Now Advisory · Buyer side guide · 2026 edition
Now Assist consumption, explained from the buyer side
How the metered assist model works, why agentic actions consume more than simple ones, and how to size a Now Assist consumption commitment that fits real usage rather than a vendor estimate.
Section 01How Now Assist consumption is metered in the 2026 model
Now Assist consumption is metered in assists, the unit ServiceNow uses to measure AI usage across the platform. In the 2026 commercial model, the AI capability is bundled into every tier, so the question is no longer whether you have access. The question is how many assists your workloads draw and whether your committed allowance matches that draw. Consumption, not entitlement, is now the variable that decides your bill.
This is a meaningful change from seat based thinking. A licence count is static and easy to forecast. Now Assist consumption fluctuates with adoption, workflow design and the mix of simple and complex actions your teams run. Two organisations with identical seat counts can draw very different assist volumes depending on how aggressively they automate. For the pricing context, see our Now Assist pricing guide and the metering detail in our ServiceNow metered assists guide.
The buyer side implication is that your commitment must be built from a usage model, not accepted from a vendor estimate. A commitment sized to the vendor's convenience is sized to maximise top up charges, not to fit your demand. Understanding what drives consumption is the first step to sizing it correctly.
Section 02What drives Now Assist consumption up
The largest driver of Now Assist consumption is the type of action. A simple action, such as summarising a record or suggesting a reply, draws a modest number of assists. A large agentic action, where the platform chains several reasoning steps to complete a multi part task, consumes materially more. As teams move from assisted suggestions toward autonomous agents, the assist draw per task climbs, sometimes sharply.
The second driver is breadth of adoption. The more workflows you enable across IT service management, customer service and HR, the more touch points draw assists. Consumption scales with the number of active use cases, not just the number of users, which is why a successful rollout can quietly outrun a commitment that looked generous on paper.
The third driver is workflow design. Inefficient automations that invoke the AI repeatedly, or agents configured to run on broad triggers, burn assists faster than tightly scoped ones. Consumption is partly a design decision, which means it is partly within your control. Our Now Assist consumption advisory service helps model these drivers so the commitment reflects how you will actually use the platform.
Section 03Sizing a Now Assist consumption commitment correctly
Sizing starts with a baseline. If you have run a pilot, the pilot data is your best evidence of per user and per workflow assist draw. If you have not, model the expected mix of simple and agentic actions across your planned use cases and weight accordingly. The goal is an expected annual assist volume with a defensible range around it.
From the baseline, build three scenarios: conservative, expected and aggressive adoption. The spread between them tells you how much headroom to negotiate into the commitment. Size the committed pool to the expected scenario with enough headroom to absorb normal growth, rather than to the aggressive case, which overcommits, or the conservative case, which guarantees overage.
The final step is to attach the right protections to the commitment: a fixed overage rate, an annualised measurement window and a conversion right so that sustained overage becomes an expansion at the committed rate. A commitment sized well but left without these protections still exposes you on the tail. The detail on forecasting the volume sits in our Now Assist budgeting guidance.
Section 04Reading the consumption model the vendor presents
When the account team presents a consumption model, read it as a sales artifact, not a neutral forecast. The estimate of how many assists your workloads will draw shapes both the commitment you buy and the top up exposure you carry. An estimate that is too low looks cheaper at signature and generates overage later. An estimate that is too high overcommits you to capacity you may not use.
The questions to ask are specific. What assumptions about the mix of simple and agentic actions underpin the estimate? How does the model treat growth across the term? What happens to unused assists at the end of a measurement period, and what is the overage rate if you exceed the pool? The answers reveal whether the model fits your demand or the vendor's revenue target.
The independent move is to bring your own model to the table. A commitment defended by your usage data, rather than accepted from the vendor's spreadsheet, changes the negotiation from a take it or leave it to a comparison of two forecasts, which is exactly the conversation you want to be having.
Section 05Benchmark observations on consumption commitments
Based on benchmark observations across real enterprise renewals, buyers who build their own consumption model typically size the committed pool closer to realistic expected usage with measured headroom, rather than to the conservative figure that maximises overage or the aggressive figure that overcommits. The result is fewer surprise top up charges and a commitment that scales with adoption rather than against it.
The bigger win is usually on the protections rather than the pool size. Fixing the overage rate at signature, securing an annualised measurement window and negotiating a conversion right together remove the tail risk that makes consumption pricing feel unpredictable. Buyers who do this convert a variable exposure into a managed line item.
These are typical negotiated outcomes used as internal leverage, not official list prices. The right commitment depends entirely on your adoption profile, which is why the model has to be yours.
Section 06Putting consumption at the centre of the renewal
The throughline is that Now Assist consumption is now a primary renewal variable, not a footnote. A renewal negotiated on seats and tiers alone leaves the fastest growing and least predictable cost untouched. Buyers who put consumption at the centre, with their own model and a clear set of protections, control the line item that would otherwise control them.
This work connects to the rest of the renewal. The tier you sit on, the uplift you accept and the overage terms you negotiate all interact with consumption, which is why we run them as one engagement rather than separate conversations. A commitment sized in isolation can still be undermined by an unfavourable overage rate or an uncapped uplift.
If your renewal includes Now Assist and you want the consumption model built before the vendor sets the terms, book a renewal assessment call and we will start from your usage rather than their estimate.
Section 07Monitoring consumption after signature
A consumption commitment is not a set and forget decision. Once the contract is signed, the value of the deal depends on tracking actual assist draw against the forecast that sized the commitment. Instrument your usage from day one so you can see, month by month, whether consumption is tracking the conservative, expected or aggressive scenario you modelled. The earlier you spot a divergence, the more options you have to respond.
Monitoring serves two purposes. The first is operational: if a workflow is drawing far more assists than expected, that is often a design problem worth fixing before it consumes the pool, not a reason to buy more capacity. Inefficient automations that invoke the AI repeatedly can be re scoped, which protects the commitment without sacrificing the outcome. The second purpose is commercial: a documented record of actual consumption is the strongest evidence you can bring to the next renewal, because it replaces the vendor estimate with your own measured reality. The detail on tracking sits in our ServiceNow metered assists guidance.
Monitoring also protects the conversion right. If your data shows consumption consistently approaching the allowance, you can trigger an expansion at the committed rate before you cross into top up territory, rather than discovering the overage after the fact. The buyers who manage consumption best are the ones who treat the signed commitment as the start of a measurement discipline, not the end of the negotiation.
The wider point is that consumption pricing rewards engagement. A buyer who sizes the commitment from a real model, fixes the overage rate, secures a conversion right and then monitors actual draw against forecast turns a variable and uncertain cost into a predictable one. A buyer who accepts the vendor estimate, signs and stops paying attention inherits whatever the meter produces. The difference between those two outcomes is rarely the headline price and almost always the discipline applied across the term, which is why we treat consumption as an ongoing advisory relationship rather than a single negotiation moment
FAQFrequently asked questions
What is Now Assist consumption?
It is the metered usage of ServiceNow AI features, measured in assists. In the 2026 model AI is bundled into every tier, so consumption rather than entitlement determines your cost. Your committed allowance must match the assists your workloads actually draw.
What makes Now Assist consumption go up?
Three things: the type of action, since large agentic actions consume materially more assists than simple ones; the breadth of adoption across workflows; and workflow design, since inefficient or broadly triggered automations draw assists faster than tightly scoped ones.
How do I size a Now Assist consumption commitment?
Build a usage baseline, ideally from pilot data, then model conservative, expected and aggressive adoption scenarios. Size the committed pool to the expected scenario with headroom, and attach a fixed overage rate, an annualised measurement window and a conversion right.
Should I trust the vendor consumption estimate?
Treat it as a sales artifact, not a neutral forecast. The estimate shapes both your commitment and your top up exposure. Bring your own model built from usage data so the negotiation compares two forecasts rather than accepting one.
Are these consumption figures official ServiceNow prices?
No. They are typical negotiated outcomes based on benchmark observations across real enterprise renewals, used as internal leverage. They are not official list prices and depend on your adoption profile.