How Databricks’ marketing team used a Genie analytics assistant to triple data use

The useful part

Genie, an AI analytics assistant | Databricks Blog Skip to main content Summary Databricks built Marge, an AI analytics assistant powered by Genie Agents, on a governed Marketing Lakehouse. It gives marketers trusted answers in seconds through natural-language questions, reducing repetitive analytics requests. Databricks marketers now use data 3x more often in decisions, adoption exceeds 85% of the marketing organization, and the team scaled access to insights without adding analytics headcount.

How it works

• Scaling self-service analytics works best when starting with one high-value use case, defining business context, then validating answers.
• Embed the assistant into existing workflows and continuously improve it through user feedback.
• At Databricks, we addressed this challenge by unifying our marketing data in a governed lakehouse and building Marge, our marketing implementation of Genie Agents.
• We built on governed data, taught Genie our business language, earned trust one use case at a time, and embedded it directly into the way marketers already work.
• I'm Liz Dobbs, Databricks' AVP of Marketing Technology, and in this video I share a practical playbook on how my team helped the Databricks marketing department use data 3X more often in decisions:

What to take from it

Which fields represent cost, engagement or attribution The approved relationships between campaign, account and sales data Clear metadata helps Genie interpret a question correctly before it generates a query. A meaningful decrease in benchmark accuracy signals that the data model, definitions or agent context may need attention. We included only the essential campaign, recipient and engagement data required to answer those questions.

Example or evidence

• Marketers access Marge through Genie One, the AI cowork experience for business users that brings together dashboards, Genie Agents, apps and deeper analysis.
• Users can see when an answer is based on verified logic, which adds an important signal of trust.
• For example, users may say "spend" or "investment" when the underlying field is named "cost." Marge needs to understand that those words refer to the same concept.
• Create continuous feedback and evaluation loops Every response gives users an opportunity to provide positive or negative feedback.

Details worth keeping

Most marketing teams aspire to be data-driven. In practice, getting a trusted answer, at the moment a decision needs to be made, can still take days. Marge lets marketers ask questions in natural language and receive governed answers in seconds.

Related coverage

• Databricks: Every marketing team sits on a goldmine of data: campaign performance metrics, customer...
• Databricks: We all know the pattern: you ask an AI tool a question and get an answer in seconds,...
• Databricks: At Data and AI Summit, we announced the evolution of Genie Spaces to Genie Agents,...
• Databricks: Ask a generic agent about revenue, and it'll likely grab the first revenue table...
• Medium: You run a business in HSR Layout and notice a familiar problem. Your website looks good, but enquiries are still slow. Your social media… Continue reading on Medium »