you 4 complete months of 24×7 operation with absolutely no problems just to run the regression test. So, what you are really buying is a package where a few of the combinations have been tested and the rest are a poke of hope.

What happens when the new business model you want to implement isn’t in the set of combinations which have been tested? That’s right, you wait months for the vendor to get around to solving your problem, OR, you have to bring in people who have expertise in this shiny new package. That’ll be cheap. Yes, you and every other company using an untested combination of those knob turns will be trying to bring in people with 3-5 years of experience on a package which has been shipping for a year. I’m sure you’ll get a deal.

How did we get here? Management never listened to their developers who said that legacy ERP package they had running the cash cow needed to be ported to a relational database to extend its life and make it easier to modify. They wanted a “quick win”. They wanted to turn a knob. They bought the shiny new thing, then found out they had nobody to do the data conversion because they had already marginalized their IT. Even if they did the conversion they couldn’t generate the exact same output because it just didn’t fit into the new system. So, for at least 7 years, they had to have two sets of staff. One set to keep the old system running so past invoices could be looked up, and one set to support the new system.

Bet you can guess what will happen to them next. Go on, take a guess.

Management gets an opportunity to do some profitable business which is just a tiny bit different from the old business that couldn’t be converted into the shiny new thing. They try turning a bunch of knobs, but it just doesn’t work. They try hiring consultants and paying the vendor for modifications, but it just doesn’t fit, or it will take a year to fit and they need to pull the trigger in months. Yes, they end up tweaking the old system and running this new profitable venture through it “just until the shiny new thing can handle it”.

We have all been here before. The spec the vendor started with was the spec you had when you started. In the year it took them to deliver, enhancements went in every week. Their deliverable isn’t even close. Management can’t turn a knob and make it fit. Vendor goes back into development mode and the handful of developers shepherding the old system into the grave get another round of enhancements from the business unit it is supporting. Management cannot unplug the shiny new thing because it did work for some lines of business they are currently in. They now have two ERP systems which can’t really talk to each other and can’t be unplugged. The really good people for the Heritage system have either retired or simply left because they were told there careers were coming to an end. Good margin.

Laugh all you want. I’ve been to companies with no less than 3 ERP systems running different business units, each was the best thing Gartner was selling in its day. Each have now been called Heritage to justify bringing in another system, and yes, they are looking at a fourth system.

What is that psychology term for a person continually repeating the same action which has the same outcome every time, but who expects it will be different “this time”?

Posted in Information Technology | Print | No Comments »