Novartis cut annual market research spend by $30 million — a 56% drop — and saved employees an average of two hours per major research query after integrating Market Logic Software's DeepSights generative-AI tool into its Sherlock insights platform in July 2024. The collaboration, recognized as a finalist in the 2024 Market Research Society Awards for "Best Data Solution," centralized knowledge management for Novartis's 7,500 internal users and replaced fragmented SharePoint sites with a single RAG-powered search and summarization hub.
"Sherlock has been critical in driving forward our commercial strategy across the entire international organization. Whilst our key use case has always been to support better, deeper and broader insights through centralized market research, our scope now extends into all areas of our internal strategy," said Ian Hook, Director of Business Operations at Novartis.
Novartis built Sherlock starting in 2020 to consolidate disparate SharePoint sites into a single sign-on insights repository. The first phase introduced five operational pillars: a single source of truth, robust governance frameworks, curated "Knowledge Zones" microsites for brands and therapeutic areas, seamless supplier integration, and aggressive user engagement — adoption grew 640% from launch, reaching 7,500 unique users and spawning more than 25 Knowledge Zones within a year. The platform, however, still required analysts to manually synthesize findings across thousands of research artifacts.
In July 2024, Novartis layered DeepSights, Market Logic's award-winning generative-AI tool for consumer insights, on top of the existing Sherlock data foundation. DeepSights uses retrieval-augmented generation (RAG) to pull from the underlying Sherlock corpus, runs a self-validation loop before displaying answers, supplies full citations with document references, and surfaces a "Watch Outs" flag when source material contains inconsistencies or contextual limitations. Novartis worked with legal and risk teams to embed compliance controls and modeled prompt-injection safeguards before procurement approval.
Operational impact scaled quickly. In one comparative test, Team A commissioned a three-month, $50,000–$100,000 patient-preference survey for a specific product's brand and formulation decision and drew conclusions from 50 respondents. Team B asked DeepSights the same question against the Sherlock archive and surfaced findings from 6,700 prior patient interactions in three weeks at zero external cost. Novartis now mandates that contracted research vendors upload deliverables directly into Sherlock to keep the corpus fresh, and a behavioral-science study inside the company found leadership endorsement drove spikes in platform logins — the "authority principle" outperformed highlighting historical pain points by a wide margin.
Looking ahead, the Novartis insights team is pairing DeepSights with Market Logic's broader portfolio (Explore, Radar, Personas, Innovate, Research) to extend AI-mediated insight workflows across competitive intelligence, trend radar monitoring, and synthetic-persona simulations, building on a knowledge-management foundation the company describes as the prerequisite for any AI deployment to deliver business value.