How to Tell Whether Your IT Provider Is Any Good

Fast ticket closure is the main one. It reads as efficiency and it has an innocent explanation and a bad one. The innocent version is a well run desk fixing simple things quickly. The bad version is a desk closing tickets the moment the symptom stops, which means the same fault comes back next week under a new ticket number, and the average closure time looks better every time it happens.

A quiet month is the second. Low ticket volume can mean everything is working. It can also mean people have given up on the process and started ringing one helpful engineer directly, or have simply stopped reporting things because nothing came of it last time. Those two situations produce identical reports and completely different businesses.

Then the all green report, the friendly account manager, and the claim that there has never been a serious incident. None of those are bad. None of them are evidence of anything either. A monthly report is written by the party being assessed, a good account manager is a sales skill rather than an engineering one, and never having had a serious incident can mean good work or can mean luck, and from the outside they look the same.

Ask for two things that nobody puts on a report voluntarily. How many tickets were reopened, and how many were the same underlying cause as something already seen. Those two say more about the quality of the work than every other measure combined, because they are the ones that get worse when corners are cut.

We are not going to tell you what a good level looks like, because it depends entirely on the age and state of your setup, and any figure quoted at you without seeing your systems is a figure somebody invented. What matters is the direction and whether anybody is watching it. A provider who can produce the list, explain what is on it, and name the three they intend to eliminate this quarter is doing the job.

If the answer is that they do not track it, that is not automatically damning. Plenty of good desks have never been asked. Ask them to start, agree a date to look at it together, and see what comes back. What they do with the request tells you more than the original answer did.

This is the single best test and it takes no preparation. Think back over the last few months. How many times did your provider tell you about something you did not already know was wrong.

A team that is actually looking at your setup finds things. A disk approaching full. A backup that has been failing on one server since a change in the spring. A licence renewing next month that nobody uses. Software going out of support. A member of staff still able to reach a system they left behind two roles ago. Any of those is normal to find. Finding none of them, ever, over a year, is not.

The reverse test is just as useful. If every single thing they bring you is something to buy from them, you have learned what the relationship is for. A provider working in your interest will sometimes recommend spending nothing, or spending it with somebody else.

Ask to see the documentation for your own systems. Not a sample, yours. You are entitled to it, it should exist, and how the request is received tells you a great deal before you have read a word.

Then read some of it and ask whether you could hand it to a competent engineer who had never seen your business. Does it say why things are set up as they are, or only what the settings currently say. Does it name the suppliers and the account numbers. Does it cover the odd one, the machine in the corner that everybody knows about and nobody has written down.

Then test it for real without announcing it as a test. Raise something moderately fiddly during the week your usual engineer is on leave. If the answer comes back competently from somebody else, the knowledge is in the business. If it waits until Monday, or the replacement rings the regular engineer at home, the knowledge is in one person, and you are exposed to their next job offer.

Ask for a restore. Not a report saying the backup succeeded, an actual file and an actual system recovered onto something, with you watching, and a note of how long it took. This is the test that most often surprises everybody involved, including the provider.

Ask for the current list of every account with administrator access to your systems, including their own staff and any former staff. A good provider produces it in a day and it is shorter than you expect. A weak one takes a fortnight and it is longer than you expect, with names on it nobody recognises.

Ask what would happen if you gave notice tomorrow. What you would receive, in what form, how quickly, what is licensed to them rather than to you, and what stops working the day the contract ends. Ask it as a planning question and note whether the answer is specific or vague.

And log a genuinely low priority ticket, something that can honestly wait. Watch what happens to it. Anybody can handle an emergency well. How a business treats the unimportant work is a much better guide to how it operates on an ordinary Tuesday.

Do they ever say no to you. A provider who agrees to everything is either extraordinarily capable or is agreeing now and disappointing you later. Somebody telling you that what you have asked for is a bad idea, and why, is a good sign even when it is annoying.

Do they tell you when they have got something wrong, before you find out. Every provider makes mistakes, including us. The difference between one you can trust and one you cannot is entirely in what happens in the hour after the mistake is noticed.

Do they know your calendar without being reminded. Do they warn you before a renewal rather than after it has automatically charged. Do you ever speak to the person doing the work, or only to somebody who relays it. Are the same faces still on your account, or has it turned over twice this year without anybody mentioning it.

Before judging the answer, check what you asked for. Some of the complaints businesses have about their provider are complaints about a contract they chose.

If you bought the cheapest tier and are measuring it against what the expensive one promises, the provider is doing what you agreed. If they have recommended a fix three times and it has not been funded, the recurring fault is not their failure. If nobody will give them the access they need to do the job properly, they will keep working around it and you will keep paying for the workaround. And if half your business bypasses the process to ring a favourite engineer, the reporting will always be wrong and the desk will always look worse than it is.

None of that excuses genuinely poor work. It does mean the honest first question is whether you have bought the wrong thing rather than bought from the wrong people, and that question is much cheaper to answer than a change of supplier.

Ask who exactly will pick up a ticket raised at four on a Thursday afternoon, and what happens to it at half past five. Ask for their list of the things about your setup that currently worry them, and be suspicious of an empty list. Ask when they last told a client not to buy something.

Ask what happens if you leave. What you receive, what is yours already, and what is licensed to them. On our side the answer is that you own the code and the accounts, and that stays true on the day you decide to go elsewhere.

Ask to speak to the engineer who would actually do the work rather than to the person selling it, and notice whether that is easy to arrange. Ask what they have got wrong for another client recently and what changed afterwards. A provider who cannot name a single mistake is either new or is managing you.

Then ask us the same things. It is a fair test and we would rather be measured on it than on a report that we wrote about ourselves.