ServiceNow ITSM Foundation, Advanced and Prime compared. Who runs agents decides the tier.

All three ServiceNow packages include the same AI floor. They differ in agentic capability, process scope and assist allowance, so the right tier depends on who actually runs agents, and in our benchmarks that was rarely everyone.

What are ServiceNow ITSM Foundation, Advanced and Prime?

They are the three AI native packages that replaced Standard, Pro, Pro Plus, Enterprise and Enterprise Plus on April 9, 2026. All three bundle the same AI portfolio: Now Assist generative AI, the Moveworks layer, Workflow Data Fabric and AI Control Tower.

What separates them is how much agentic capability each one includes and how large an assist allowance it ships with. AI is present at every level, so the old argument for climbing to Pro Plus to get AI no longer applies. The climb is now sold on autonomy.

What does Foundation include?

ServiceNow's ITSM pricing page lists Incident, Service Catalog and Request, Asset Management and CMDB, Virtual Agent, Workflow Data Fabric and Now Assist at this level. The AI works inside the ticket, summarizing the record and drafting the resolution note. For a desk whose work is tickets in and resolutions out, that is the whole requirement.

What does Advanced add?

Advanced adds the rest of the ITIL process set (Change, Problem, Major Incident and On Call) along with AI Voice Agents, Process Mining and Platform Analytics Advanced. On the AI side it goes from assisting a person to executing tasks: an agentic workflow can take a step, check the result and continue without a human pressing each button.

What is only available in Prime?

Prime is the only package with fully autonomous agents and the right to build your own AI skills and agents. For ITSM it adds AI Agents for ITSM, the L1 Service Desk AI Specialist, an AI agent for digital end user experience and DevOps Change Velocity.

If no one in your organization will build or run agents in this term, and no team needs DevOps Change Velocity or Digital Product Release, Prime buys nothing you will use.

How do the legacy tiers map to the new ones?

The mapping is not one to one, because the process boundaries moved. Under the old packaging, Standard covered incident, problem and change. Under the new packaging, Problem and Change sit in Advanced, so a desk on legacy Standard that runs change management in ServiceNow maps to Advanced, and the conversion quote carries a tier step.

Legacy SKUs reached end of sale on July 1, 2026. Existing contracts keep their SKUs until a renewal, a new purchase or another contract change converts them, so most customers meet the new packages at renewal. Our tier migration cost guide shows what that step usually does to the quote.

Which tier does each of your teams actually need?

Each team needs the lowest package whose capabilities it uses today or on a dated plan, and the deciding question is who runs agents. The shared AI floor already serves everyone who might benefit from AI, and a mention on the transformation roadmap does not count as a plan.

Across our benchmarks the answer was narrow. A small build and run team needed Prime. Operations populations needed Advanced where task execution, voice and the wider process set applied to their work. The incident and request teams, often the largest single population, were correctly served by Foundation.

Why is Foundation so often undersold?

Account teams push pure incident and request teams past Foundation because the higher packages carry a higher rate. For a team that never opens a change or problem record, every seat moved to Advanced pays for capabilities its queue never invokes.

Under the legacy packaging, Now Assist started at Pro Plus, so a Standard desk had no generative AI. Foundation now includes Now Assist, which removes the old reason to climb for AI alone.

Can you put different teams on different tiers?

Usually not within one product line. ServiceNow has generally required one package for all licensed fulfillers of a product line such as ITSM, with a few exceptions like App Engine, SecOps and IntegrationHub. If one team needs Change in ITSM, every ITSM fulfiller is on Advanced.

The team by team map still decides the money. It shows which package each product line needs, how few people drive a proposed step to Prime, and where product lines such as ITSM, HR Service Delivery and Customer Service Management can sit on different packages.

How do you build the team by team map?

Build it from instance data before the account team sends its own mapping. The method is the same one we use in the fulfiller reconciliation, run against the new packages. The 2026 pricing tiers pillar has the packaging detail for the other product lines.

• Fulfiller roster. Pull sys_user_has_role for itil and the other fulfiller roles, and remove inactive users in sys_user. This is the seat count every package is priced on.
• Process use. Count who worked change_request and problem records over twelve months through the assigned_to field, and check whether any incidents were promoted to major incidents (the major_incident_state field). If those counts are zero for a product line, Foundation is on the table.
• Agent builders. List by name who builds or runs agents and custom skills in AI Agent Studio and the Now Assist Skill Kit, and which of those agents run in production.
• AI consumption. Read assist usage by skill and instance in Subscription Management. Anyone with usage_admin or admin can see it.

How do the assist pools change the total cost?

Every package ships an assist allowance and bills overage per unit past it, which makes the pool a sizing decision separate from the package. A larger pool bought through a higher package only pays if measured usage fills it, and overage on a lower package only hurts if it exceeds the price difference between packages.

The allowance is pooled at tenant level, so development and test instances draw from the same pool as production. A multi step agentic action consumes far more assists than a summary. Our note on consumption and overage in 2026 covers the meter in detail.

What does the pool arithmetic look like in practice?

Say you license 1,000 ITSM fulfillers. Assume monthly rates of $90, $120 and $160 per fulfiller for Foundation, Advanced and Prime, allowances of 1,500, 3,000 and 6,000 assists per fulfiller a year, and a top up rate of $0.08 per assist. None of these are ServiceNow prices. A pilot shows annual burn of 2,000,000 assists.

The extra pool is worth at most the extra assists times your top up rate. Moving from Foundation to Advanced adds 1,500 assists a seat, worth $120 a year at $0.08, against a price step of $360 a year. From Advanced to Prime the extra 3,000 assists are worth $240 against a $480 step.

In this example the larger pool never pays for itself. The top up rate would have to pass $0.24 per assist before Advanced's pool covered its own step, and $0.16 before Prime's did. Capability has to justify the step, and the capped top up rate sets how much the pool can ever be worth.

Price the package on what your teams do in it, and size the pool on what your pilot burned.

Prime also costs $480,000 a year more than Advanced here. If 40 people build and run agents, that is $12,000 a year for each of them, a figure the agent program's sponsor should see before anyone signs.

How should you size the pool?

Run the pilot before you commit to a package. Once real consumption exists the sizing is mechanical, and without it buyers pay for AI that sits on the shelf.

1. Measure burn. Run the skills you plan to use with a pilot group for at least a quarter and read consumption by skill.
2. Scale it to the rollout. Multiply by the number of people you will activate each month, which in year one is lower than the licensed count.
3. Phase the rollout. License 50 to 60 percent of the target population in year one and grow into the rest at a price fixed now.

What should the order form say before you migrate?

It should price the package, the AI uplift and the assist allowance on separate lines, because a single blended number hides which part rose. It should also fix the terms that decide what the meter costs you.

• Three line quote. Package, AI uplift and allowance priced separately, so you can compare the conversion with your legacy price.
• Allowance from measured burn. The pool size stated in assists and tied to the pilot, with a right to resize it at each anniversary.
• Capped overage rate. A per assist top up price fixed for the whole term, including any prepaid top up pack.
• Meter audit rights. Access to consumption data by skill and instance, and a way to dispute usage from misconfigured or looping agents.
• Next package option. The price of stepping a product line from Foundation to Advanced, or Advanced to Prime, at any anniversary.
• Conversion trigger. A written statement of which events move legacy SKUs to the new packages, so a small purchase mid term does not force an early conversion.

The migration also reprices everything else on the contract, so negotiate it as part of the wider renewal, where the package mix can be traded against price, term and quantity.

What will the ServiceNow account team say about the tiers?

Expect the pitch to lead with the bundled AI and the agent roadmap, and to treat the package mapping as settled. These are the lines we hear most, with replies.

• "AI is in every package now, so the uplift reflects value you already receive." Ask for the three line quote and the usage data behind the uplift. The AI floor was bundled for every customer, so price it against what your teams consume.
• "Companies your size run Advanced as standard." Show the process counts. If a product line has no change or problem work in ServiceNow, Foundation covers it.
• "Prime's larger pool means you will never pay overage." Multiply the extra assists by the capped top up rate and set the result against the package step. Unless your top up rate is high, the extra pool will not close that gap.
• "The mapping to the new packages is fixed." Treat the mapping as a proposal. The renewal reopens price, term and quantity, and your process and agent data decide the package.

Should you buy Prime now so you are ready for agents?

Account teams and some internal sponsors often advise taking Prime across the product line now, to lock the rate and avoid a second negotiation. We advise against it. A product line wide Prime rate charges every fulfiller for autonomy that a small team uses.

Buy the package your current, dated requirements support, and write the Prime step into the contract as an option priced now and exercisable at any anniversary. If the agent program arrives on schedule, you exercise it. If it slips, you have paid nothing for it.

What have we seen in ServiceNow tier comparisons in 2025 and 2026?

Across the ServiceNow renewals Fredrik Filipsson and the team benchmarked from 2025 to 2026, the tier comparison came down to the agent question almost every time, and the answers followed the same pattern.

• The Prime population was small. The people who needed full autonomy and custom skills were a build and run team in the tens, set against seat counts in the thousands.
• The pool decided more money than the package price. Consumption allowances set against real usage, with overage billed past every pool, moved the total further than the difference between package rates.
• Foundation was pushed aside without evidence. Account teams moved pure incident and request teams past the package that served them, and the good news of the bundled AI became an upsell that went unchecked.

The customers who held their totals ran the team by team map first and let the packages follow the evidence. Accepting the vendor's mapping as the starting point means every later discount is taken off its number instead of yours.

What to do next

1. Twelve months out, answer the agent question with names. List who builds and runs autonomous agents, who executes agentic tasks and who needs generative assist alone.
2. Nine months out, map each product line to a package. Use role, process and assist data from the instance, and hold Foundation wherever a pure incident and request population is served by it.
3. Nine months out, scope Prime to the build and run team. Refuse a product line wide Prime proposal the first time it appears, and ask for a priced Prime option instead.
4. Six months out, size the pools from piloted burn. Cap the overage rate in the order form and phase the rollout at 50 to 60 percent in year one.
5. Six months out, request the three line quote. Compare the package, the AI uplift and the allowance with your legacy price, line by line.
6. Three months out, run the migration inside the renewal. Trade the package mix against the whole contract. The ServiceNow practice and the rightsizing tool run the mapping with you.

Frequently asked questions