White paper
This Now Assist consumption model guide explains how metered assists work, why large agentic actions cost materially more, how overage is triggered, and the buyer side moves that keep consumption from becoming an invoice you did not plan for.
White paper · 2026 edition
How metered assists, agentic actions and overage really work.
Tell us who you are and the paper opens immediately. No download, no waiting.
Corporate email only. Free mailboxes will not unlock the paper. No newsletter, no sales sequence. We may follow up once, personally.
Contents
Overview
This Now Assist consumption model guide exists because the 2026 commercial model turned AI from a fixed purchase into a metered cost, and most buyers are pricing it for the first time. Since April 2026, AI is bundled across the Foundation, Advanced and Prime tiers, and the assists that power Now Assist are metered. The capability is included; the consumption is not. That single distinction is where the new cost lives.
The consumption model deserves its own guide because it behaves unlike the rest of a ServiceNow agreement. Licenses are counted; assists are consumed. A license cost is stable once agreed; consumption cost moves with how your teams actually use the platform. A buyer who negotiates the tier carefully but ignores the consumption model has secured only half the deal, and usually the smaller half over a multi year term.
We are independent ServiceNow negotiation advisors with no vendor partnership and no reseller margin. This guide reflects benchmark data from real enterprise renewals rather than list price theory. The aim is to give procurement, ITAM and finance a clear model of how assists are consumed, where overage exposure sits, and how to negotiate an allowance that reflects real usage.
Assists
The unit of the consumption model is the assist. A Now Assist feature consumes assists when it runs, and the count is the meter that drives cost. Routine actions, a short summary, a suggested response, a quick classification, are inexpensive in assist terms. The model is designed so that light, everyday use stays comfortably inside the allowance that comes with the tier.
The important nuance is that not every action consumes the same number of assists. The consumption of a given feature depends on what it does and how much work it performs. This means two teams on the same tier, running different workflows, can consume very different volumes. The tier label tells you the allowance you start with; the workflows tell you how fast you draw it down.
For a buyer, the consequence is that you cannot price the consumption model from the tier alone. You have to look at the specific Now Assist features your teams use and how often they run them. Our Now Assist pricing analysis sets out the assist economics that underpin this guide.
Agentic
The single most important fact in the consumption model is that large agentic actions consume materially more assists than routine ones. An agentic action is one where the platform does not just suggest but acts: it plans a sequence, calls steps, and completes a task with limited human input. That autonomy is the value of the feature, and it is also what makes it consume so much more.
The multiplier between a routine assist and a large agentic action is where the consumption model surprises buyers. A workflow that looks affordable in a demo, where it runs a handful of times, can generate a very different number at production volume across thousands of cases a month. The demo shows the capability; only a consumption model shows the cost. The two are easy to confuse and expensive to confuse.
For the buyer this means the agentic workflows you plan to deploy are the real drivers of consumption cost, not the tier label or the headline that AI is bundled. Estimating the assist volume of each intended agentic workflow before you respond to a quote turns the allowance conversation from a guess into a defensible position.
Allowance
Each tier in the 2026 model carries an assist allowance, an included volume of consumption that comes with the bundle. The allowance is sized to cover light usage comfortably while leaving real agentic adoption to push beyond it, at which point overage charges begin. It is a negotiated number, not a fixed law, and treating it as negotiable is the most useful stance a buyer can take.
Because the allowance is negotiable, the first renewal under the consumption model sets the baseline that every later renewal measures against. Accept a thin allowance to secure an attractive tier price and you have simply moved cost from a line you negotiated to a line you will discover through overage. Negotiate the allowance up front, with headroom for the workflows you actually intend to run, and you keep the model honest.
The discipline is to treat the allowance as a primary commercial term, not a footnote in the order form. Ask what the allowance is, what it is based on, and what the overage rate is, and pin all three in writing before you sign. Our Now Assist consumption advisory builds that allowance model with you.
Overage
Overage is where the consumption model becomes a hard cost. When consumption exceeds the allowance, top up charges apply, and those charges are far less negotiable after signature than before it. The bundle gives the impression that AI is paid for, which is exactly why overage catches teams off guard when the first invoice above the allowance arrives.
The exposure compounds because adoption tends to grow. A workflow that sits comfortably inside the allowance in year one can cross the threshold in year two as more teams adopt it. Without a fixed overage rate written into the contract, that growth is priced at whatever rate the vendor sets, which is rarely the rate you would have agreed under competitive pressure before signing.
The buyer side move is to model consumption with headroom, negotiate an allowance that fits, and pin the overage rate in writing as a hard number. The overage rate governs the cost of every unit of growth, so it deserves the same attention as the tier price itself, not a glance at the close of the negotiation.
Modelling
The core buyer side discipline in the consumption model is to model usage before you respond to a quote. A consumption model estimates the assist volume of the workflows you intend to run, weights the agentic actions heavily, and produces a projected annual consumption with headroom for growth. That projection is what lets you negotiate an allowance that fits rather than accept the allowance the tier happens to provide.
Modelling does not require perfect data. It requires an honest inventory of the Now Assist features in use or planned, an estimate of how often each runs, and a heavier weighting for the agentic workflows. The output is a range, not a single number, and a range is enough to negotiate an allowance and an overage rate with confidence. The absence of a model, by contrast, leaves the allowance as a vendor default.
Where consumption is still uncertain, the model also informs the term length. Committing to a long term before you have real production data on assist consumption locks in an allowance you sized on estimates. A shorter initial term or a built in review point protects you, and that decision belongs in the consumption model, not at the close of the negotiation.
Benchmarking
Benchmarking turns the consumption model from a vendor explanation into a buyer position. A quote implies that a given allowance and overage rate are standard for your tier and volume. Useful benchmarks test that against comparable enterprises: what allowance similar estates secured, what overage rate they fixed, and how their consumption compared to the allowance they negotiated.
The benchmark here has to account for the kind of workflows you run, because a comparable enterprise running mostly routine assists is a different benchmark from one running agentic workflows at scale. The useful comparison matches both size and workflow profile. With that benchmark in hand, an allowance below range becomes contestable and an overage rate above range becomes negotiable, on evidence rather than opinion.
Based on benchmark observations, the allowance and overage rate are where the most value is available in the consumption model, because buyers focus on the tier price and let the consumption terms pass. The wider benchmarking method sits in our ServiceNow renewal negotiation service, and the consumption specifics in our Now Assist pricing analysis.
Framework
Putting it together, the Now Assist consumption model is negotiated through a short sequence. First, inventory the Now Assist features your teams use or plan to use. Second, estimate how often each runs and weight the agentic workflows heavily, producing a projected consumption range with headroom. Third, compare that projection against the allowance each candidate tier provides.
Fourth, treat any gap between the allowance and projected consumption as a negotiation point, whether by adjusting the tier or negotiating the allowance up. Fifth, pin the overage rate in writing as a hard number. Sixth, benchmark the allowance and overage rate against comparable enterprises with a similar workflow profile, and concentrate the negotiation where the gap is largest.
Each step feeds the next. Skipping the consumption model means negotiating the allowance blind. Skipping the overage rate means leaving the cost of growth open. The sequence is what turns the consumption model from a framing you accept into a set of terms you control. For the wider 2026 picture, read the ServiceNow 2026 commercial model guide.
Mistakes
The most common mistake in the consumption model is treating the assist allowance as a fixed feature of the tier rather than a negotiable term. Buyers who accept the default allowance, and an overage rate left unstated, hand the vendor control of the cost of every unit of growth. The allowance and rate are exactly as negotiable as the tier price, and they deserve the same scrutiny.
A second error is modelling consumption from the demo rather than from production. A workflow that runs a handful of times in a proof of concept consumes a trivial volume of assists. The same workflow at production scale, across thousands of cases a month, can consume orders of magnitude more, especially where it is agentic. Pricing the model on demo volume understates the real cost and the allowance you need.
A third mistake is committing to a long term before consumption is understood. The consumption model rewards real production data, and signing a multi year allowance on an estimate locks in a guess. Where the workflows are still maturing, a shorter initial term or a built in review point protects you from baking an under sized allowance into a commitment you cannot easily revisit.
After signature
A well negotiated allowance and overage rate erode unless they are monitored across the term. Assist consumption grows as more teams adopt Now Assist features and as agentic workflows expand, so an allowance that fit comfortably in year one can come under pressure by year two. The fixed overage rate written into the contract is what protects you when consumption crosses the threshold, which is why it has to be there from the start.
The practical mechanism is a regular reconciliation between contracted allowance and actual consumption, ideally quarterly. That cadence flags consumption trending toward the overage threshold while there is still time to act, whether by adjusting workflows, renegotiating the allowance, or planning for the next renewal. Consumption monitored quarterly is consumption you control; consumption reviewed only at renewal is consumption that controls you.
This monitoring also builds the evidence base for the next negotiation. An organisation that has tracked its assist consumption across the term arrives at its next renewal with real data on what it uses, which is the strongest position from which to negotiate the allowance. The discipline that produced the first fair outcome is the same discipline that sustains it, which is why monitoring is part of the consumption model rather than an afterthought.
Advisory
The consumption model is new enough that most enterprises are negotiating it for the first time, while the account team has run it repeatedly since launch. That asymmetry is the gap independent advisory exists to close. An advisor who has seen assist allowances and overage terms across a range of recent enterprise renewals brings pattern recognition to a team facing the consumption model once.
Independence matters because the model rewards consumption growth, and the parties around the deal tend to benefit when consumption rises. An advisor with no vendor partnership, no reseller margin and no implementation revenue has only one incentive: the allowance, overage rate and tier you actually need. That alignment is what keeps the consumption model from quietly inflating your cost across the term.
For the supporting service, see our Now Assist consumption advisory, and when you are ready to model your own consumption against benchmark data, the ServiceNow renewal negotiation service is the place to start.
Questions
It is the 2026 model in which Now Assist features consume metered assists. AI is bundled into every tier with an included assist allowance, and consumption above that allowance triggers overage top up charges.
Agentic actions plan and execute multi step tasks autonomously, which performs much more work than a routine summary or suggestion. At production volume that multiplier is the main driver of consumption cost, so agentic workflows should be weighted heavily when modelling usage.
Model your projected assist consumption with headroom, negotiate an allowance that fits, and pin the overage rate in writing before signing. Overage charges are far less negotiable after signature, so the allowance and rate should be settled up front.
Yes. The allowance and overage rate are real commercial terms, and comparable enterprises with a similar workflow profile provide the benchmark. Based on benchmark observations, the allowance and overage rate are often where the most value is available, because buyers focus on the tier price.
NowNegotiations Advisory Team. Independent ServiceNow negotiation advisors, buyer side in hundreds of enterprise software negotiations. Guidance based on real enterprise renewal engagements. Last updated 30 March 2026.