ComparisonManaged Services

FTE, fixed fee, per transaction or outcome: pricing a managed service once AI does the work

The pricing model decides what a provider is paid to do. Under FTE pricing, a provider earns more by staffing more, which runs directly against automation once software can do part of the work. This comparison sets FTE, fixed monthly fee, per-transaction and outcome-based models side by side on incentives, baselines, risk and predictability, then covers the mechanics that make each one work.

Reviewed 10 October 2026 6 min read

On this page

Why the pricing model matters more once AI does the work
Four pricing models, and gainshare, in plain terms
The four models compared on incentive, risk and predictability
Setting the baseline and choosing outcomes the provider can influence
Contract mechanics to settle before signing
Choosing a model for your situation
A hypothetical insurer moves claims intake off FTE pricing
Questions and answers
Sources

Why the pricing model matters more once AI does the work

Traditional business process outsourcing priced labor. Productivity improved slowly, so a price per person held up for years. When software agents can take on a large share of routine items, the cost of doing the work changes faster than most contracts are written to handle.

If the price is tied to headcount, someone loses: either the provider gives up revenue by automating, or the client keeps paying for capacity it no longer needs. Both outcomes slow automation down. ColdAI's comparison pages state that it prices managed operations on outcomes rather than per FTE¹. The trade-offs below apply to any provider's proposal, including ours, and this page quotes no prices.

Four pricing models, and gainshare, in plain terms

FTE-based pricing: The client pays a rate per full-time-equivalent person assigned to the service, sometimes with a management fee. Simple to buy and easy to compare between providers, but it measures effort rather than results.
Fixed monthly fee: One price for an agreed scope and volume range. The provider keeps any efficiency gains during the term, which motivates automation, and carries the risk of under-pricing the work.
Per-transaction pricing: A unit rate for each invoice, claim, case or contact handled. Cost moves with volume, and unit definitions have to be watertight so that work is not split or reclassified to raise the count.
Outcome-based pricing: Payment tied to a measured result, such as cost per resolved case, days to close the books or backlog age, usually with service levels as a floor. It needs a reliable baseline. See the outcome-based model glossary entry.
Gainshare: A variant in which client and provider split a measured benefit, such as reduced cost or recovered leakage, against an agreed baseline, often on top of a base fee and within caps.

The four models compared on incentive, risk and predictability

Criterion	FTE-based	Fixed monthly fee	Per transaction	Outcome-based
Incentive to automate	Negative: automation shrinks billable headcount	Positive within the term: savings become provider margin	Depends on whether unit rates fall as automation rises	Strong: the provider earns by improving the measured result
Baseline needed	Low: a staffing plan	Medium: volumes and scope to size the fee	Medium: unit definitions and volume history	High: measured performance and an agreed method
Who carries volume risk	Client pays for capacity whether used or not	Provider, within agreed bands	Client pays per unit as volumes move	Shared, depending on how the outcome is defined
Budget predictability	High while staffing is stable	High	Varies with volume	Lower unless capped
Typical gaming risk	Padded teams and slow productivity gains	Quiet erosion of quality to protect margin	Splitting or re-routing work to inflate counts	Optimizing the metric instead of the business result
Suits	Short or undefined work that cannot yet be measured	Stable, well-scoped processes	Countable work with stable definitions	Processes with a result both sides can influence and measure

Many contracts combine models, for example a fixed fee for the core service, unit pricing above a volume band and a capped gainshare on one defined improvement.

Setting the baseline and choosing outcomes the provider can influence

An outcome price is only as fair as its baseline. Measure it before the service starts, ideally during discovery and the parallel run, from system data rather than estimates. Normalize it for volume, mix and seasonality, and write down the calculation method, the data source and who runs the numbers.

Choose outcomes the provider can actually move. Cost per invoice processed, time to resolve a customer case or the age of the claims backlog sit largely within its control. Revenue, market share or loss ratios do not, because your own decisions dominate them. Pair every outcome with quality floors from the service levels for AI-run operations, so a provider cannot earn outcome payments by producing fast but wrong output.

Contract mechanics to settle before signing

0 of 8 checked

Volume bandsA price per band, with re-pricing triggered when volumes move outside it for a sustained period.
Collars and capsLimits on gainshare upside and downside per period, so neither party faces a shock.
True-upsA periodic reconciliation of estimated and actual volumes or outcomes, with timing and a dispute process.
A productivity pathIn fixed-fee or unit deals, scheduled rate reviews as automation matures, or an agreed way to share the savings.
A quality gateOutcome payments count only in periods when service levels and quality thresholds are met.
Baseline reset eventsAcquisitions, new product lines or regulatory changes that justify re-baselining, and who decides.
Measurement rightsYour access to the data needed to verify every calculation the provider invoices against.
Exit termsTransition assistance rates and ownership of automation built during the term.

Choosing a model for your situation

If
The process is poorly measured and volumes are uncertain
Then
Use a fixed fee for a defined discovery and stabilization period, with a contractual point to move to unit or outcome pricing
You cannot price outcomes you have not yet measured.
If
The work is countable and the unit definition is stable, such as invoices, claims or onboarding cases
Then
Use per-transaction pricing with volume bands and a scheduled unit-rate review
It ties cost to demand while keeping automation savings on the table.
If
There is a clear, attributable result and a reliable baseline
Then
Use outcome-based pricing, or a base fee plus gainshare, with a quality gate and caps
It aligns the provider's revenue with the result you are buying.
If
A provider proposes FTE pricing for a process it says it will automate
Then
Ask in writing how the price changes as automation coverage rises
Otherwise the savings either stay with the provider or never arrive.
If
Budget certainty matters more than anything else
Then
Prefer a fixed fee with volume bands, and accept less upside from improvement
Predictability has a price, and it is usually paid in shared gains.

A hypothetical insurer moves claims intake off FTE pricing

Questions and answers

Why do providers still offer FTE-based pricing?

Because it is familiar, easy to compare and low-risk for the provider. It suits work that cannot yet be measured, such as a short discovery period. For a stable, high-volume process that software can partly handle, it rewards effort rather than results, so treat it as a starting point to move away from.

Is outcome-based pricing more expensive?

Not necessarily, but providers price the risk they carry. If the outcome is volatile or the baseline is weak, expect a premium or a cap on the provider's downside. The cost falls when the baseline is solid, the outcome is within the provider's control and measurement is transparent to both sides.

Which outcomes work well for outcome-based pricing?

Outcomes that are frequent, measurable from system data and largely within the provider's influence: cost or time per case, backlog age, first-time-right rates or recovered overpayments. Avoid outcomes driven mainly by your own decisions or the market, because neither side can then tell what the provider contributed.

Can we combine pricing models in one contract?

Yes, and most mature contracts do. A common shape is a fixed base fee for the core service, unit pricing above a volume band and a capped gainshare on one or two measured improvements. The more models you combine, the more important clear measurement rights and a single reconciliation process become.

Sources

ColdAI vs Accenture: delivery and pricing models compared — ColdAI

More in Managed Services

Back to Managed Services

Next step

Bring the pricing proposal you are weighing up

Send the scope, current volumes and the pricing structure you have been offered. We will point out where its incentives and baselines are likely to cause friction, and what an outcome-based alternative would need to measure.

Review a pricing model