Many MSPs are wondering how to price AI and automation services. Our recommendation: price AI and automation services on what the outcome is worth to the
client, not on the hours it took to build it. Four models hold up in
practice: value or outcome based pricing, a monthly recurring fee for an
operating automation, per workflow or per Digital Worker pricing, and a
project plus retainer structure for larger builds. Pick the model that
fits the engagement, anchor the number to the client’s measurable gain
(cash freed, headcount avoided, days cut off a cycle time), and
structure it to bill every month. Hourly rates and per-device fees do
not carry over from break-fix work, and pricing this kind of work like a
service ticket is the fastest way to undersell what you built.
MIPLY was founded by the team from Simpatico, a managed service provider (MSP) that has build an established and growing AI and Automation practice. Simpatico made the transition to sell AI before we built MIPLY to teach it,
and pricing was the part we got wrong first. Here is what actually
worked, and what it cost us to learn it.
Why
Hourly and Per-Device Pricing Breaks Down for This Work
Hourly billing and per-device fees are pricing models built for a
specific kind of work: labor that scales with time, and infrastructure
that scales with device count. Break-fix, help desk, patching, endpoint
management. All of it is priced on inputs because the inputs are what
the client is buying: your hours, your coverage, your seats.
AI and automation work is different. When you build a workflow that
eliminates a full time role, or an AI agent that handles scheduling
calls a staff member used to answer, you are not selling hours or
devices anymore. You are selling a result. If you price that result the
same way you price a help desk ticket, you cap your own upside at your
labor cost and hand the client the rest of the value you created for
free.
This is also where the industry is headed whether individual MSPs
choose it or not. Fewer devices and fewer seats under management is
coming as AI takes over routine work, and a per-device or per-seat model
shrinks along with it. The MSPs building recurring revenue around
outcomes now are the ones who will not watch their revenue base erode as
that happens.
The Four Pricing
Models That Actually Work
Value or outcome based pricing. You price a fee tied
directly to a measurable result: cash freed, cost removed, revenue
captured. This model has the most upside and is the hardest to sell
until you have a few case studies of your own to point to.
A monthly recurring fee for an operating automation.
Once a workflow or system is built and running, you charge a flat
monthly recurring revenue (MRR) fee to keep it operating, monitored, and
improved. This is the closest structure to your existing managed
services model, which makes it the easiest one to sell first.
Per workflow or per Digital Worker pricing. A
Digital Worker is an automated system or AI agent that performs tasks a
human employee used to do: ticket triage, scheduling, reporting, payment
reminders. You license each Digital Worker as its own billable unit and
charge MRR for what it does, not for the time it takes to run. This
keeps pricing simple for the client and scales cleanly as you add more
workers to an account.
Project plus retainer. You charge a fixed fee to
build the initial automation or system, then convert to an ongoing
retainer for monitoring, iteration, and expansion. This works well for
larger builds where the client wants to see the thing working before
committing to a recurring number.
Most MSPs end up using a mix. A monthly Digital Worker fee for the
first deployment, a retainer once you are managing several workflows for
one client, and outcome based pricing reserved for the engagements where
the financial impact is large enough and clean enough to measure.
Anchor the
Price to the Client’s Gain, Not Your Cost
The habit to break is pricing from your cost up. Break-fix pricing
starts with your labor cost, adds margin, and stops there. Outcome
pricing starts with what changed for the client and works backward.
Here is an illustration of the math, with made-up numbers to show the
shape of it. Suppose a client spends $200,000 a year of staff time on a
manual process, and your automation cuts that to $50,000. The client’s
gain is $150,000 a year, every year. If you charged that build out at an
hourly rate, you might invoice $15,000 once and walk away. Priced
against the gain instead, even a modest share of that recurring $150,000
is several times your hourly number, and the client is still keeping
most of the value. That is not extraction. It is pricing that matches
what actually happened.
There is no universal percentage that works across every engagement,
and anyone who hands you one number for every deal is guessing. What
does hold across engagements is the discipline: quantify the client’s
before state in dollars, quantify the after state, and price against the
difference, not against your build time.
Wondering how you will measure how to price for your projects? MIPLY provides its clients a platform that maps business processes, builds out project plans and estimates the outcome based value to the customer.
The
Mistake of Underpricing Your First Few Engagements
The first outcome based deal you do sets an anchor, whether you mean
it to or not. Price it too low because you are nervous, or because you
want a reference story, and you teach that client, and often the next
few after them, what this kind of work costs. Renegotiating up from an
underpriced anchor is much harder than pricing it right the first
time.
The fix is not to guess higher and hope. It is to do the value math
before you quote, even on the first deal. Quantify the before state,
project a realistic after state conservatively, and price against that
math instead of against what feels comfortable to charge a client you
have billed hourly for years. A client who has never paid you for
outcomes has no anchor yet either. That is the moment to set the right
one, not the moment to play it safe.
Structure
for Recurring Revenue, Not One-Off Projects
A project fee, even a large one, is still a one-time transaction. The
entire point of moving into AI and automation services is to build a
second recurring revenue line next to your existing managed services
contracts, not to run a bigger consulting shop.
The industry numbers make the same argument. The Service
Leadership Index Q4 2024 report found MSP project and professional
services gross margins dropped from 23 percent to 12.9 percent year over
year, while the Q2
2024 report put managed service gross margins at 46.2 percent.
Recurring service work carries more than three times the gross margin of
project work in those benchmarks. Building your AI services as projects
means building them on the eroding side of that gap.
Practically, that means most engagements should land on an MRR number
somewhere in the structure, whether that is the Digital Worker license
fee, the operating and monitoring retainer, or an outcome based fee
billed monthly against an ongoing result rather than paid once. Even
project plus retainer deals should be built so the retainer, not the
project fee, is where the relationship lives after month one. Freed
operational capacity is not just a cost saving for the client. Converted
into a Digital Worker and priced right, it becomes a new, durable
revenue stream for you.
Why Listen to This
We are not writing this from a theory of how MSPs should price AI work.
We have learned a lot from our experience, and MIPLY exists because other MSP owners kept asking how we did it. MIPLY is a coaching and skills transfer program, built inside a
working MSP, for MSP owners making this same transition. We do not teach
a framework we have not run ourselves first.
One honest caveat: outcome pricing only works if you can measure the
outcome. If your client will not share the numbers, or the workflow’s
value is too fuzzy to quantify, start with a flat monthly fee and earn
your way to the data. Get the pricing right and you build a second
recurring revenue line on work you were about to give away. That is the
whole game.
Frequently Asked Questions
Should I ever go back to hourly billing for AI or automation
work? For scoping and discovery, sure. Hourly or a flat
discovery fee makes sense while you are figuring out what you are
building. Once the automation or Digital Worker is live and producing a
measurable result, hourly billing undervalues the work and caps your
revenue at your labor cost.
What if I can’t measure the client’s before state
accurately? Then the assessment is not finished yet. Value
based pricing depends on a documented before number, in dollars, hours,
or headcount. If you cannot produce that number, price on a monthly
recurring fee for the operating system instead, and move to outcome
pricing once you have the data.
How do I introduce recurring pricing to clients used to
project quotes? Start with a Digital Worker fee or an operating
automation fee, framed the same way clients already understand MRR from
their existing managed services contract with you. It is a familiar
structure applied to a new category of work, which makes it an easier
sell than opening with a percentage of value captured.
Is per-Digital-Worker pricing the same as per-seat
pricing? No. Per-seat pricing scales with headcount, which is
exactly what shrinks as AI takes over routine work. Per-Digital-Worker
pricing scales with the number of automated systems you deploy and
manage, and that number can grow independently of how many people the
client employs.
Recent Comments