The first ninety days after buying software

The decision everyone agonises over is which system to buy. The decision that determines the outcome is what happens in the twelve weeks afterwards, and it gets a fraction of the attention.

Weeks one to three: get one thing completely right

The instinct is to switch everything on. Resist it. Pick the single process that hurts most — usually billing, sometimes stock receiving — and make that one process work properly, end to end, with the real people who do it.

A business running one process well on a new system has momentum. A business running six processes badly has a rebellion.

Weeks four to eight: fix the data, not the software

By now the complaints will have started, and most of them will sound like software problems. A good proportion will be data problems wearing a costume: duplicate customers, items with two names, opening balances that were never right in the old system either.

Migrating bad data and then blaming the new system for surfacing it is the most common failure mode we see. The old system hid it. That is not the same as it not existing.

Weeks nine to twelve: check who has quietly stopped

Someone always reverts. Not loudly — they keep a parallel notebook, or they enter things in a batch at the end of the week, or they get a colleague to do it. Find that person and ask why, without an audience.

The answer is almost always specific and fixable: a step that takes too long, a field they do not have at the moment they need it, a screen that assumes a desk when they are standing. It is also, almost always, information nobody else was going to tell you.

What not to do

• Do not run the old system in parallel "just in case" for more than a few weeks. Two systems means neither is trusted.
• Do not train everyone once, at the start, in a room. Train the people who use each part, at their station, when they need it.
• Do not judge the decision at week two. Everything is worse at week two.