From Service Catalog to Operational Data: Why the MSP Pricing Model Has to Change in 2026
UK SMEs are being pitched AI-enhanced IT services at a premium, but buyers expect AI to lower costs, not raise them. Here's how to spot an unjustified 'AI tax' and what fair, outcome-based pricing actually looks like.
A Nottingham manufacturing firm we spoke to recently had a renewal quote land on the finance director's desk with a new line on it: 'AI Operations Premium — £840/month.' No explanation of what it did. No measurable change to response times, ticket volumes, or anything the business could actually feel. Just a higher number and a confident sales pitch about 'AI-driven service delivery.'
The finance director's reaction was the right one: "What am I getting for this that I wasn't getting last year?" The MSP couldn't answer. The renewal didn't happen.
That conversation is going to repeat itself thousands of times across the UK in 2026, because two things are colliding. MSPs are spending real money on AI tooling and want to recover it. Customers, meanwhile, have read the same headlines about AI making everything faster and cheaper — and they expect to share in those savings, not subsidise them. The old per-device, per-user service catalogue can't hold both of those truths at once. Something has to give.
The squeeze nobody at the MSP wants to talk about
For twenty years the managed services model has been beautifully simple. Count the devices. Count the users. Multiply by a monthly rate. Add a few line items for backup, email security, and a help desk. Bill it. The whole industry runs on this, and it works because it's predictable for both sides.
AI breaks the maths in an awkward way. The tools that genuinely improve an MSP's operations — automated ticket triage, anomaly detection, scripted remediation, AI-assisted documentation — all reduce the number of human hours needed per device. That's the point of them. But the per-device price was calibrated around those human hours. If you automate 30% of the labour and keep charging the same per-device rate, your margin goes up. Good for you.
The problem is your customer knows this. They've heard 'AI will cut our costs' so many times that when their IT bill goes up and the work clearly involves more automation, the story doesn't add up. They suspect, correctly, that they're being asked to pay twice: once for the tool, and once for the labour the tool replaced.
So MSPs face a genuine bind. Absorb the AI tooling cost and watch margins erode. Or pass it on as a premium and watch customers walk at renewal. Neither is a strategy.
Why the 'AI tax' fails on contact
Let's be blunt about what an 'AI tax' actually is. It's a price increase justified by an input rather than an outcome. "We now use AI, therefore you pay more" is the same logic as "we bought a faster server, therefore you pay more." Customers never accepted that, and they won't accept this either.
The reason it grates is that buyers don't purchase technology. They purchase results — fewer outages, faster fixes, less risk, lower total cost. AI is a means to those ends. When an MSP charges for the means while the ends stay flat, the customer is right to push back.
There's also a credibility trap. If your AI genuinely makes service better, you should be able to point at the improvement: tickets resolved before the user noticed, mean time to resolution halved, security incidents caught at 2am instead of 9am. If you can show that, you don't need an 'AI line item' — the value speaks for itself and it's baked into a service the customer happily pays for. If you can't show it, then the premium is just a margin grab dressed up in fashionable language, and sooner or later someone's finance director works that out.
A buyer's framework: when is an AI premium ever justified?
If you're an SME being quoted more for AI-enhanced services, you don't need to become a technologist to assess it. Ask four questions.
1. What outcome changes, and can you measure it? A justified premium is tied to a result you can verify — uptime, resolution speed, incidents prevented, hours of your own staff freed up. If the answer is about the provider's internal tooling rather than your experience, that's a red flag.
2. Who captures the efficiency? If automation saves the MSP labour, ask how that saving is shared. A fair arrangement passes at least some of it back to you, or reinvests it visibly in faster service. An arrangement where the MSP keeps 100% of the efficiency and charges you a premium is paying twice.
3. Is the premium for the tool or the value? "We licence an AI platform" is a cost line, not a benefit. "We catch and remediate 40% of issues before they reach a human" is a benefit. Only pay for the second kind.
4. What happens if it doesn't work? Any provider confident in AI-driven outcomes should be willing to stand behind them with a measurable commitment. If they want the premium but won't put the result in writing, the confidence is one-sided.
Run those four questions and most 'AI taxes' fall apart. The genuine value cases survive — and you'll recognise them because the provider leads with your numbers, not their software stack.
What MSPs should do instead: pricing tied to operational data
The way out of the squeeze isn't to hide the AI cost or slap a premium on top. It's to change what you sell. Move from billing for the catalogue — devices, seats, line items — to billing for the operational outcomes your tooling now makes possible. The data exists to do this; most MSPs are sitting on it already.
Here are four models that protect margin without triggering the 'tax' reaction.
Outcome-based tiers. Instead of 'standard' and 'premium support', define tiers by what the customer gets: guaranteed resolution times, proactive incident prevention rates, security response windows. AI is what lets you hit the higher tiers profitably. The customer pays for the tier, not the technology behind it. Your automation becomes a margin lever you control, invisibly.
Efficiency-share pricing. Be upfront that automation reduces labour, and split the saving. If you're now resolving a class of tickets at a third of the old cost, offer a lower per-ticket rate and take a smaller absolute margin on a larger volume of resolved work. Customers love a provider who actively lowers their bill — they renew, they refer, and they stop shopping around.
Risk-and-prevention pricing. Price around what you prevent. An SME that suffers no ransomware incident, no extended outage, no data loss across a year has received enormous value that the old model never captured. Detection and prevention driven by operational data and AI monitoring is exactly where this lands. Bundle a measurable prevention commitment and charge for the outcome of nothing bad happening — which is harder to sell but far more defensible than a tooling line item.
Capacity-and-growth pricing. For SMEs that are scaling, sell headroom: the ability to add users, sites, or services without a step-change in cost or a drop in service. AI-driven operations make this elastic in a way human-only delivery never could. You're charging for the customer's freedom to grow smoothly, which they'll pay for gladly.
The honest conversation wins
The MSPs who get through 2026 in good shape will be the ones who have the direct conversation. "We've automated a lot of what we used to do by hand. Here's how that shows up for you — faster fixes, fewer incidents, and here's where we've passed savings back." That's a renewal conversation, not a confrontation.
The ones who struggle will be the ones reaching for the premium line item and hoping nobody asks what it does. Someone always asks. And the answer 'because AI' has a very short shelf life with anyone who controls a budget.
The service catalogue had a good run. But pricing the labour of yesterday while delivering the automation of tomorrow is a trick that only works until the customer notices. Better to change the model on your own terms — and let the value, measured and shared, do the selling for you.