FactSet: Who Owns the Financial AI Workflow?

Topic

Corporate Strategy, Capital Allocation, Risk Management

Industry

Financial Exchanges & Data

Internal ID

S&F-STR-250

Source

Strategy & Finance

Publication date

5 October 2026

Pages

6

License

Not specified

Download case PDF (6 pages)Request teacher’s pack

Student strategy case · Information cutoff: August 3, 2026, 11:59 p.m. Eastern Time

All company information in this case was public by the cutoff. The executive assignment and decision described below are constructed for classroom discussion; FactSet has not disclosed a meeting or pending resource-allocation decision matching this case.

A new partnership, an open question

On June 30, 2026, FactSet Research Systems announced a strategic partnership with Google Cloud. The announcement described a plan to bring Google's enterprise Search and Gemini model capabilities into FactSet Workstation through Gemini Enterprise Agent Platform; add FactSet's Model Context Protocol (MCP) and agent-sharing functionality to Gemini Enterprise; and jointly develop agents for portfolio operations, deal advisory, and corporate finance. Google grounding would supplement FactSet financial data. Google Cloud would also join FactSet's existing group of cloud providers. The announcement did not disclose launch dates for each planned product, contract economics, exclusivity, data-retention terms, or client results. (FactSet–Google Cloud announcement, June 30, 2026)

For this exercise, assume you advise FactSet's executive strategy team on how to sequence its next 12 months of AI product and commercial investment. Your assignment is to recommend which capabilities to prioritize, what FactSet should own, where it should rely on partners, and what evidence should govern broader rollout. You may recommend a portfolio across product lines. This is not a claim that FactSet has a fixed AI budget, that a specific product has failed, or that management has reached an internal disagreement.

The decision arrives as FactSet reports broad client use of AI products and continues to invest in data, workflow integration, and AI. Yet the public record does not show how much AI features contribute to client retention or subscription growth, how much value clients will pay for, how reliably AI outputs improve professional judgment, or how the economics and data rights of the Google relationship are structured. FactSet must decide what to learn and build next while protecting the trust clients place in its financial information.

The business behind the interface

FactSet serves investment managers, bankers, wealth managers, asset owners, hedge funds, corporations, and private-capital professionals. Its subscription platform combines workstations, portfolio analytics, enterprise data solutions, feeds, APIs, and managed services across research, portfolio construction, execution, performance measurement, risk, and reporting. In its 2025 Form 10-K, FactSet said it aggregates content from third-party data suppliers, news sources, exchanges, brokers, and contributors into managed databases. It also produces, curates, and normalizes data and analytics of its own. Some data sets have limited suppliers. Thus, “FactSet data” can refer to a mix of owned or curated content and content licensed or aggregated from other parties; rights to use a source in a product do not automatically establish rights to expose it to every model, client, or agent. (FactSet FY2025 Form 10-K, Business and risk factors)

FactSet had already introduced AI features before the Google announcement. In December 2023, it launched Mercury in beta as a conversational, large language model-based knowledge agent aimed initially at junior bankers. Its release described access to FactSet fundamentals, pricing and bank/branch regulatory data, with a Chart Creator feature. In 2025, FactSet launched Pitch Creator, a workflow for investment-bank pitchbooks with search, charting, source-linked insights, templates, and Microsoft Office integration. FactSet has also developed APIs and MCP capabilities that expose data or workflow functions in other systems. These are existing product initiatives; the 2026 Google partnership's future agent offerings are announcements of planned collaboration. (Mercury announcement, December 14, 2023; Pitch Creator announcement, January 15, 2025)

The partnership therefore divides work across several layers. Google brings Gemini models, enterprise Search and grounding, the agent platform, and cloud infrastructure. FactSet brings its financial content, analytics, interfaces, client workflows, and financial-domain integration. FactSet's contribution includes content that it may license from third parties, not only information it owns. The announcements do not settle which party controls each client interaction, how data and outputs may be retained or reused, how the partners share economics, what happens if model or cloud terms change, or how clients can move an AI workflow to another provider.

An established business with a new adoption signal

For the third quarter of fiscal 2026, ended May 31, FactSet reported revenue of $622.9 million, up 6.4% year over year, and organic Annual Subscription Value (ASV) of $2.486 billion, up 7.1%. FactSet defines ASV as forward 12-month revenue from subscription services currently supplied to clients; its organic measure adjusts for acquisitions, dispositions, discontinued lines, and foreign currency movements. The company reported annual ASV retention above 95%. These results describe the existing subscription business; they do not identify AI's effect on revenue or renewal behavior. (Q3 FY2026 earnings release, July 1, 2026; Q3 Form 10-Q, MD&A)

FactSet also reported that more than 90% of its 50 largest clients used four or more FactSet AI products as of quarter end. The release does not define active use, minutes saved, output quality, willingness to pay, or AI-attributable ASV for that statistic. The company reported adjusted operating margin of 34.0%, down from 36.8% a year earlier. It attributed the decline mainly to higher compensation and technology-related expenses. The 10-Q says technology expense included cloud hosting, licensed software, and AI-token consumption, but it does not isolate AI's incremental cost. Adjusted margin is a non-GAAP measure. These figures inform the resource question, but do not by themselves establish that AI caused either growth or cost pressure.

FactSet's stated strategy joins three priorities: expanding proprietary and third-party data offerings and delivery, embedding more deeply in client workflows, and innovating with AI. Its Q3 filing says it believes enterprise AI depends on trusted, high-quality data, secure integration with models and workflows, and domain expertise. Whether this combination leads to durable value capture is the question students must assess. (Q3 Form 10-Q, “Business Strategy” and “Third Quarter in Review”)

Three possible investment paths

The strategy team could prioritize one path, combine them, or propose another sequence.

1. Bring more AI into Workstation. Accelerate Google-enabled agents within FactSet's interface for research, portfolio operations, deal advisory, or corporate finance. This could keep AI close to financial data and existing client workflows. It would also make the customer experience more dependent on partner capabilities and raises questions about usage economics, data handling, model performance, and who owns the client relationship.
2. Build a model-flexible intelligence layer. Prioritize FactSet-controlled data lineage, permissions, retrieval, evaluation, and workflow orchestration, exposed through APIs/MCP and selected client or partner surfaces. This could preserve optionality as models and platforms change and extend FactSet into client systems. It would require integration and support across multiple interfaces and data-rights regimes; modularity could increase reach while also multiplying coordination and testing demands.
3. Reinforce trusted content and verification first. Direct more effort to data coverage and quality, licensed-content governance, provenance, review tools, and existing high-value workflows before broadening agentic automation. This could protect reliability and trust in consequential work. It could also defer product learning and adoption while client expectations and competing AI offerings change.

These paths are not mutually exclusive. A decision could sequence them by workflow or persona, set different release gates for low- and high-consequence use, or run limited pilots before committing to wider availability. No public evidence in this case ranks one path as superior.

What does AI add—and what might it make harder?

Research on workplace AI offers mixed lessons. A field study of customer-support agents found that access to a conversational assistant increased average issues resolved per hour, with the largest gains among newer or lower-skilled agents. The study measured a comparatively structured support task at one firm; its productivity estimate is not a forecast for FactSet clients. (Brynjolfsson, Li, and Raymond, “Generative AI at Work,” NBER Working Paper 31161)

A November 2025 working paper studied analyst reports around FactSet's 2023 Mercury launch. Its first version reports broader and richer information in FactSet-referenced reports, as well as faster reporting, but also a rise in forecast errors, especially among analysts facing heavier information-processing demands. The paper is an unrefereed preprint. Its treatment measure is report citation to FactSet after Mercury's launch, not verified use of Mercury by an individual analyst. It studies sell-side reports and forecasts, not FactSet revenue, user productivity, or investment returns. This case uses the paper's initial version only. Students should treat the findings as provisional evidence about a possible tradeoff, not as a settled verdict on FactSet's products. (Xue, Zhang, and Zhu, “Generative AI for Analysts,” arXiv v1, submitted December 12, 2025)

Strategy research on ecosystems also highlights partner alignment. A useful AI workflow requires more than a capable model: it may depend on data permissions and provenance, current content, secure integration, a clear interface, client adoption, and human review. Platform research suggests that interfaces and modular design can affect how costly it is to work across multiple systems, but its empirical setting is mobile-app ecosystems—not enterprise financial software. It offers a lens for evaluating interoperability and coordination cost, not a prescription for FactSet. (Adner, “Ecosystem as Structure,” 2017; Chen et al., “Platform Governance Design in Platform Ecosystems,” 2022)

FactSet itself lists risks associated with AI development and use, including cost, uncertain monetization, inaccurate outputs, confidentiality and security exposure, intellectual-property and data-rights issues, and the possibility that free or lower-cost information sources reduce demand. These are disclosed risks, not statements that any particular event has occurred. They make the design of product evaluation and release decisions central to the investment choice. (FY2025 Form 10-K, Item 1A)

Your assignment

Prepare a short recommendation to FactSet's hypothetical executive strategy team. Choose an investment sequence across the three paths, or propose a defensible alternative. Your recommendation should:

• identify where FactSet should own capabilities and where it should partner;
• compare the customer, strategic, operational, and trust consequences of the alternatives;
• specify evidence gates before expanding an AI workflow, including measures of quality and human review as well as speed and adoption;
• identify the most important data or contract terms to request from Google and from internal product, sales, legal, risk, and client teams; and
• name evidence that would cause you to change your recommendation.

You may propose hypothetical targets, release thresholds, resourcing, or pilot designs, but label them as your own assumptions. Do not infer AI-attributable revenue from ASV growth or product-use counts. Do not treat the Google partnership announcement as proof of a released product or settled commercial terms. The case's historical information stops on August 3, 2026; later outcomes are not part of the assignment.

Exhibit 1. Product and partner map at the cutoff

Sources: FactSet 2025 Form 10-K; June 30, 2026 Google Cloud partnership; December 2023 Mercury and January 2025 Pitch Creator releases. “FactSet data” should not be read as meaning that FactSet owns every data item used in its products.

Exhibit 2. Q3 FY2026 indicators

Source: FactSet Q3 FY2026 earnings release and Form 10-Q. The figures are historical indicators for strategic discussion, not a financial model.

Exhibit 3. Public timeline

The cited source versions and locators are listed in the research source register. The independent research review accepted the dossier for case drafting with open limitations on partner terms, AI-attributable economics, content rights, and provisional academic evidence.

Back to strategy case studies