Databricks Genie for Marketing

Summary

• Natural Language Analytics: Genie enables marketers to ask questions in natural language without SQL, accessing real-time insights from unified marketing data.
• Marketing Questions Evolved: Unlike dashboards that answer predefined questions, Genie handles evolving marketing questions as campaigns run and optimize in real time.
• AI BI Capabilities: Genie is part of Databricks' AI BI platform, delivering accessible data intelligence for everyone on the Databricks Data and AI platform.

Databricks Genie for Marketing

Talk with your marketing data - Get real-time insights simply by asking questions in natural language.
Databricks Genie empowers non-technical users such as marketers, to derive insights and make data-driven decisions without needing to write SQL to access real-time answers.

Chapters

• 0:00 Marketing data challenges
• 2:01 Introducing AI/BI Genie
• 3:34 Dashboard and Genie overview
• 5:25 Data performance analysis
• 8:05 Root cause and optimization
• 12:16 Summary and conclusion

FAQs

What can non-technical marketers do with Genie?

Genie allows marketing professionals to ask natural language questions on unified customer data without writing SQL, getting instant answers to questions like which channels drive the most revenue or which customers are at risk of churn.

How is Genie different from dashboards?

Dashboards answer predefined questions and require technical setup, while Genie handles evolving questions in natural language. This allows marketers to investigate and react to issues quickly as campaigns run and strategies change.

What has to be in place before non-technical users can successfully ask data questions in natural language?

This video shows how to enable Databricks Genie for marketing on the Databricks Data and AI platform. Before non-technical users can effectively ask questions, you need well-organized data in your lakehouse, proper access controls through Unity Catalog, and semantic understanding through Genie's data indexing. The video emphasizes data governance, documentation, and testing to ensure Genie returns accurate, relevant insights.

How can non-technical teams ask questions about company data with AI?

This video demonstrates how marketing teams use Databricks Genie to ask natural language questions about campaign performance, customer behavior, and market trends without SQL expertise. Genie interprets questions contextually, queries your unified data, and returns answers in business-friendly formats. The video showcases workflows where marketers interact directly with data through conversational interfaces.

How can marketers create complex customer segments using natural language?

This video shows how Genie enables marketers to define complex customer segments through natural language queries on the Databricks Data and AI platform. Rather than waiting for data teams to write SQL, marketers can describe segments like 'high-value customers in the US who purchased in the last 30 days' and Genie translates this into queries. These segments become actionable for targeting, analysis, and personalization workflows.

Full transcript

0:00 Marketing data challenges

Hi everyone. I'm an industry and product marketing lead at Databricks. For this demo, I'm going to show Genie from the perspective of a non-technical marketer like myself. Someone who wants to understand marketing performance and make decisions quickly, but doesn't live in SQL or data tools. This could be a marketing practitioner or an executive who needs immediate insights to steer their business. Because ultimately, the people making decisions about customer experience and campaign budgets are often business users and they need access to data to make those decisions.
Marketing today is incredibly data-driven. Every campaign generates massive amounts of data across channels. Things like impressions, clicks, conversions, revenue, and customer behavior. As marketers, we make decisions based on customer behavior. So, we're constantly trying to answer questions like which channels or customer segments are driving the most revenue? Which high-value customers are at risk of churn? What behaviors predict our most loyal customers?
And where should we shift focus or budget next? The challenge is that even when companies invest in unifying their data into a customer 360, it's still difficult to answer those questions quickly. And there are a few reasons for that. First, most marketing users don't write SQL or build complex queries. So, we often rely heavily on analytics and data teams to provide the insights we need for campaign planning and quarterly reporting.
Second, dashboards answer predefined questions. They're great for monitoring metrics, but marketing questions evolve constantly as campaigns run and teams experiment and optimize. And third, customer and campaign data is often a fragmented across martech tools, regions, channels, and agencies, which makes it difficult to investigate and react to issues quickly. As a result, even when data is unified in a platform like Databricks, answering simple and essential questions can sometimes take hours or days and multiple cycles with marketing ops and analytics teams.
That's what AI BI was designed for, to give intelligent analytics to everyone. The real value of Customer 360 isn't just unifying data, it's making that data accessible to the people who need to act on it. AI BI is native to the Databricks platform and includes two complimentary capabilities. The first is dashboards to give your entire organization fast and secure visual insights. And number two is Genie, which I'm going to demo today.

2:01 Introducing AI/BI Genie

A natural language chat interface to converse with your data. And the best part is, you can start using it today with no extra license fees or seat based pricing. Well, Genie is the conversational interface to your Customer 360 data. It enables you to go beyond the what to understand the why. Instead of relying only on dashboards, you can ask questions in natural language and get governed secure answers from AI. Because Genie sits directly on the Databricks Lakehouse, those answers are governed, accurate, and the best part, built on the same trusted data foundation that powers the rest of your business.
And even better, Genie is accessible anywhere. It's built directly into every AI BI dashboard, so you can ask follow-up questions without leaving your reports. And for even broader reach, Genie is accessible in Slack, Microsoft Teams, or any app your your team is already use via our conversation APIs. So, let me show you what Genie looks like in practice. So, I'm starting here in Databricks One, which is designed to be the entry point for our business users like marketers and executives.

3:34 Dashboard and Genie overview

Instead of needing to understand all of the data available in my Databricks workspace, I can ask any question and get routed to the right AI BI a app, or Genie space. I can also find curated spaces designed for me like this marketing campaign performance dashboard. So, this dashboard is a typical starting point for CMOs or marketers who want a quick pulse on campaign performance across channels. Things like spend, revenue, return on ad spend, and conversions.
Dashboards like this are great for monitoring key metrics and performance and keeping everyone aligned. But, in reality, dashboards can't answer every marketing question and you often need deeper exploration. That's where Genie comes in. So, I'm going to open my Genie space right here. This is the digital campaign Genie that's associated with the dashboard I just showed. So, Genie enables marketers to talk to their data and explore data conversationally.
Instead of opening multiple dashboards, writing SQL, or submitting a request to my marketing ops team, I can just start asking questions. And actually, the first question I want to ask Genie is to tell me about my data. Excuse the typos. I might be opening a new dashboard or maybe someone sent me this link and I want to understand more about what data this Genie has access to. So, I can see that we have three tables um related to marketing campaign performance, um details about campaigns, daily metrics like spend, revenue, ROAS, etc. and aggregated weekly metrics.
And it covers multiple channels and sub-channels and regions. Great. So, I want to understand what's working well. So, which channels are performing best? So, Genie needs help, which is good. In marketing, performance could mean a lot of things. It could mean conversions, revenue, ROAS, click-through rate, etc. So, I love that Genie understands that this question is a little bit ambiguous and it's smart enough to know when to ask for clarification.

5:25 Data performance analysis

But, I want ROAS. Great. So, Genie's analysis is that digital and social channels are performing the best and print has the lowest average ROAS among our channels. I can also expand this analysis and see the exact table that Genie analyzed to give me this result. So, already we're getting a quick answer to a question that normally might require pulling a report or asking an analyst. Genie gave me a quick result table as well as a nice visualization.
And I can click edit to actually change the color of the charts or even download this as an image and add it to a slide deck. But, I want to dive in even more. So, I'm going to ask, "How have digital and social sub-channels been performing over time? Show by month." Marketers rarely stop at analyzing just a single metric. So, the next thing I want to understand is how performance is changing over time. Within digital marketing, especially, there's a lot of complexity.
Different platforms, audiences, creatives, and regions. For example, the sub-channels that we have in our data set within the digital umbrella include multiple platforms. You can see them here. Um Google, Meta, Instagram, LinkedIn, TikTok, YouTube, etc. Um as marketers, we're constantly trying to figure out which of those channels are improving and which are declining. So, here we can actually start to see some trends.
YouTube and Google Search are trending strong, but at the bottom Google Display and Programmatic have a significant decline in return on ad spend or ROAS. That's the kind of signal that would immediately get a marketing team's attention. So, now the question is, what should we actually do about this? So, this is where Genie becomes even more powerful. Instead of just answering questions, we can switch into agent mode where Genie can perform deeper analysis and recommend actions.

8:05 Root cause and optimization

So, a question I want to ask is >> channel spends and why? And this takes a minute to run. Uh it's kind of like deep research mode. So, I preloaded the question and I'm just going to switch to that. So, one thing that I really like about Genie, uh that's helpful for me, is that Genie will show the reasoning behind the analysis and the steps that it took to evaluate the data. So, it looked at overall channel performance metrics, subchannel performance breakdowns, and ROAS trends over time.
So, campaigns. Programmatic performance has collapsed. It's crashed from 4 to 1. 6, a 60% decline. Genie is recommending that we reduce spends in Programmatic and reallocate budget towards channels that are performing well, such as Google Search and YouTube. >> performing well. Um but given that this is performing so poorly, um I really wanted to dig deeper into the why of Programmatic collapsing. So, I asked, "Can you elaborate on why programmatic is doing so poorly?"
Genie explains that the root cause is audience saturation. In other words, people are seeing the same ads over and over again. So, ad fatigue, as Genie calls out here. So, programmatic advertising distributes ads across a huge network of websites and apps via automated auctions. So, think of seeing ads like on a news site, a recipe blog, mobile app, or even a streaming TV auction streaming TV platform. Google Display is a good example of this, where ads can appear across thousands of publisher sites across the web.
Because those networks rely on shared audience pools, it's easy for the same users to be exposed to the same campaign repeatedly. And it's actually quite difficult to limit that exposure because programmatic ads are served across many different systems and exchanges. Each one might enforce its own frequency cap independently. So, always a single global limit that marketers can set on how many times a viewer sees that ad.
So, when that happens, engagement starts to drop. You can see here, people stop clicking because they've already seen the ad too many times. >> rates and ROAS are declining. Essentially, we're paying for a lot of ad impressions that users aren't converting on. >> >> difficult to detect quickly when you're looking across dozens of dashboards and campaign reports. In many organizations, advertising is managed by a multiple agencies, each running campaigns across different platforms and reporting back on their own cadence.
That means that marketing teams are often stitching together fragmented reports, sometimes days or weeks after the campaigns are already running. So, by the time a problem like audience saturation becomes obvious, a lot of budget may already be spent inefficiently. So, that's the power of Genie on top of a unified customer 360 within Databricks. With Genie, instead of waiting for reports or digging through multiple dashboards, marketers can explore data directly and realize and respond to issues like this much earlier.
In the demo, I showed the power of Genie. In just a few conversational questions, I've gone from a high-level view of campaign performance to identifying a specific anomaly, understanding the root cause, and even getting recommendations on how to optimize our channel spend. In a traditional workflow, this level of analysis might require opening multiple dashboards, submitting several requests to an analytics team, or going back and forth for hours, sometimes even days, on Slack or Teams.

12:16 Summary and conclusion

With AI BI Genie, marketers and business leaders can explore their data directly, ask follow-up questions instantly, and move from insight to action much faster. The real value of customer 360 isn't just storing it in one place. It's making it accessible to the people who need to act on it. Thank you.

Learn more about the Databricks Data and AI platform.