Choosing business software without letting the software choose your business

The demonstration is designed to make you buy. Writing down what has to be true when the system is working — before anybody shows you anything — is the only reliable defence, and it takes about half a day.

Why "we bought something and nobody uses it" is so common

Because the requirement was written after the demonstration rather than before it. Once you have watched a good product being shown by somebody who knows it well, your idea of what you need has been rearranged into the shape of what you were shown. That is not a criticism of vendors — a demonstration is supposed to be persuasive — but it means the order of events decides the outcome, and most businesses get the order wrong.

The second reason is that the business bought a system to replace a mess, without first working out what the mess was doing. Informal systems are load-bearing. The spreadsheet everybody complains about is also the only place three exceptions are handled, and a new system that cannot handle them does not remove the spreadsheet — it adds itself alongside it.

The half day, before anybody sees a product

Sit down with the two or three people who would actually use it and write four lists.

• The jobs. What has to be possible. Not features — jobs. "Somebody at the counter can see whether we can promise Thursday." "The invoice goes out the day the job is signed off, without anybody retyping anything."
• The Monday questions. What you want to be able to answer on a Monday morning without asking a person. Be specific and be short: five is plenty.
• The exceptions. Every awkward thing the current informal system quietly handles. This is the list nobody writes and it is the one that decides whether the new system is used.
• What must move across. What genuinely has to come from the old world, and what everybody says has to and does not. Data migration is where these projects die, and half of what gets migrated is never looked at again.

That document is the requirement. It is one or two sides. Nothing on it is a product name.

Then, and only then, look at products

Score each candidate against your own lists rather than against each other, and do the scoring in writing before the next demonstration, because after three of them they blur. Ask each vendor to show you your exceptions specifically — not the standard demonstration, your third and fourth awkward cases — and watch what happens. A vendor who says "that is a configuration question, let me show you" is a good sign. A vendor who explains why you should not want to do that is telling you something true about the product.

The four questions nobody asks and everybody regrets

1. What does it cost in year three?

   Not the first-year price. Licences per person as you grow, the modules that are extra, the support tier you will end up on, and what happens to the price at renewal. Ask for it in writing.

2. How do we get our data out?

   In a form somebody else could load. Ask to be shown an export, not told about one. A business that cannot leave a system is a business whose supplier sets the price.

3. Who inside our business will own this?

   By name, before you buy. Not the owner and not the most junior person. A system with no internal owner reverts to spreadsheets within a year, and the implementation is what makes the difference.

4. What happens to the thing it replaces?

   Decide the date it is switched off, and who is allowed to say no. Running both is the most expensive outcome available and it is the default one.

Implementation is most of the value and nearly all of the risk

The gap between two shortlisted products is usually small. The gap between a good and a bad implementation of either is enormous. Budget for the implementation as if it were the purchase, protect the time of whoever inside the business is going to own it, and expect the first eight weeks after go-live to be worse than the old system rather than better. That is normal, it is not a sign of having chosen wrong, and knowing it in advance is what stops businesses reverting in week three.

A note on who is telling you this

We do not sell software and we take no commission, referral fee, rebate or reseller margin from any vendor. That is not a boast, it is the only thing that makes advice about which system to buy worth having, and the reason it is written on several pages of this website is that you should be asking it of anybody who advises you on this — including, especially, anybody who also implements.

We take nothing from any software vendor. We do not sell software and we take no commission, referral fee, rebate or reseller margin from any vendor. It is the only thing that makes advice about which system to buy worth paying for.