An Outcome Is Not a Price

The client says a sentence that seems impossible to argue with: I don’t want to pay for hours, I want to pay for the result. Within seconds the supplier finds itself on the wrong side of the conversation. Defend the billable days and you look like someone who wants to be paid for inefficiency. Ask for a fixed fee and you look like someone who doesn’t believe in their own work. Try to distinguish what you control from what depends on the client, and you already look like someone building an alibi.

The sentence works because it contains a truth. For too long we sold activities as if their completion coincided with value. Delivered software can go unadopted, an integration can leave the process unchanged, a consulting engagement can produce no decision at all. The client is right not to want an inventory of hours. The client is right to ask the supplier to share at least part of the risk.

I know this well because I made that argument myself. In March I wrote that time & materials is dying and that pricing for results is the most adult model we have. I still think so. But in the months since, I have sat through enough negotiations, on both sides of the table, to notice that between wanting that model and knowing how to write it into a contract lies a distance almost nobody is measuring. This essay is the second half of that argument, and it is the less comfortable half.

The trouble begins precisely when you try to put the sentence in writing. Which result? Measured by whom? Against what starting point? Over what window? What would have happened without the intervention? Which decisions stay with the client? What obligations will the client have around adoption? What happens if prices, staff, processes or strategy change while the supplier waits to find out whether it will be paid?

At that point it becomes clear that the discussion was never only about price. It was about deciding who would have the right to tell the story of what caused whatever happened next.

Which Result, Exactly

The word outcome is now applied to almost any billing unit other than hours, and that is the first confusion to dismantle. Public policy evaluation has used a distinction for decades that B2B would do well to steal: the OECD results chain separates inputs (the resources used), activities (what gets done), outputs (the product directly delivered), outcomes (the effect produced or made possible) and impacts (the longer-term transformation). This is not pedantic taxonomy: it is a causal chain, and precisely for that reason it forces you to spell out assumptions, risks and the links between what the supplier does and what eventually happens.

Translated into software and services, the ladder looks like this. The day and the hour are inputs. The delivered feature, the integration, the report are outputs. The active user, the token, the transaction are metered consumption. The resolved ticket, the completed case, the identified anomaly are operational outcomes. Revenue growth, churn reduction, margin are economic outcomes. Health, employment, safety are impacts. The further you move along the chain, the more the billed unit resembles the client’s real value. But the time needed to observe it grows too, along with the number of external variables, the cost of verification and the difficulty of attribution.

The rule condenses into this: the desirability of an outcome grows with its distance from the supplier’s work. Its contractability shrinks with its distance from the variables the supplier controls.

Then there is a second confusion, more insidious because it suits everyone. In value-based pricing, the price is set in relation to the expected economic value: an intervention potentially worth a million can be sold for a hundred thousand euros, but payment does not depend on the million materializing. In outcome-based pricing, payment is contingent on the result occurring. What changes is not the method for calculating the price: it is who bears the risk. Many offers presented as outcome-based are actually value-based pricing, per-output prices, usage-based pricing, sales bonuses, SLAs with penalties, success fees on easily counted events. This is not lexical fussiness. Calling an output an outcome lets you promise alignment without actually taking on the uncertainty of economic impact.

The Counterfactual

The heart of the problem is that the observed result does not coincide with the result the supplier produced. If sales rise after a CRM rollout, the increase may come from the CRM, from the new campaign, from a discount, from a talented new salesperson, from a competitor’s exit or from seasonality. To attribute value you have to answer a question no dashboard shows: what would have happened over the same period, under the same conditions, without the intervention?

In most commercial relationships there is no control group. The counterfactual has to be estimated, normalized or negotiated, and the moment it becomes negotiated it stops being a technical fact. An outcome without a counterfactual is an opinion with an economic clause attached.

And before the counterfactual comes the baseline. To pay for a result you need to know the starting value, its historical variability, the quality of the data, the relevant segmentations, the concurrent interventions, the seasonality, the changes that occurred during the period and the source authorized to produce the final number. Two dashboards can show different values without either being technically false: a different time window, deduplication rule or definition is enough. Before discussing the price, you have to establish which administrative reality will carry contractual weight. An outcome without a baseline is not a result. It is retrospective storytelling.

Then there is control, and here the argument stops being statistical and becomes political. The principle is familiar to anyone who writes serious public contracts: risk should sit with the party best able to manage it. The UK Cabinet Office guidance, updated in June 2026, says it without hedging: a supplier loaded with risk it does not govern will almost certainly produce a risk premium in the price, underperformance as its focus shifts to cost cutting, or a contract so onerous it collapses. And when it recommends payment-by-results mechanisms, the same guidance adds a condition I almost never hear mentioned in Italian negotiations: the supplier must have broad scope to determine how to achieve the outcome. If the fee depends on conversion growth, who controls the prices, the campaigns, the product range, the lead quality, the response times, the sales force? A supplier can be paid on the result only to the extent that it holds the levers to produce it.

Finally, cooperation. Most outcomes are co-produced: a system cuts administrative time only if the client redesigns procedures and enforces adoption, a sales platform generates opportunities only if someone follows them up, a predictive system reduces breakdowns only if maintenance acts. The client is not the passive beneficiary of the result: the client is one of its producers. A real outcome-based contract therefore contains client obligations with the same economic dignity as the price: access to data, response times, dedicated staff, adoption levels, stability of operating conditions. Without those obligations, the contract contains unilateral responsibility for a bilateral result.

A Market That Wants to Buy What It Doesn’t Measure

So much for the mechanism. The next question is whether the Italian market has the infrastructure to make it work, and the available data suggests an uncomfortable but precise answer.

In 2025, according to ISTAT, the Italian national statistics institute, 56% of Italian firms with ten or more employees used at least one business software package and 68.1% bought intermediate or advanced cloud services. But only 42.7% performed data analysis, in house or through external providers: 41.9% among SMEs, 83.6% among large firms. ERP was used by 48.8% of SMEs, CRM by 21.1%, and Business Intelligence was present in 16% of firms, with no significant change since 2023. These numbers do not prove that Italian companies are incapable of measuring. They prove something more interesting: the capacity to collect and interpret economic and operational data is distributed very unevenly. A time-based contract can be administered even by a firm with fragmentary data. A contract based on churn reduction or administrative cycle times cannot. When the baseline is missing, the problem is not just that you cannot measure the result: you cannot even establish what would have happened without the supplier.

The closest available picture comes from the servitization of Italian machinery. The observatory run with ASAP Service Management Forum across roughly two hundred firms shows a market that understands the strategic direction perfectly and struggles to turn it into organization: 57% declare a dedicated services strategy (83% among large firms, 48% among SMEs), but only 41% have dedicated service roles, 34% have specific responsibility for new service development and just 29% assign it a budget. Digital and connected services are worth about 1% of revenues, recurring maintenance contracts about 3%. One firm in five claims to offer product-as-a-service models, but with still negligible revenues. This is not conservatism: it is the physiological distance between a slide and a P&L. In Italy, as-a-service has often already arrived in the board presentation. It has not yet arrived in the income statement.

In the public sector the tension is even more visible. An exploratory study from Politecnico di Milano on outcome-based contracts in a public-private pilot finds that the main obstacles are not legal but infrastructural: the skills and the data storage and management systems needed to administer instruments of this kind are missing. It is a circumscribed academic finding, not a national statistic, but it is consistent with the rest of the diagnosis. A public authority can write indicators, penalties and incentives; the problem is administering an invoice that depends on a counterfactual, on data scattered across systems and on the behavior of actors other than the supplier. To be purchasable, a result must be not only true, but administratively defensible.

Finally, there is a signal worth more than many surveys. In 2026 Assonime, the association of Italian joint-stock companies, dedicated a position paper to the servitization of Made in Italy, including pay-per-use and pay-per-outcome models, and its central proposal is not a new price list: it is an enabling infrastructure, a platform integrating production capacity, digital services, financial tools and expertise that are currently fragmented. When a business association feels the need to propose infrastructure to make a pricing model possible, it is implicitly admitting that the model is not a commercial trick a single company can adopt on its own. It requires an ecosystem of data, finance, insurance and verifiable trust.

Risk Dressed Up as a Price

There is one aspect of results-based pricing that almost never gets said out loud: the supplier is not simply selling differently. It is financing the interval between the work done and the result observed, and it is insuring part of the client’s variance. An outcome that matures over twelve months means twelve months of locked-up capital. If the result can be contested, the period stretches. If the contract is entirely variable, the supplier also carries the risk of zero revenue. The client gets financing and an insurance policy embedded in the price; if the price contains no premium for those risks, the supplier is either giving them away or has not understood them.

Here the structure of the Italian production system stops being a statistical detail. In the permanent business census, covering firms with three or more employees, 78.9% are micro-enterprises with fewer than ten people, 18.5% small, 2.2% medium and 0.4% large. Not all fragile, not all undercapitalized. But accepting results-based remuneration means financing the work upfront, absorbing revenue volatility, absorbing measurement errors, provisioning for disputes and, above all, spreading the risk across a sufficiently large portfolio. A global platform can spread risk across thousands of clients and millions of events. A software house of ten, thirty or fifty people cannot behave like an insurance company without insurance capital, actuarial data and diversification. Why do we consider it innovative to ask a small technology firm to finance the project, insure the client’s decisions and collect only when a contestable metric moves in the right direction?

The resulting pathologies are two, and they mirror each other. The first is the proposal the supplier receives: “I don’t want to pay for the project, I’ll give you a percentage of the result.” It can look like trust; it is often a request for an asymmetric transfer of risk, and the signs are recognizable. The outcome is far from the intervention, the baseline doesn’t exist, the data is controlled solely by the client, the decisive decisions stay with the client, there is no adoption obligation, the payment is entirely variable, there are no exclusions for external events and no premium for the locked-up capital. The buyer is asking the supplier to underwrite the technological, organizational and commercial risk all at once. The client wants the pricing of the outcome and the governance of a fixed-scope brief.

The second pathology is the offer the supplier makes to win the tender, stand out or get past the price objection: an ambiguously defined outcome, a metric that can be gamed, necessary conditions left unstated, a bonus for the sales team and the risk dumped on delivery, the required capital never calculated. And a phenomenon sellers systematically underestimate: adverse selection. The clients most eager to pay only on results may be precisely the ones who know better than the seller how fragile their organization is, how poor their data is, how unlikely adoption is. The supplier does not receive a random sample of opportunities: it receives a concentration of the cases the client prefers not to finance directly. In results-based pricing, the client often knows more than the supplier about how improbable the result is.

Both roads end at the same place: many outcome-based offers are not innovative prices. They are insurance policies written by people who don’t know they have become insurers.

Where Outcomes Already Work

It would be false, though, to say that Italy is simply not ready, and the exceptions are the most instructive part of the story, because they show exactly which conditions are required.

The first is pharmaceuticals. In the monitoring registries of AIFA, the Italian medicines agency, risk-sharing agreements tied to therapeutic outcomes have existed for years: under Payment by Results, the pharmaceutical company fully reimburses the treatments of patients who do not respond; under the Success Fee model, the national health service initially receives the drug for free and pays for the dispensed packs only after therapeutic success has been verified. These are contracts in which the invoice literally depends on the clinical result, and they work. But they work because before the price there exist an eligible population, clinical criteria, a time window, a definition of response, a national platform, actors authorized to certify, and payback procedures. Nobody signed a generic “we’ll pay if the patients get better.” Somebody built an infrastructure capable of translating a clinical outcome into an administrative event. The outcome does not precede the measurement infrastructure. It is its product.

The second is energy. In Energy Performance Contracts the energy improvement is defined, measured and monitored across the whole life of the contract: the guidelines from ENEA, the Italian agency for energy efficiency, insist on baselines, guaranteed savings, precise obligations for each party, verification procedures and rules for handling changes, from energy prices to how intensively the systems are used. Here the outcome is contractable because consumption is measurable, the baseline can be built from history, normalization techniques exist, the supplier controls a substantial share of the levers, and the result can be observed repeatedly rather than once. Consip’s €1.4 billion Integrated Energy Service tender for local authorities, launched in March by Italy’s central purchasing body, combines fees, guaranteed energy savings, a renewable energy quota, comfort, response times and a contract extension tied to service quality. A hybrid model, not a binary bet on generic “improvement.”

These two cases allow a conclusion far more precise than the culturalist complaint: Italy knows how to use results-based contracts when the result has been institutionalized. What it does not yet know how to do at scale is improvise that same infrastructure inside an ordinary commercial negotiation.

The AI Plot Twist

The AI debate seems to suggest that outcome pricing is now inevitable: the marginal cost of execution is collapsing, agents work on their own, the client doesn’t want to buy tokens or seats but work done. The direction is real. But if you look at the actual offers of the global vendors, a much more interesting fact emerges: even the companies with more data, more scale and more technological control than anyone on the planet carefully avoid pricing broad economic outcomes.

Intercom bills Fin at $0.99 per outcome, but the outcome is defined as a conversation resolution, a handoff to a procedure or a disqualification: a resolution can be counted when, after the last answer, the customer asks for no further help. HubSpot announced for April 2026 $0.50 per resolved conversation and one dollar per lead for outreach: what gets billed is not customer loyalty or a closed sale, but an operational event close to the vendor’s own system. Salesforce uses Flex Credits: each Agentforce action consumes twenty credits, roughly $0.10, and the unit actually measured is the action, updating a record, summarizing a case, running a flow. The communications speak of investment aligned with value; the meter counts something else.

This is not a criticism of the vendors: it is probably the rational choice. They are searching for the smallest unit of value that is atomic, frequent, observable, attributable, governed by their own system and billable without endless disputes. Every requirement on that list is a direct answer to one of the problems above: attribution, baseline, control, transaction cost. The sting is semantic: the term outcome is expanding in the rhetoric exactly as it shrinks in the price lists, down to events we would have called outputs, transactions or successful automations a few years ago. The commercial frontier of AI is not pricing the big business results. It is shrinking causal distance until it finds an event small enough to be called an outcome without becoming a lawsuit.

This lets us bring the discussion back to Italy without provincialism. The problem is not that Italian firms lag behind a world where everyone pays for AI-produced EBITDA, because that world does not exist: even the most advanced vendors prefer resolved conversations, qualified leads, updated records. So the sensible prediction is that the Italian market will be ready much sooner for narrow operational outcomes, the correctly classified document, the case completed without human intervention, the reconciled invoice, the verified anomaly, the process time under a threshold, than for fees tied to revenue growth, churn reduction or the overall success of a digital transformation. I suspect this distinction, between the outcome you can count and the outcome you can only narrate, will matter more than any debate about the future of pricing.

Don’t Go Back to the Hours

The best objections to this essay deserve to be taken seriously, because two out of three are well founded.

First: the supplier must have skin in the game. True, and it is the strongest objection. Too many suppliers sell days, deliver formally correct artifacts and leave the client with all the risk that they produce no value. But the right answer is not to defend the billable day: it is to demand that risk sharing be proportional to control. It becomes extraction when one party keeps the decisions and transfers the consequences to the other. You do not transfer responsibility without transferring authority.

Second: hours reward inefficiency. Also true, and I wrote it myself. A supplier paid by time can earn more by working more slowly. But eliminating an imperfect, observable unit does not license replacing it with a unit that is morally attractive and causally undecidable. The billable day survives not because it is intelligent. It survives because it is verifiable. The true opposite of the billable day is not the result: it is observability.

Third: AI will make everything measurable. Here I disagree. AI makes it cheaper to count what happened; it does not by itself establish why it happened. The counterfactual, the attribution, the client’s cooperation, the lag between intervention and effect and the external events all stay exactly where they were. As I argued about metrics, counting better is not understanding better.

So the direction is not back to the days, but a maturity ladder, which is how I work whenever I get to choose. First you sell a paid observability phase: defining the baseline, mapping the sources, verifying data quality, identifying the levers, designing the verification mechanism. The provocation is that the first thing a serious supplier should sell, in an outcome-based project, is the right to find out whether the outcome is actually sellable. Then a fixed component covering what the supplier bears regardless: setup, infrastructure, integration, operations, non-controllable risk. Then a bounded variable component, indicatively 20-30% of the value, tied to one or two operational outcomes, not a constellation of KPIs. With four non-negotiable properties: symmetry, because a contract where the client keeps all the upside and transfers only the downside is not alignment; a cap and a floor, because risk without a ceiling and a fee without a sustainable minimum are not a price but a free option granted to the client; client obligations treated as economic conditions; and causal change control, because a change in prices, process, staff or strategy can invalidate the baseline, and the contract must say when the metric gets recalibrated. Better, finally, frequent measurements and quarterly settlements on a ledger both parties can query than a single verification after twelve months on a spreadsheet produced by the economically interested party.

A commercial relationship can start from outputs and SLAs, move to operational outcomes once the baseline exists, and eventually reach a share tied to the economic result. Results-based pricing should not be the starting point of trust. It should be one of its end products.

A Risk Someone Can Govern

Outcome-based pricing will reach the Italian market too. In some sectors it already has, but it did not appear thanks to the courage of a salesperson willing to waive the fee. It appeared where someone had first built registries, baselines, eligibility criteria, measurement systems, mutual obligations and verification procedures. The outcome did not replace governance. It became contractable because governance had turned it into an administrative fact.

AI will accelerate this trajectory, but probably in a form less heroic than the one being announced. We will pay less and less for tokens, seats and days. We will pay for processed documents, resolved conversations, completed cases and executed actions. We will call outcomes events ever closer to the machine, because those are the ones the machine can measure and the supplier can control. The bigger results will keep being produced by systems in which technology, organization and human decisions remain inseparable, and no price list will split them onto an invoice.

The final diagnosis, then, is less provincial and harsher than “the Italian market doesn’t get it.” The Italian market does not reject results-based pricing because it doesn’t believe in results. It rejects it, or practices it badly, because it tries to make the price do the work that governance, data and organization have not done. We want to move past the billable day because it rewards what the supplier controls instead of what the client wants. But in doing so, we risk paying the supplier for what the client wants and what neither party can attribute to it with any certainty.

Refusing to guarantee a result you do not control is not a lack of faith in your own work. It can be the sign of having finally understood the client’s work. In the same way, asking the supplier to share the risk is not unfair. It becomes unfair when the client keeps all the levers and considers it innovative to transfer the consequences of its own decisions elsewhere. A mature market is not one where every supplier agrees to be paid only upon success. It is one where the parties can tell apart the risk each of them controls, measure it and price it.

You don’t pay for a result. You pay for a risk someone is actually able to govern.

The author

Andrea MargiovanniAdvisory

I do independent advisory for those buying IT services: vendor assessment, contract review, second opinion on RFPs. I don't sell hours. I sell readings of the context and defensible decisions.

See the guide →