The SKUs your contract names ended sale on 1 July 2026, and the replacement is quoted at Prime

On 9 April 2026 ServiceNow replaced five module tiers with three capability tiers, and set legacy end of sale for 1 July. That is not a rename. It retires the SKUs your contract was written against, so every customer renegotiates at their next renewal, on paper the vendor drafted.

Prepared by Redress Compliance · August 16, 2026 · ServiceNow advisory. 30 to 40 renewals benchmarked, 2025 to 2026.

Two dates drive planning. The 9 April 2026 launch made the new tiers available. The 1 July 2026 end of sale means any renewal quoted after that point is written on Foundation, Advanced, or Prime. Contracts mid term keep their legacy SKUs until renewal, and not one day longer.

• Bundled into every tier: Now Assist generative AI, the Moveworks layer, Workflow Data Fabric, and AI Control Tower governance. None of these is sold separately any more, which is the mechanism that raised the base.
• Foundation is the floor: incident, request, and asset management with the CMDB, Virtual Agent, and Now Assist for summarization, insight, and data extraction. Right for teams that want assistance but not autonomous execution.
• Advanced adds execution: change and problem management, AI voice agents, process mining, on call management, and higher assist allowances, supporting agents that carry out specific defined actions.
• Prime is the only tier with autonomous agents, custom AI skill building, and the Level 1 service desk AI specialist. That is a labor substitution pitch, and it is priced accordingly.
• Benchmarked deals run roughly $70 to $100 per user on Foundation and $160 to $200 or more on Prime, before any consumption. List pricing remains unpublished, which makes independent comparison the only reference point you have.
• The tier name now signals ambition, not module breadth. Pay for the capability you will run, not the capability the demo showed, and scope Prime to the population that builds and runs agents.

The standard reseller advice is to standardize the whole estate on Prime so every user is agent ready. We disagree, and the benchmarking is unambiguous. In roughly six of ten estates reviewed across 2025 and 2026, autonomous agents were needed by a small operations team rather than the whole company, and full Prime standardization inflated the bill by about a third with no added value attached to it. The pitch is attractive because it is simple to quote, simple to administer, and simple to forecast. It is expensive for exactly the same reason: it prices ambition uniformly across a population whose ambition is not uniform.

The second cost is quieter and harder to see at signature. Each tier ships a bundled pool of assist capacity, and once that pool is spent, top up packs bill at a per unit rate. Overage was quoted as a rounding error in the deals we reviewed and landed at 15 to 30 percent of tier spend once agents and Now Assist actually went live. That is not a modeling error on ServiceNow's side. It is the predictable result of pricing a fixed component that buyers scrutinize alongside a variable component that buyers accept on assurance.

Underneath both sits a change in behavior that legacy licensing never punished. Development and sub production instances draw on the same assist pool as production, so a busy test cycle moves the annual bill on its own. Teams that iterate on agents burn the pool without noticing, because under the old model a development instance carried no incremental cost and nobody was watching. Seat based forecasting understates the bill every time, since consumption scales with what users and agents actually do rather than with how many licenses exist.

So the buyer side move has three parts and none of them is a discount request. Buy Prime for the agent building and agent running population, hold the rest on Foundation or Advanced, and negotiate a documented right to expand Prime later at a locked rate. Build an actions based forecast that separates interactive prompts, agentic tasks, and autonomous runs, then add non production load on top of it. And treat the migration itself as the moment to secure price protection, because it is the one renewal where the vendor needs your signature on new paper. The tier comparison sits in Foundation, Advanced, and Prime compared, the meter mechanics in Now Assist consumption and overage, and the wider library in the ServiceNow practice.

Across roughly 30 to 40 ServiceNow renewals benchmarked once the new tiers hit the table, the tier price is the part buyers focus on and the consumption meter is the part that moves the bill:

The third pattern was the one nobody modeled. Development and sub production assist usage drew on the same pool as production, and it was rarely priced before signature. Model dev and test consumption alongside production before you agree a pool size, because the pool is the only buffer between you and the per unit rate.

The renewal levers on the new packaging are tier fit, consumption pool size, overage unit rate, and price protection on the migration. Three of those four are invisible if you negotiate the per user price alone.

Watch the briefing · 4:17ServiceNow's New AI: Assists, Tiers, and the Meter You Are About to SignHow the three tiers bundle Now Assist and where the consumption meter starts billing.

What changed in ServiceNow pricing in April 2026?

On 9 April 2026 ServiceNow replaced its five legacy tiers with three AI native tiers and set legacy end of sale for 1 July 2026. The retired tiers were Standard, Pro, Pro Plus, Enterprise, and Enterprise Plus. The replacements are Foundation, Advanced, and Prime, ordered by AI capability rather than module count.

Why does the repackaging force a renegotiation?

Because it retires the SKUs your contract was written against. Contracts mid term keep their legacy SKUs until renewal, but any renewal quoted after 1 July 2026 is written on Foundation, Advanced, or Prime. Every customer therefore renegotiates at their next renewal, on paper the vendor drafted.

What is bundled into every tier?

Now Assist generative AI, the Moveworks layer, Workflow Data Fabric, and AI Control Tower governance are bundled into all three tiers rather than sold separately. That bundling is the mechanism that raised the base price, and it is why the entry tier costs more than Standard did.

Where do autonomous AI agents sit?

In Prime only. Prime is the sole tier that permits fully autonomous agents, custom AI skill building, and the Level 1 service desk AI specialist. That gating is the most consequential design choice in the repackaging, because it is what the upsell is built on.

Should we standardize the whole estate on Prime?

Usually not. In roughly six of ten estates benchmarked, autonomous agents were needed by a small operations team rather than the whole company, and full Prime standardization inflated the bill by about a third with no added value. Scope Prime to the population that builds and runs agents.

How much does consumption overage add?

It was quoted as a rounding error and landed at 15 to 30 percent of tier spend once agents and Now Assist went live. Each tier ships a bundled assist pool; past the pool, top up packs bill at a per unit rate, so the annual bill has two moving parts rather than one.

Does development usage count against the pool?

Yes. Sub production and development instances draw on the same pool as production, which is a real change from legacy licensing where a development instance carried no incremental cost. A busy test cycle can move the annual bill on its own, so model non production load before agreeing a pool size.

What are the renewal levers on the new packaging?

Four: tier fit per population, the size of the bundled consumption pool, the overage unit rate, and price protection on the migration. Negotiating the per user price alone leaves three of the four untouched, and those three are where the variable cost lives.