Technology
You modernize everyone else's operation. When did you last look at your own?
IT firms are the hardest industry on this list to sell to, and the most interesting to audit. You know what an integration costs. You know what an agent can and cannot do. You will check the arithmetic. Good. This page is written on the assumption that you will.
You sell this work. You defer it on yourself.
Every trade on this list has the same problem in a different costume: work that a system should be doing is being done by a person, and nobody has counted it. What makes technology firms different is that you already know that. You say it to clients weekly. The gap is not knowledge, it is that the work you sell is the work you defer, because your own operation is the only account that never files a ticket.
So the audit is not here to explain automation to you. It is here to do the thing you do not have a spare senior engineer to do: go through nineteen operational domains, count what is being carried by hand, and put a number against it. The findings tend to be uncomfortable in a specific way: most of them are things you would have flagged in a client's environment inside an hour.
There is a second reason, and it is the commercial one. The largest untapped AI opportunity in your book is not your own back office. It is sitting inside your clients' estates, in the line-of-business applications everyone works around. Most of those do not need replacing. They need an interface. If you are not the firm that offers that, someone else will be.
None of this is a crisis. That is the problem.
Every item below is survivable on its own. Together they are the reason the week is full and the forecast is a guess.
- 01The queue at 8am, and the third instance this quarter of a ticket that resolves the same way every time.
- 02A senior engineer answering a question only they can answer, because the runbook was never written down.
- 03A statement of work that has been nearly ready for nine days.
- 04A renewal discovered because a card declined, not because anything warned you.
- 05An onboarding checklist that exists, mostly, in a folder someone maintains out of goodwill.
- 06A client's 2009 line-of-business application that everybody has quietly agreed to work around.
The audit domains, read through this trade.
Each of these is one of the nineteen operational domains the audit examines, stated in the terms this industry actually uses.
The repetitive third of the service desk
Triage, categorization, routing, the first reply, the status update nobody has time to send. Industry reporting puts the repetitive volume at 30–50% of a technician's day. Automating the triage layer is not the same as deflecting tickets: it is deciding, in the first thirty seconds, what kind of ticket this is and who should hold it.
Humans acting as middleware
The PSA does not talk to the RMM, the RMM does not talk to billing, and somebody reconciles the three by hand on a Friday. Every one of those handoffs is a documented, boring, automatable interface, and every one of them is currently a person with a second screen open.
The knowledge that only exists in senior heads
An agent can only be as good as what it can read. If your procedures, escalation paths, standard builds and past resolutions are not in a form a machine can retrieve, no amount of AI will help, and neither will your next hire. This is the groundwork domain, and in technology firms it is almost always the one holding the rest back.
The statement of work that ages in a drafts folder
Scoping is the bottleneck in most IT firms, and the cost is invisible because it is measured in deals that closed slowly rather than deals that were lost. A proposal drafted from the record, priced from live rates, sent with signature attached and chased without a human remembering to chase it, moves the whole pipeline forward at once.
Feast, famine, and no forecast
Project work makes a lumpy pipeline, and a lumpy pipeline makes hiring a guess. The problem is rarely the CRM: it is that stage changes are entered after the fact, so the forecast reflects what somebody remembered rather than what actually happened.
Licenses, true-ups and the margin nobody watches
Seat counts drift, renewals arrive unannounced, and a per-client margin figure exists only in aggregate at quarter end. This is the domain where an audit most often pays for itself before anything is built, because the leak is arithmetic rather than strategy.
Buyers now ask an assistant first
Referral is still the strongest channel in this trade and it is not going anywhere. But the first search a new buyer runs is increasingly a question put to a model, and the firms that appear in that answer are the ones whose sites are structured to be read by one. Referrals are compounding; being absent from the machine answer is compounding too, in the other direction.
A site a technical buyer bounces off
Stock imagery, a services grid and a contact form is the house style of this trade, and it converts the way you would expect. The buyer evaluating you is technical: they want to see how you think, what you have actually run, and what happens in the first hour of an incident. Almost none of that survives contact with a template.
You hold other people's data and prove it by hand
Evidence collection, access reviews, offboarding proof. If you carry a framework, you already know the audit-prep scramble. Continuous evidence is the difference between a compliance posture and a compliance fortnight.
Churn shows up at renewal, not before
The signal was there for months: falling ticket engagement, unread reports, a quarterly review that slipped twice. A client-facing surface that shows work delivered and value returned is retention infrastructure, not decoration.
Most legacy applications do not need replacing. They need an interface.
Gartner and PwC put 60–80% of IT budgets into maintaining legacy systems, and 73% of companies name legacy as their single biggest transformation blocker. The usual recommendation is a rewrite, and the usual outcome is that the rewrite is never funded, so the system stays, the workarounds stay, and everyone agrees to look at it next year.
There is a cheaper move. Put an API in front of the old application. Give an agent the ability to read and write through it. The twenty-year-old system carries on doing the one thing it does well, and everything modern can finally talk to it. No migration, no retraining, no weekend.
This is a domain in the audit rather than a footnote to it, and for technology firms it points in two directions at once: at your own estate, and at every client estate you already have access to.
Third-party research, not our own claims.
We have no case studies to show you and will not invent any. These are published figures for the trade, and they are the reason the audit asks what it asks.
of a technician's day spent on repetitive volume
MSP industry reporting, 2026
expected ticket-volume reduction from service desk automation
MSP industry trends, 2026
of managed service providers plan to increase AI investment
MSP industry trends, 2026
of IT budgets consumed by legacy maintenance
Gartner / PwC, 2026
of companies name legacy systems their biggest transformation blocker
Legacy modernization research, 2026
Start with the score.
You can run the FusionScore in about four minutes and see where your own operation sits. If the result tells you nothing you did not already know, that is a legitimate outcome and costs you nothing. If it tells you where the hours are going, that is the point of it.