Case study · Retail · Now Assist
This ServiceNow retail Now Assist cost control case study shows how a national retailer turned a padded AI forecast and an open metered consumption line into a capped assist pool with a fixed overage rate, using benchmark data from real enterprise renewals.
Reduction on the metered Now Assist line
Saved over the three year term
Overage rate fixed at signature with rollover
A national retailer with a mature ServiceNow estate reached renewal under the 2026 commercial model and was handed a Now Assist forecast that sized the committed assist pool well above anything the platform had used. The account team framed the number as conservative, built from projected adoption across service and HR workflows. The retailer brought us in buyer side to test the forecast against benchmark data from real enterprise renewals before it became a multi year commitment.
The estate had grown into customer service and HR over two terms, and the AI story was genuinely attractive to the business. That enthusiasm was the risk. Internal sponsors wanted Now Assist switched on broadly, and the vendor forecast turned that intent into a large committed pool of metered assists with an open overage rate for anything beyond it. With a renewal date inside two months, the default was to accept the AI line as the cost of moving fast.
Our first task was to separate the appetite for the capability from the size of the commitment. We mapped the proposed assist pool against actual consumption signals, weighted the agentic actions that draw the pool down fastest, and broke the AI line out of the bundle so it could be sized on evidence rather than on enthusiasm.
The forecast counted assist actions without weighting them. Large agentic actions, the kind that resolve a case end to end, consume materially more assists than a simple generative reply, and the model treated both as equal. That understated real consumption per workflow while oversizing the committed pool to cover a usage curve the retailer would not reach for at least a year. The overage rate sat open, so any month above the pool would bill at an undefined top up price. In short, the retailer was being asked to overcommit on volume and stay exposed on rate at the same time.
We built the strategy around four moves, sequenced so the consumption model was agreed before the commercials. First, a weighted consumption model that priced agentic actions separately from generative ones, set out in our guidance on Now Assist consumption. Second, a right sized assist pool matched to a realistic first year adoption curve rather than a full rollout. Third, an overage rate fixed at signature, so any consumption beyond the pool billed at a known top up price rather than an open one. Fourth, rollover for unused assists, so a conservative pool carried no penalty for underuse.
The retailer's team led every conversation. We stayed behind the table, reviewing each proposal revision, modelling the consumption math, and briefing executives before each session, in the pattern set out in our ServiceNow renewal negotiation advisory.
The agreement signed three weeks before deadline. The Now Assist line closed roughly 31 percent below the opening quote, with the committed pool sized to a realistic first year curve, the overage rate fixed at signature, and rollover secured for unused assists. The retailer kept the option to scale the pool upward in later years on benchmarked terms rather than locking the full rollout in on day one. Across the three year term the metered AI line saved the retailer in the region of 1.4 million dollars. The mechanics behind the consumption work are set out in our broader ServiceNow renewal guidance and in our spoke on Now Assist overage exposure.
Three lessons carry beyond this engagement. A metered AI line must be sized from a weighted consumption model, because counting agentic and generative actions as equal oversizes the pool and understates the real draw at the same time. Business enthusiasm for a capability is not a reason to commit to a full rollout in year one, because the pool can scale on benchmarked terms as adoption proves out. And an open overage rate is the most expensive line in any Now Assist agreement, so fixing it at signature is worth more than a headline discount on the pool.
The retailer reduced a padded Now Assist forecast to a weighted consumption model, capped the committed assist pool, and fixed the overage rate at signature. The metered AI line closed roughly 31 percent below the opening quote with rollover secured for unused assists.
The case study is anonymised. It is based on real enterprise renewal engagements, with the client profile, estate and figures presented as plausible and internally consistent ranges rather than naming any organisation.
Assists are metered, and large agentic actions draw the pool down materially faster than simple generative requests. A forecast that counts actions without weighting the agentic ones understates real consumption while oversizing the committed pool, so both the size and the overage rate need to be negotiated.
No. All figures are typical negotiated ranges based on benchmark observations across real enterprise renewals, used as internal leverage rather than published as official list prices.
Your renewal