Now Advisory · Buyer side guide · 2026 edition
Now Assist budgeting: a buyer side guide
Now Assist budgeting is the discipline of planning a metered line that moves with usage. This guide shows how to budget AI consumption as a range, fund the overage tail and avoid the surprise top up, with benchmark data from real enterprise renewals.
Section 01Why Now Assist budgeting is different
Now Assist budgeting is different from licence budgeting because the cost moves with usage. In the 2026 model, AI is bundled into the tier and metered in assists, so the committed pool is a fixed line but consumption beyond it is variable, billed at the overage rate as top up charges. A budget that treats AI as a flat annual figure will be wrong in one direction or the other; the discipline is to plan a line that has both a fixed floor and a variable tail.
The variable tail is where budgets fail. Because large agentic actions consume materially more assists than simple generative requests, consumption tends to rise as adoption deepens and agentic workflows move into production. A budget set at signature against early, mostly generative usage will be exhausted by production adoption, and the gap arrives as an unbudgeted top up at the worst moment. Now Assist budgeting plans for that tail rather than being surprised by it.
This guide sits under the Now Assist pricing pillar and pairs with our Now Assist consumption advisory. It treats the AI line as a managed range, not a single number.
Section 02Budgeting a metered line
A metered line is budgeted in two parts: the committed pool, which is fixed and known, and the expected overage, which is variable and modelled. The committed pool is the prepaid floor; it is the assists you have already bought as part of the tier. The expected overage is the consumption beyond the pool, priced at the overage rate, and it is the part that has to be forecast rather than read off the contract.
Budgeting only the floor is the common error: it produces a tidy number that the first production quarter overruns. Budgeting the floor plus a modelled overage tail produces a range that finance can plan against, with a central estimate and a contingency for higher adoption. The range is more useful than a point because it tells finance the likely cost and the worst plausible case, which is what a metered line actually presents.
The most common Now Assist budget failure is funding the committed pool and nothing beyond it, then meeting production adoption with no allowance for the overage tail. The tail, not the pool, is where budgets break.
Section 03Modelling the consumption range
The consumption range comes from modelling assist draw across a realistic adoption curve rather than a single point. The inputs are the workflows adopting AI, the volume each runs, and the assist weight of each, with agentic actions weighted heavily. Running this model under conservative and optimistic adoption produces a band of likely consumption, which translates directly into a band of likely cost once priced against the pool and overage rate.
That band is the budget. The conservative end funds the expected case; the optimistic end sizes the contingency. Crucially, the model also shows the inflection point where consumption crosses the pool boundary into overage, which is the moment the marginal cost of an assist changes. Knowing where that boundary sits tells finance when the variable tail begins to bite. The modelling method is detailed in our Now Assist consumption advisory.
Because adoption is the main driver, the range should be revisited as adoption data arrives, turning the budget into a living forecast rather than a one time estimate.
Section 04Funding the overage tail
The overage tail is the consumption beyond the committed pool, and it has to be funded explicitly rather than absorbed by surprise. The size of the tail depends on two things: how far adoption pushes consumption past the pool, and the overage rate the consumption is billed at. A budget that funds a contingency for the tail, sized from the optimistic end of the consumption model, meets production adoption with a plan rather than a crisis.
The overage rate is therefore a budgeting variable, not just a negotiation one. A high or open overage rate makes the tail expensive and uncertain; a fixed, negotiated rate makes it calculable. This is why the rate should be fixed at signature: an open rate leaves the most variable part of the budget priced by the vendor mid term. Our Now Assist overage guide covers how to bound the rate before it bites.
Funding the tail is not pessimism; it is recognising that successful AI adoption is precisely what generates overage, so a growing tail is often a sign the investment is working and should be planned for accordingly.
Section 05Sizing the committed pool
The committed pool is a budgeting decision as much as a negotiation one. Too large a pool is prepaid waste: assists bought and not consumed, raising the effective price of every assist actually used unless rollover protects them. Too small a pool pushes consumption into overage sooner, increasing the variable tail. The right size sits where the prepaid pool and the modelled consumption meet, with a margin chosen deliberately rather than by default.
Rollover changes this calculus materially. With rollover or true forward treatment, a conservative pool is protected because unused assists carry forward to offset later peaks, so under sizing is less punishing. Without it, the pool has to be sized closer to expected consumption to avoid waste. The budgeting decision and the rollover term are therefore linked, and both belong in the renewal. The mechanics connect to our Now Assist true up guide.
Size the committed pool from the consumption model, not the vendor's adoption hopes, and pair it with rollover so a conservative pool is protected rather than wasted.
Section 06Governing the budget through the term
A Now Assist budget is only as good as the monitoring that maintains it. Because consumption is observable, a quarterly review of actual assist draw against the budgeted range shows early whether the estate is tracking toward the conservative or optimistic case, in time to adjust funding or workflow design rather than discovering the overrun in an invoice. The organisations that get caught are usually those that budget at signature and never compare actuals to the plan.
Governance also turns the mid term resize right into a budgeting tool: if consumption settles well above or below the pool, the commitment can be adjusted so the budget and the contract stay aligned. Monitoring closes the loop between the forecast and reality, keeping the variable line under control. The monitoring practice is detailed in our Now Assist usage monitoring guide.
Treat the AI budget as a forecast to be revised quarterly, not a number to be set once, because the thing it measures changes every quarter.
Section 07The Now Assist budget, line by line
A defensible Now Assist budget reduces to a small set of lines, each modelled rather than guessed.
- Committed pool
The prepaid floor, sized from the consumption model and paired with rollover.
- Expected overage
The central estimate of consumption beyond the pool, priced at the fixed overage rate.
- Contingency tail
The allowance for the optimistic adoption case, funding the variable tail rather than absorbing it by surprise.
- Boundary inflection
The point where consumption crosses the pool into overage, marking where marginal cost changes.
- Quarterly revision
The monitoring cadence that moves actuals against the range and triggers funding or resize adjustments.
Built this way, the budget is a range with a plan for its tail, which is the only honest shape for a metered line.
Section 08The Now Assist budgeting checklist
Before finalising the AI budget, confirm each item below from the consumption model rather than a flat estimate.
- Consumption modelled as a range across conservative and optimistic adoption.
- Committed pool sized from the model and paired with rollover.
- Expected overage estimated and priced at a fixed overage rate.
- A contingency funded for the optimistic adoption tail.
- The pool to overage boundary identified so marginal cost changes are visible.
- A quarterly monitoring cadence in place to revise the forecast.
If the budget is a single number with no tail, it is a licence budget applied to a metered line, and it will be wrong by the first production quarter. Plan the range and fund the tail.
FAQFrequently asked questions
Why is Now Assist budgeting different from licence budgeting?
Because the cost moves with usage. AI is bundled and metered in assists, so the committed pool is a fixed floor but consumption beyond it is variable, billed at the overage rate. A flat annual figure will be wrong; the budget needs both a fixed floor and a modelled variable tail.
How do we budget the overage tail?
Model assist consumption as a range across conservative and optimistic adoption, identify where it crosses the committed pool, and fund a contingency sized from the optimistic end. Fix the overage rate at signature so the tail is calculable rather than priced by the vendor mid term.
How large should the committed pool be?
Size it from the consumption model, not vendor adoption hopes. Too large is prepaid waste; too small pushes consumption into overage sooner. Pair the pool with rollover so a conservative size is protected by carrying unused assists forward rather than losing them.
How do we keep the budget accurate through the term?
Review actual assist draw against the budgeted range quarterly, so you can see early whether the estate is tracking to the conservative or optimistic case and adjust funding, workflow design or the commitment through a resize right before an overrun appears in an invoice.
Are these figures official ServiceNow prices?
No. All ranges are typical negotiated figures based on benchmark observations across real enterprise renewals, used as internal leverage rather than published as official list prices.