Lenovo puts data agents in hands of marketers across 170 countries
Lenovo deployed an AI data agent to give marketers and UX designers direct access to fragmented customer data spread across 35 countries and multiple internal systems. The move lets non-analysts explore campaign insights without waiting for data teams to aggregate information.
Derek Gominger, Lenovo's global e-commerce chief operating officer, said the company's core problem was simple: marketing data lived in silos. Campaign information from different regions, website traffic metrics, conversion rates, and unstructured content performance data existed separately, making it difficult for marketers to spot patterns or test new campaign ideas.
Lenovo tried traditional approaches like data lakes, but those tools couldn't effectively combine structured metrics with unstructured content data. An agentic AI solution-specifically Adobe Data Insights Agent-could process both types of information simultaneously and surface answers faster.
Who uses it and why
Three groups now use the agent: data analysts, regional and local marketers, and UX designers focused on customer experience. Marketers can ask questions about campaign performance without analyst assistance. UX teams use the same data to simplify product searches on Lenovo.com, where a dense catalog can overwhelm buyers.
Small businesses account for 70% to 80% of Lenovo's online traffic regardless of where they eventually purchase. Small efficiency gains in how customers find products compound across the business. Gominger said even a couple of basis points of improvement in media spend efficiency scale significantly to the bottom line.
The broader shift to conversational search
Gominger sees agentic AI as the future of how customers interact with e-commerce sites. Traditional search will give way to conversational agents that act like concierges, asking clarifying questions to narrow options rather than forcing customers to browse entire catalogs.
He compared the experience to a wine sommelier: instead of memorizing every bottle on the list, a customer answers four questions and gets a good-better-best recommendation. Applied to Lenovo's product range, an agent could ask about use case, budget, and performance needs, then surface relevant options.
Keeping multiple teams aligned
Gominger acknowledged the hardest part of scaling agentic AI: preventing teams from building disconnected solutions that pull in different directions. Without coordination, one group solves one workflow, another team tackles another, and results become inconsistent.
His approach centers on three elements: guardrails and guidelines to keep development within a defined framework, shared data standards across the organization, and formal governance that prioritizes specific use cases rather than letting every team experiment independently.
Lenovo also invested in building its own on-device agent, called Qira, rather than relying solely on third-party models. Owning the agent and the device lets Lenovo control data security and build internal competency around the technology.
Measuring success goes beyond KPIs. Strategic decisions like building Qira reflect where the company wants to own core capabilities, even if immediate ROI metrics don't fully capture the value of that investment.
Learn more about AI for Marketing and AI Data Analysis to understand how these tools apply to your work.
Your membership also unlocks:
