ServiceNow: Prove the Value or Expand the Platform?
Download case PDF (5 pages)Request teacher’s pack
Decision date and information cutoff: July 24, 2024, 4:15:05 p.m. ET
Decision scope: Where to focus incremental product, infrastructure, and customer-success effort for Now Assist
Case convention: This is an author-created management decision exercise. The assignment and hypothetical role are not an actual ServiceNow proposal or private deliberation. Use only information publicly available by the stated cutoff.
The decision
On July 24, 2024, ServiceNow reported that subscription revenue had grown 23% year over year in the second quarter. The same release said net-new annual contract value (NNACV) for Now Assist, the company's generative-AI product family, had doubled sequentially. Eleven Q2 transactions had more than $1 million in Now Assist NNACV.
The signal was promising, but incomplete. ServiceNow did not disclose the absolute Now Assist NNACV base, how many customers had moved from evaluation to paid production use, how widely the features were being used, or whether productivity gains exceeded serving, implementation, and support costs. The doubling statistic described growth in net-new contract value—not recognized revenue, total AI revenue, or realized customer savings. [S1]
Assignment: Imagine you are advising a hypothetical ServiceNow strategy committee in July 2024. Recommend where the company should put its next increment of product and customer-success effort to make Now Assist a durable source of customer value and profitable growth. Choose or combine a focus: (1) deepen a few high-volume workflows and prove results; (2) broaden the product family quickly across workflows and customers; or (3) make the platform more extensible through customer-selected models and partner-built capabilities. Explain your choice, what evidence would change it, and what measures should govern the next stage. No actual budget, company decision, or board discussion is represented.
A business built around workflows
ServiceNow sells cloud software that connects people, processes, data, and devices through enterprise workflows. Its Now Platform spans information technology, employee and customer service, security, operations, and application development. Customers may use multiple products over time; the company's 2023 filing says many customers pursue multi-year digital-transformation plans that expand their use of products and services. [S2]
The model depends on subscription sales and ongoing service delivery. In 2023, ServiceNow recorded $8.68 billion in subscription revenue: $7.68 billion from digital workflow products and $1.00 billion from IT Operations Management (ITOM). The company says digital workflow products are generally priced per user, while ITOM offerings are predominantly priced by subscription unit. Its 2023 subscription gross margin was 82%. These are company-wide figures. They do not show the price, cost, margin, or return on a Now Assist customer or deployment. [S2]
ServiceNow has invested in product development, data-center capacity, and its partner network. Research and development expense was $2.12 billion in 2023, up 20% from 2022 and equal to 24% of revenue. R&D includes employee costs, infrastructure depreciation, and data-center capacity; the filing does not separate spending on generative AI. The company also works with partners on customer implementation and product integrations. Its 10-K recognizes competition from new technologies and entrants, alternative workflow approaches, and vendors that could bundle products or create pricing pressure. [S2]
From AI feature to product family
ServiceNow introduced Now Assist in 2023 and widened the portfolio in 2024. At its May Knowledge conference, it announced generative-AI features for IT service, customer service, human resources, IT operations, strategic planning, and application creation. Examples included summarizing cases, drafting knowledge articles, simplifying alerts, generating applications and playbooks, and helping users create strategic demands. These announcements varied in availability; the release described some as upcoming. [S3]
The product strategy allowed customer choice. Customers could use ServiceNow's Now AI models, bring their own large language model, or use a general-purpose model. ServiceNow also announced integrations with Microsoft Copilot and IBM watsonx, and described work with systems integrators and other partners. These announcements signal that ServiceNow saw value in combining its workflow context with outside models and services. They do not establish the commercial terms, customer use, or economics of those arrangements. [S3, S4]
ServiceNow also reported internal results. It said Now Assist was being used in more than 20 internal use cases and was producing roughly $10 million of annualized tangible benefits. The company described monthly improvements in self-service deflection and estimated that, if every IT agent around the world using ServiceNow used Now Assist, the productivity enhancement would equal one billion hours. The release gave no underlying method, baseline, cost, time period, or confidence range for those figures. The one-billion-hour figure was a forecasted scale scenario, not recorded savings. [S3]
The design challenge was not just choosing a model. Each workflow has different users, records, permissions, risk tolerances, and measures of success. A case summary may reduce time spent closing a ticket; a generated knowledge article could help users solve later cases themselves; an app-building assistant may increase the number of workflows customers create. Yet the value depends on adoption, the accuracy and usefulness of outputs, customer data quality, implementation work, human review, and who captures the economic benefit.
What the market signal did—and did not—say
The July 24 release reported Q2 subscription revenue of $2.542 billion, up 23% year over year. Current remaining performance obligations (cRPO) were $8.78 billion, up 22%; total remaining performance obligations (RPO) were $18.6 billion, up 31%. ServiceNow said NNACV for Now Assist had doubled sequentially and that 11 deals exceeded $1 million in Now Assist NNACV. Across all products, it reported 88 transactions above $1 million in net-new ACV, up 26% year over year. [S1]
The company also said its May Knowledge event had generated more than $1 billion in pipeline. Pipeline represents potential sales activity, not signed contracts or revenue. The earnings release did not give an absolute Now Assist NNACV figure, total AI ACV, number of paying AI customers, customer conversion, seat or use rates, renewal behavior, or AI product costs. [S1]
RPO is a contract measure, not an AI backlog. ServiceNow's 10-K explains that cloud subscription revenue is generally recognized ratably over contract terms that are typically 12 to 36 months. Contract timing and mix affect the period when ACV, RPO, and revenue appear. A larger contract value therefore does not establish in-period revenue, active use, or customer return. [S2]
Exhibit 1. ServiceNow's disclosed growth indicators
| Measure | Q2 2024 | Year-over-year change | What it measures |
|---|---|---|---|
| Subscription revenue | $2.542 billion | +23% | Recognized subscription revenue across product families |
| Total revenue | $2.627 billion | +22% | Subscription plus professional services and other revenue |
| Current RPO | $8.78 billion | +22% | Contracted revenue expected to be recognized in the next 12 months |
| Total RPO | $18.6 billion | +31% | Remaining contracted performance obligations |
| Company transactions above $1m NNACV | 88 | +26% | Company-wide net-new contract value metric |
| Now Assist NNACV | Doubled sequentially | Not disclosed | Net-new ACV for the product family; dollar base not disclosed |
| Now Assist deals above $1m NNACV | 11 | Not disclosed | Count of Q2 deals crossing the reported threshold |
Source: ServiceNow Q2 2024 earnings release. Dollar amounts and growth rates are company-reported. NNACV means net-new annual contract value; it is not recognized revenue, an AI-specific ARR figure, cash receipts, or profit. RPO and cRPO are company-reported contract measures, not AI-specific. [S1]
Learning from AI use at work
A 2023 working paper by Erik Brynjolfsson, Danielle Li, and Lindsey Raymond studied the staggered introduction of an AI conversational assistant at a Fortune 500 business-software company's customer-support operation. In the authors' preferred analysis, AI access increased resolved problems per hour by 0.30, about 13.8% from a baseline of 2.18. Results differed by worker group: the lowest pre-AI skill quintile improved about 35%, while the highest showed no productivity increase. Average handling time fell, and the authors did not find an economically significant average change in customer satisfaction. [S5]
The study involved one company's support agents and one task environment. It examined employee-facing recommendations, not ServiceNow's Now Assist portfolio or ServiceNow customers. Its estimates cannot be assumed to apply to ServiceNow, to other workflows, or to customer willingness to pay. The result does suggest questions for a deployment: Which task? Which user group? How will output quality be measured alongside speed? Does saved time become better service or lower cost? [S5]
Three ways to focus the next stage
1. Prove a few workflows deeply
Concentrate product engineering, implementation capacity, and measurement on a short list of repeatable workflows, such as IT service or customer service. Set customer baselines and measure deployment time, use, resolution, deflection, quality, and cost. This path could produce stronger evidence of value and make customer renewals or expansion easier to assess. It would leave some announced use cases less developed while competitors and customers move quickly.
2. Expand breadth across the platform
Continue extending Now Assist across workflows and user groups, supported by broad sales availability. The range of processes and installed customer relationships could make the platform more useful and reveal unexpected high-value uses. This would require capacity across many product teams and could spread measurement, support, and model-governance resources thinly. Public information does not show the incremental return or usage by workflow.
3. Prioritize extensibility and partners
Invest in customer-selected models, development tools, APIs, and partner capacity so customers can adapt Now Assist to their own data and processes. This could reduce reliance on one model and let partners solve specialized use cases. It may also add implementation and governance complexity, and could make it harder for ServiceNow to capture value if customers or partners perform much of the work. The company had announced partnerships and BYO model support, but had not disclosed partner economics or adoption. [S3, S4]
The choice
ServiceNow's core business was growing rapidly, and early Now Assist contract signals were positive. But the public evidence did not show which features customers were using, what outcomes they achieved, whether AI sales would recur or renew, or what it cost ServiceNow to serve them. The company had options: focus and learn, expand rapidly, or make its platform more extensible.
What should the hypothetical strategy committee prioritize? What should it measure before scaling further? What customer, product, and financial evidence would cause you to shift course?
Sources and information boundary
S1. ServiceNow, Second Quarter 2024 Financial Results, filed July 24, 2024, 4:15:05 p.m. ET (Exhibit 99.1 to Form 8-K, accession 0001373715-24-000269). The case cutoff is the SEC acceptance time. The earnings call occurred later and is excluded.
S2. ServiceNow, 2023 Form 10-K, filed January 25, 2024. Sections: business and competition, “Key Business Metrics,” subscription revenue, gross margin, R&D, and revenue recognition.
S3. ServiceNow, “ServiceNow powers enterprise-wide productivity and innovation with new generative AI capabilities”, May 7, 2024. Company-reported claims and forward-looking availability statements are attributed as such.
S4. ServiceNow, Knowledge 2024 announcement, May 7, 2024. Partner announcements are not evidence of commercial terms or realized results.
S5. Erik Brynjolfsson, Danielle Li, and Lindsey R. Raymond, “Generative AI at Work,” NBER Working Paper 31161, April 2023 full-text draft. The case uses the study as external evidence on a separate task setting, not as a ServiceNow performance estimate.