Now Assist arrives as two charges, and most buyers model one
Now Assist is ServiceNow's AI family across ITSM, CSM, and HRSD, and the bill has two layers: a Pro Plus or Enterprise Plus tier uplift of 25 to 40 percent on the base SKU applied to every licensed subscription unit, then a committed credit pool that draws down as people use it. Across our engagements the AI was sold as a platform wide upgrade while the measurable value sat in a handful of workflows, and the first year commitment was set before anyone had a burn rate.
Prepared by Redress Compliance · August 8, 2026 · ServiceNow advisory. Based on 30 to 40 ServiceNow estates advised or benchmarked on Now Assist 2024 to 2025.
Executive summary
The tier uplift is the larger charge, and it applies to everyone, not just AI users.
Only Pro Plus and Enterprise Plus carry Now Assist, Standard and base Pro carry none.
And the step up runs 25 to 40 percent on the base SKU per subscription unit across the whole licensed base: applying it for narrow AI use lifted the platform bill 30 to 60 percent in our files, which makes the scope of who gets the tier the largest single cost decision.
Settled long before anyone opens the credit model.
The Enterprise Plus delta, platform analytics and governance tools, is real and often irrelevant: price Pro Plus as the default and make the account team defend the delta feature by feature.
The worked model: $550 per fulfiller in year one, both layers.
A 1,000 fulfiller ITSM estate models at $550,000 in year one at full deployment, $300 per fulfiller of tier uplift plus $250 of consumption, holding flat per head across pilot, phase two, and full rollout, with a CSM expansion adding another $300,000.
Run the uplift backwards as a sanity check: $300 at 40 percent implies a $750 base SKU, at 25 percent implies $1,200, and a quoted base outside $750 to $1,200 means the uplift percentage you were quoted is not the one in the model, a gap worth raising in the room.
The credit pools were over committed 20 to 50 percent, because forecasts replaced measurement.
Generative actions, summaries, drafted responses, natural language search, draw from the committed annual pool with overage billing on top, and first year pools ran 20 to 50 percent over measured burn while broad enablement burned credits 2 to 3 times faster than staged rollouts.
The buyers who measured one workflow first cut their committed pool 25 to 40 percent, which makes the staged path, one workflow, then a two product pilot, then breadth, the only sizing method that prices from evidence.
The contract terms decide whether the AI can ever scale down.
Where the Plus uplift was bundled into the multi year platform commitment, the option to reduce AI spend later was gone, so the paper matters as much as the sizing: the telemetry obligation, the true down right, and the action metering definition in writing, plus negotiated credit rate, pool size.
And rollover terms.
BYOLLM shifts some model control but not the platform charge and is not an alternative to Now Assist, and the credit pool runs as a budget with a named owner or it runs as an overage generator with nobody watching.
The commitment scenarios, both layers priced
| Scenario | Tier uplift | Estimated consumption | Total year one |
|---|---|---|---|
| Pilot, 100 fulfillers | $30,000 | $25,000 | $55,000 |
| Phase 2, 400 fulfillers | $120,000 | $100,000 | $220,000 |
| Full deployment, 1,000 fulfillers | $300,000 | $250,000 | $550,000 |
| Full deployment plus CSM expansion | $450,000 | $400,000 | $850,000 |
The per head rate is the model's gift: it holds flat, so the scenarios compare cleanly. $300 per fulfiller of uplift and $250 of consumption in every ITSM row means the pilot prices the full deployment honestly, and the backwards check on your own quote, the implied base SKU between $750 and $1,200.
Catches the deals where the quoted uplift percentage and the quoted base do not reconcile.
These are deal model numbers, not a price list: put your quoted base through the same arithmetic before the account team frames the conversation as a rate discussion.
The rollout approaches, by credit risk
- Staged, one workflow: low risk and measured, the first year default, and the method that cut committed pools 25 to 40 percent.
- Two product pilot: moderate risk, earned by a proven first workflow with a real burn rate behind it.
- Broad enablement: high risk and hard to forecast, burning 2 to 3 times faster than staged rollouts, justified only on a mature, measured estate.
- The tier scope follows the rollout: the Plus tier goes to the teams in the staged plan, not the whole base, because the uplift for narrow use lifted bills 30 to 60 percent.
The Now Assist AI pricing brief
The two layer model, the credit mechanics, the staged rollout method, and the contract terms worked on a representative estate.
Get the white paper →The contract terms that keep the exit open
Three terms belong in writing before any commitment: the telemetry obligation, ServiceNow providing the consumption reporting that makes the burn rate measurable at all.
The true down right, the ability to reduce the committed pool and the tier scope at renewal, which vanished wherever the uplift was bundled into the multi year platform commitment.
And the action metering definition, exactly which generative actions draw credits and at what rate, because an undefined meter is an unbounded bill.
The negotiable numbers ride alongside: the credit rate, the pool size evidenced by the staged rollout, and rollover terms that stop an over committed pool expiring worthless.
The consumption mechanics and overage patterns in production run in the Now Assist consumption analysis, the seat layer underneath in the ITSM pricing guide, and the renewal calendar the whole decision lands in via the renewal playbook.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
What we saw across Now Assist engagements, 2024 to 2025
Across roughly 30 to 40 ServiceNow estates we advised or benchmarked on Now Assist between 2024 and 2025, the AI was sold as a platform wide upgrade while the measurable value sat in a handful of workflows:
From applying the Plus tier uplift across the full base for narrow AI use.
Credit consumption against staged rollouts, with the forecast error billed as overage.
The sequencing error was universal: the first year commitment was set before anyone had a burn rate, which handed the sizing to the vendor forecast, and the correction is procedural rather than clever, stage the rollout, measure one workflow, and size both layers from the telemetry.
BYOLLM deserves its honest sentence too, bringing your own model shifts some control over the inference layer but the platform charge remains, so it is a governance choice, not a cost escape.
The credit pool is a budget: it gets a named owner, a monthly burn review, and rollover terms, or it becomes the line item that surprises finance at the first true up.
Your first five moves
- Scope the Plus tier to the staged rollout, not the base, against the 30 to 60 percent lift narrow use paid for.
- Measure one workflow before committing the pool, the method that cut commitments 25 to 40 percent.
- Run the backwards check on your quote: the implied base SKU between $750 and $1,200, or ask about the gap.
- Put telemetry, true down, and metering definitions in writing, and keep the uplift out of the multi year bundle.
- Negotiate credit rate, pool size, and rollover, with a named owner on the burn. The ServiceNow practice runs the deal with you.
Frequently asked questions
How is ServiceNow Now Assist priced?
In two layers: a tier upgrade, since only Pro Plus and Enterprise Plus carry Now Assist, running 25 to 40 percent on the base SKU per subscription unit across the licensed base; and a committed annual credit pool that generative actions draw down, with overage billing on top.
Most buyers model one layer and get billed for both.
What does Now Assist cost for a typical estate?
Our deal model for a 1,000 fulfiller ITSM estate lands at $550,000 in year one at full deployment, $300 per fulfiller of tier uplift plus $250 of consumption, holding flat per head from pilot through full rollout, with a CSM expansion adding roughly $300,000 more.
The backwards check: $300 of uplift implies a base SKU between $750 and $1,200 depending on the percentage.
Which ServiceNow tiers include Now Assist?
Pro Plus and Enterprise Plus only: Standard and base Pro carry no AI.
Enterprise Plus adds platform analytics, advanced performance analytics, and governance tools over Pro Plus, a real delta that is often irrelevant to the buyer, so the default position is Pro Plus with the account team defending the Enterprise Plus delta feature by feature.
How do Now Assist credits work?
Generative actions, case summaries, drafted agent responses, and natural language search, draw from a committed annual pool at defined rates, with usage drawing down the pool and overage billing on top.
First year pools were over committed by 20 to 50 percent against measured burn in our files, because commitments were set from vendor forecasts before anyone had a burn rate.
Should Now Assist be rolled out broadly or staged?
Staged: broad enablement across all products raised consumption 2 to 3 times faster than staged rollouts and is hard to forecast, while buyers who measured a single workflow first cut their committed pool 25 to 40 percent.
One workflow, then a two product pilot on proven adoption, then breadth on a mature measured estate, with the Plus tier scoped to the same stages.
Is BYOLLM an alternative to Now Assist?
No: bringing your own large language model shifts some control over the inference layer but the platform charge remains, so it is a governance and data choice rather than a cost escape.
The cost controls live elsewhere, the tier scoped to actual AI users, the pool sized from measured burn, the true down right preserved outside the multi year bundle, and the metering definition in writing.
