Databricks on Databricks: Powering Marketing Insights with Lakehouse
Summary
- Internal Case Study: This video shows how Databricks' own marketing team uses the lakehouse on the Databricks Data and AI platform to power marketing insights and analytics.
- Lakehouse Benefits: Learn how a unified lakehouse architecture simplifies data integration, analytics, and AI for marketing use cases compared to traditional approaches.
- Marketing Analytics: Discover practical applications of the Databricks Data and AI platform for customer analytics, campaign optimization, and market insights.
Databricks on Databricks: Powering Marketing Insights with Lakehouse

Talk by Anoop Muraleedharan, Sr Director Marketing Analytics, Databricks ; Elizabeth Dobbs, AVP, Marketing Technology, Data + Growth, Databricks
This presentation outlines the evolution of our marketing data strategy, focusing on how we’ve built a strong foundation using the Databricks Lakehouse. We will explore key advancements across data ingestion, strategy, and insights, highlighting the transition from legacy systems to a more scalable and intelligent infrastructure. Through real-world applications, we will showcase how unified Customer 360 insights drive personalization, predictive analytics enhance campaign effectiveness, and GenAI optimizes content creation and marketing execution. Looking ahead, we will demonstrate the next phase of our CDP, the shift toward an end-user-first analytics model powered by AIBI, Genie and Matik, and the growing importance of clean rooms for secure data collaboration. This is just the beginning, and we are poised to unlock even greater capabilities in the future.
Here’s more to explore:
Economist Impact: Unlocking enterprise AI: https://www.databricks.com/resources/webinar/unlocking-enterprise-ai
The State of Data + AI: https://www.databricks.com/resources/ebook/state-of-data-ai
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements
Chapters
0:00The 10x marketer challenge4:35Four pillars for data growth8:08Architecting the lakehouse11:53Medallion architecture layers16:18Marketing analyst workflow20:11Empowering with Lisa25:40Lisa performance demo30:42Lessons and future vision
FAQs
How does Databricks use the lakehouse for marketing?
This video explains how Databricks' marketing team leverages the lakehouse on the Databricks Data and AI platform to consolidate data and create unified customer views.
What marketing insights can you derive from a lakehouse?
According to this video, a unified lakehouse enables analysis of customer behavior, campaign performance, and market trends for strategic marketing decisions.
What is the advantage of unified architecture?
This video demonstrates how combining data lakes and warehouses eliminates data silos and enables seamless analytics and AI applications.
How has this approach transformed marketing at Databricks?
This video shares how adopting the lakehouse architecture improved marketing team productivity and decision-making speed at Databricks.
Full transcript
0:00 The 10x marketer challenge
All right. Hi everybody. Thank you for joining. Uh I'm very excited uh to represent our marketing analytics team. Uh I run I'm aping technology data and growth team here at data bicks and I'm joined by our senior director of marketing analytics and engineering. uh and we are actually have the opportunity to represent an amazing analytics team who have spent the last two and a half years on an very scrappy journey where we've had to really fix the plane while keeping it flying when it came to our marketing data foundation.
So, we're going to talk a little bit today about where we were, the journey we went on, and how we kind of built that north star, um, and have a couple product demos we actually see, uh, in real life about where we're going and how we're actually using data bricks today from a marketing perspective. So, today marketers are being asked to do a lot. They are being asked to predict what their customers are going to do next.
Optimize the campaigns while they're still running and deliver this Netflix-like personalization, which used to be the default at a BTOC company, but now your B2B buyers have those same expectations. And by the way, the expectation for your marketers is to do this all at scale and real time. And but that this is the new standard. But honestly, most data teams and most marketing teams are really not set up for success in this new paradigm because as the external expectations are getting higher and higher and higher, the headwinds that they're facing on the external market around uh cost and process is getting higher as well.
Most people in this room may be familiar with the idea of the 10x engineer, the one who writes better code, ships faster, and really raises the bar for all the engineers on the team. But here's the shift. The 10x engineer is now not the only unicorn in your organization. The idea of the 10x marketer is becoming more and more real every month that passes. This 10x marketer now should be making 10 times better decisions by trusting real time insights that they're getting.
They should be moving faster. They should be using their own natural language to be able to interact and engage with the data that you already have. And they need to be able to scale. They need to be able to do more using a agents to do what they're doing, but do it bigger and better and faster every time. So what does this mean for your data team? Your data teams are already managing a lot of complexity if you're servicing a marketing organization.
They're actually powering that 10x marketer. But behind the scenes, things are getting relatively overwhelming. They're already handling batch data from legacy systems, streaming data from your ad platforms and product, your structured CME, CRM records, your unstructured data from reviews, chats, and none of this arrives in a clean, tidy manner that's easy for your data team to make sense of. Instead, they need to do all the work to take all of these things and make it make sense for the marketer.
But marketing can't wait. They need to act in real time on every signal. So this is chaos. And if this is a video of actually a noop and his team managing this about two and a half years ago, and I know I'm preaching to the choir because you know the theory, you know the complexity, you know when you're getting that slack from a marketer that's saying, "Hey, this dashboard doesn't feel right or hey, I just bought a new technology.
Can you just integrate it? No big deal." Um, but about two and a half years ago when Data Brooks was really scaling, our CMO challenged us to, by the way, we need to do more with less. But we knew what we were doing was not actually taking advantage of the secret weapon we had, which was our first party data. So while our analytics team was doing kind of maybe lower value work like juggling dashboards, patching our pipelines and firefighting all the bugs that would jump up, we kind of needed to service the marketing team but also fix and design for the vision in the future.
So we needed to get past this bubble gum and ducttake architecture and figure out how we were going to write the ship and set our marketing team up for success in the next three, five, 10 years. And so we really challenged ourselves to think about how can us as a data team power the growth of our marketing organization. And it came down to these four pillars which we kind of co-developed with our marketing leaders on the business side and said really where can marketing data act as a secret weapon.
4:35 Four pillars for data growth
So we have four areas but first we want power better personalization. A lot of traditional CDPs are really good at customer data and a lot of other platforms are really good at campaign data but we realized we didn't have a great way to have a unified view of both our campaigns and our customers at the same time. And those two things are actually the secret weapon when it comes to personalization. And fundamentally, when we talked to the business, it wasn't that our marketers were struggling for ideas.
They're struggling because the data doesn't show what they should say, when they should say it, and how they should say it. We're going to talk a little bit later about how we took this concept and really made this a core tenant in our architecture design. But at the end of the day, it's about moving beyond this kind of subject line AB testing. This point in time kind of personalization is something we could already do on our fragmented kind of hairy architecture.
But the idea of this ongoing personalization, building journeys, continuing to personalize on someone's path with us was the northstar we realized we needed to build towards. Next, we knew we needed to build smarter campaigns because we knew if we did this right and had a really clear architecture that every machine learning model we'd build, every AI insight we'd find should be better than anything we could buy on the market.
The best campaigns didn't really start with that creative. It started with the confidence that hey, we feel like we have a lot of data and it tells us what we need to do. We used to like use our own product and third-party tools to build smarter campaigns. And where we decided to go is that we needed to be take those thirdparty tools, remove them from our data set because we felt really confident that we needed to be able to build smarter campaigns ourselves.
Next, this is a really big tenant we're going to be talking a little bit about later, which is how do we empower our marketers? I think in this world where data is front and center, not all of your marketers have that same level of data literacy. So if you drop them into a dashboard or a new report, sometimes they're not able to get the insights as quickly as others or maybe are really struggling to see the forest of the trees when they get that weekly email that says here here's how your campaign is performing.
So our goal was really not to give just marketers access to the data. Every marketer has access to more data than they know what to do with today. And finally, this vision of agentic and AIdriven execution. We really are big believers that AI should be helping build that 10x marketer and supercharging the team that you already have. So, in order to build that confidence and the trust and really help our marketers get inspired and excited about what you're able to do, we tried to focus our kind of ex experiments here on taking and reducing that manual work your marketers were already doing.
So think about product tagging, segmentation, all of the things that take a lot of time for someone to digest, understand, and let's be honest, maybe aren't always done in the same way to make your taxonomy super powerful. But reducing that human effort is seen as a huge win from your marketers. They appreciate the time back and it's a great use case that isn't too too big to bite off as you get going on this journey.
So I'm going to hand it over to who's going to walk us through how we built our first marketing lakehouse to answer all these problems. Thank you. All right. So let's take a look at uh the problems that we had when we started this journey. So we had three fundamental data challenges that were holding us back. Um number one, our data was totally fragmented. I think this is very familiar to all of you. So all these systems generate tons of data but they all are different granularity.
8:08 Architecting the lakehouse
They speak this different language. The taxonomy is different. So because of that we will never get a complete picture of your buyers journey. The second challenge was uh data duplication. So it's a continuation of the first problem because um you know these analysts get a lot of requests for business insights. Now what they do is they go and pull data from all these individual systems into the notebooks and start providing insights.
So what happens is they end up creating dozens of notebooks. Um data is getting duplicated. So we know that when we duplicate it's going to impact your uh processing and storage cost but more importantly you are actually creating multiple sources of truth in that process. That's the main problem there. Lastly again a continuation of the second problem you have to deliver these insights in some form to the business. So the only way or most of the time you end up creating reports and dashboards for each of those insights that you deliver to the business.
So what happens you'll end up having hundreds of dashboards. At one point we had 300 plus dashboards. Um it is a nightmare to operate operate on on on that environment. So all these resulted in um underperforming capability. So what happened was um uh because of these foundational issues we were stuck. we could not confidently move towards advanced capabilities like uh personalization, multi multi-touch attribution, lead scoring um and so on.
So our uh infrastructure simply was not ready for any of these advanced use cases. All right. So what do we do next? So before we do uh decide on what has to be the data architecture or infrastructure, we have to start from the first principle. So we have to look at and see what are the goals are we trying to solve with this architecture. So number one um we needed a system that can be the single source of truth for customer and campaign data.
So what that means is so we can eliminate duplication unified data from um you know different systems not just the engagements but engagement spend outcome. So we wanted the system to track performance holistically and measure ROA with confidence. So that's the number one goal. Number two we wanted an improved decision- making framework. So what that means is um marketers were spending too much time hunting for data and not enough time acting on it.
So we wanted a system that can take us from reactive analytics to proactive insights. And finally we needed a system that is uh both modular and compliant. So everyone who works in marketing knows that systems come and go, right? You you see that new systems are getting added. Um and you want an architecture that can actually um you know take data from any system that that you have to tap into but make sure that fits into your architecture, your schema and whatnot.
You don't want to you know engineer a lot of things when we see a new system and also you want to make sure that you meet strict security compliance requirements. So that's where u lake house comes into picture. So we have um the Medelinian architecture that can provide good uh data quality management, good governance uh and also easier access of data. All right. Um so people who have worked on Lakehouse and familiar with Medelan architecture know that it's a it's a three-step process.
Um the first is figuring out the injection. So raw data we have to tap into the source systems and put pull them into the uh uh you know the bronze layer. Um so we used a combination of third party tools and custom APIs to ingest the data and this data is basically the row storage. We don't do any processing at this level. The second is the trans transformation stage and this is where the magic happens. So we have to clean uh standardize and enrich all the data that we collect.
11:53 Medallion architecture layers
We have to unify all the identities that we collect from different systems. So that includes account identities, user identities and whatnot. And then also all the standard business rules are applied at this layer. And third and the final stage um is the activation which is how do we convert these transformed data sets into business ready products. So we do not want analysts to come in and start churning these you know uh new data sets again but we give business ready products for uh end users to operate on and also how can we create all the base attributes that we need the CDP to be operating on and lastly we needed an infrastructure where people can easily explore and access the data.
So let's see how does all this come together um in our lakehouse right so we followed a medallion architecture like I said um broad layer has all the row data and that is all preserved historically in our case we have the first party data coming from marketing automation all the event management systems web and mobile activities CRM product and telemetry data and we also have third party data which is actually pulled using tools like fra I talked about uh which includes all the paid media data that we pull from ad platforms content syndication and also the enrichment.
Second is the transform layer. So this is where the productization happens. So these are the actual products that we have created. In the silver layer we have created product for identity that is how do you connect how do you create a unique ID for all the systems that we bring together for analytics and that ID is very unique to our platform so no other systems would understand that ID um so it's it's it's it's comp it's not PII it's compliant to all the security governance aspects and we have data sets for web session so anything that you want to know about web data you have a product for that and so And finally our machine learning processes.
So we have uh lead scoring, account scoring, look like models. All those uh dat processes are running out of the silver layer. So we don't have to do a lot of feature engineering because it's already featurerich. We can use machine learning processes on top of that. And finally activation we have a composible CDP. So we use that composible CDP. We have already profiled our audiences accounts and we have created segmentation data sets as well that goes into our activation layer.
Plus we also feed all these performance reporting data sets into AIBI and Genie spaces. Um so instead of data leaving in uh in separate system everything flows into one unified environment uh where website visitor data get automatically enriched with CRM. your recent engagement immediately uh is used for scoring leads and send to sales. Uh digital media interactions become part of the complete customer profile and eventually what is more important is all these touch points together helped us to create a multi-touch attribution model and uh because we follow the median architecture and lakehouse uh everything is governed through un catalog and I'll show you an example of that.
Um so ultimately this composible architecture helps us to scale personalization, drive ROI and deliver insights at every step. So it's not just collecting data and moving around but we were we had specific use cases we used to build products that can be activated for uh various use cases. Now let's see one of these use cases in action. All right. So I'm I'm trying to demonstrate um the experience of an analyst working lakehouse today.
16:18 Marketing analyst workflow
So remember an analyst has to go and tap in the data from each of the individual systems in the beginning and had to build those analysis inside a notebook. But let's see how that happens today. Click one more time. There you go. So in this case analyst is responding to a business request and is looking for a consolidated marketing engagement data set. So what they do is they go to catalog and search for the data set.
When they search they find all the details I need to know about the data set. So they have the description, they have all the columns, description of all the columns, everything is outlined neatly in this uh platform. So now the analyst is interested in a specific column. They want to know if marketing channel is a column in this data set. So they search for marketing channel. They're able to find that. they're able to find the description of the of the column.
The next step is to see some sample data. With one click, you can actually move to the sample data field and make sure that some of the columns that you are interested in is available in the data. Now that is not enough also want to interact with this data. So we have the genie room. So with one click of button it opens a genie room and they can and Genie recommends some some questions based on uh you know previous activities.
So in this case analyst is interested in knowing uh details about the channel. So select the question which is more relevant. But then they realized oh you know this is not the granularity I need. I was looking for marketing channel but it's showing me channel group. All they have to do is edit the query change it to marketing channel. All right. So now I see the list of values I was interested in seeing marketing channels perspective data looks good.
Now I can go back to the catalog and let's look at the lineage. So I want to understand where exactly data is being sent to this this data set and where exactly uh the data is being sent out of this data set. So that kind of gives you the uh details in a in a lineage graph. So this tells you not just the tables but also all the processes workflows that are actually impacting the table and also data moving out of the table to downstream systems.
Okay. Now since I have all the data then last step I want to know is I want to visualize some of it. I want to see if the data is complete. So I can click on open dashboard and the assistant will give me some suggestion. This time I'm interested in knowing if the data uh is complete. Is there any um you know gaps in the data if I look at engagements over time looking at there's no gap so I think it is covering the entire time frame I'm looking for now since I have figured out the data now the analyst can go and do more analysis on this data set because everything is in one single environment don't have to go and open multiple notebooks and whatn not so they can go and start analyzing the data in detail So what you seen is actually a a seamless integrated experience of analyst working on the platform today.
Uh this approach allows analyst to spend more time generating insights for business rather than uh you know wrangling data pulling it from multiple sources. So this is how an analyst works in uh our marketing lakehouse today. With that I'll hand it over to to take it from here. Awesome. Thank you. Uh I don't know if this is working anymore, so I'll just stick over here. Uh great. Thank you, Anoop. And so I'm actually gonna walk everyone through.
20:11 Empowering with Lisa
You've kind of seen the back end, but as everyone in this room knows, probably too technical for most marketers to make sense of, nor would we want them to spend their time there because that's really the place for our analysts to hopefully start doing the more uh int uh intense insight analytics and really finding the patterns in the data, not just kind of looking for quick updates on how campaigns are performing. And so what we've solved for that is with.
So Lisa, we have a parent marketing genie. Her name is Marge. Uh, and is trained on all of our marketing data. So, we really have trained her on every engagement, every account insight we have. And she's really good at answering kind of those first level questions. So, how many DBUs is company consuming? Who's the AE on this account? What webinar was the best last quarter? All of these things Emerge is great at. But where a lot of these, you know, AI solutions tend to fall down is when you have to ask really deep questions.
Because when they're trained on a ton of data and you're trying to go deep, a lot of times they may hallucinate. They may not really realize exactly what the end user is trying to accomplish. So we've introduced, which is the idea of these kind of child genie spaces, which are really trained on a more narrow data set. So in this example I'm going to walk you through today, actually trained on just digital advertising data.
So she would be something that our digital advertising team would use and really able to go deep and answer questions in their terminology and their slang uh that they use in a daily basis. So building a genie is is actually pretty easy. So it seems really hard and complex, but we kind of really had a couple things that we focused on when we were doing the initial design. First is the idea of natural context. We really wanted to make sure that we were taking the way that marketers were talking about their business today and making sure that our models understood what that meant and that they knew the SQL to run to get that when they were saying certain things like MQL or return.
And next we we we wanted to make sure that we were training the model on the data we had and we weren't just replicating work again. And that's what an talked about back to the lakehouse the unity catalog everything's there everything's tagged and so what we're doing is we're just borrowing that architecture and then we're building an interface on top of it that makes it easier for a marketer to engage with. And finally the idea of decision support.
I think all of us are are familiar with that kind of blank page problem. It sometimes it's really hard to get started. And even when you have something like Genie, you don't really know what questions to start asking. So what we're trying to do now is in this year is actually design to complement the dashboards we've already built for our businesses. So imagine your digital marketing team has a dashboard that they live and breathe in and imagine Marge is right there with them.
So with they have a question, they can actually just ask Marge within that dashboard experience instead of having to go to a different system, open a ticket, kind of figure it out. We're trying to again bring the data to the marketer. And finally, I think a lot of us who are familiar with chat GPT and all these amazing innovations coming out, the idea of trust is still really intangible. And I think a lot of people are worried that if I get an answer from a system like this, how do I know it's right?
And to be honest, like I think we're the first to say, it's not always going to be right. It will be wrong. It needs to feed back. It will get better over time. But it's like the TSA. If you don't see something, see something, say something. And we wanted to make sure that we were designing our genies to have that same ability for a marker to feel like their feedback was being heard and was using to improve the model.
So, first we wanted to be very transparent by design. And within Marge or, if you're asking a question with no context, she will literally tell you, I don't know. Which seems kind of dumb, right? You're like, I don't want a product to say I don't know. But from a enduser experience, it's actually very comforting because you have more confidence that she's not going to give you an answer that she's unsure about. Secondly, we wanted to make sure we were utilizing this built-in feedback.
So very simply and I'll show you in the demo in a second. You'll see every engagement with Marge is asking for feedback at every point in the journey. So yes, is this good? This will tell us do more of this. No, this was bad. Oh, let's figure out why and we'll make it better for you. All of these kind of negative comments actually go back to our team and we actually capture all of that and we can take those and figure out why it didn't work and kind of constantly be iterating improving Marge and on the back end.
And finally, something that we're really excited about is the idea of certified answers. A lot of times what will happen is marketers will do some analysis. They'll figure out a trend they're very excited about, but then there's this little like, oh, maybe I should just double check before I do something about that. So, I think the goal we're solving for here is actually built in this idea of certification within the Genie experience.
So, when the marketer is asking a question and getting an answer, there's like a little green bar and says, "Hey, this is 100% right. You call our analyst team, you call a nuke, he's going to give you the same data. So go go ahead and run that campaign. So I'm going to walk you through a little bit of the experience of what Mar or actually looks like. So this is just a you know visualization of what the Genie hub looks like and you actually can see welcome to gives you a little bit of Q&A about it.
25:40 Lisa performance demo
But something a lot of tools miss is the idea of uh understanding what is this data really telling me? So by clicking that little box there, it will show you like the table names actually describes you what trained on, what values you can ask about and all of that I think is really helpful. So they kind of want to know what's available. So let's just ask a little bit about the data set and give me the cost breakdown.
So, as you can see here when she loads, um, we asked for the data over time with how much we're spending in each region. And as you can see here in our nice bar chart because our table's helpful, but the bar tells us a little bit more of a visualization. AMIA looks a little bit off. So, it seems like there's a big jump quarter over quarter. So, let's ask Marge uh to help us understand what's going on in that data in AMIA.
And I realize you can't see any of my queries because the words are over it. So I'm sorry. Um, so we actually asked how many lead or how many leads did we generate in this period? Because spend is fine. We just want to make sure we're getting the impact we expect from it. So now we want to click and understand was that spend really worth it? Because if we're again spending a lot of money, getting good leads, great. So let's figure out how much that cost per lead is.
All right. All right. So now we have our answer here and really what it is helpful is okay we know where we are but a data and a point is not that helpful. You need the context to understand is it up is it down. So let's actually ask Marge or how did this do in 2025 to give us a context here. Oh look okay. So you actually are spending way more money and a cost per lead and I can see what that little green box that 2025 value is certified.
So this has given me the data as an analytics team to say our cost per lead jump. We should talk to our digital team. Is it audience saturation, creative fatigue? Is it time to rethink our targeting? Uh the great thing about is it we can now get this answer and share it with our team to do something about it. Next, we're going to walk through a quick video here that talks about an engagement with that's a little bit higher level.
So, within, we really wanted to build a way for anybody like a CMO could come in and really engage with the data the same way an analyst could. So, we asking the question of which channel is performing best. But what actually doesn't know is what does best mean? So we actually she asks you give us some more data. Tell us what best means to you. And we're going to tell her that we want rorowaz because that's a very simple metric.
Everyone should know about it. And Lisa's going to tell us, hey, looks like programmatic Google is your best rorowaz. Congratulations. But again, we want that context. So we're going to tell, tell me all the marketing channels because I want to know if is it better by a lot, better by a little bit. And so we're going to ask for that data. And look, looks like gave me the channels, but she didn't carry that rorowaz value through.
No problem. I'm just going to tell that I need to fix it. And I expected to add the rorowaz. Please add that back in. And she's now rerunning. And she'll add that column back in for us. No problem. So now we can see all of our channels, all the rorowaz, and then you can still see that spike, and it matches up in our table below. Uh but again, it's a data point. It's not that context, that time series data that's really going to help validate or invalidate any hypothesises we have.
So we're going to ask her for this data for the past year by month. And so you can see here, great, we have this data, but by the way, we can see this is just over time. And we see this big dip in this kind of year. All right, we should probably dig into that. Is that all of our channels? Is that one channel? So, we're going to ask to show us this data, but actually give me the same view, but with a channel lens on top of it.
And so, now, you can see here has our channel, and you can see on the top bar there that again, it is in fact our consistently higher performer and a good use of our money because it does give us the highest return on our advertising spend. The great part about this is it, you know, you're it's not lost work. There is a great download button here. If I want to share this with any of my peers, you just click download. will take that actual data into a sharable format.
You can share the code. It will show you the code it pulled. So you can share that code with any of your co-analytics teams so they can use that. Um and anytime you know she's not working, it's easy to give her feedback and help her rerun. So looking ahead, what did we learn? I think this is the part of the track where I stopped pretending this was all smooth sailing and we had like a perfect map and a perfect northstar and everything went great.
30:42 Lessons and future vision
Uh in fact, you know, this was a journey and we're still on that journey. Things are not as perfect as we'd love them to be. And this is a constant test, learn, and iterate process that our team is going through on a daily and weekly basis. But what do we learn? First is standardization is not sexy. If you try and get a meeting with your CMO to talk about taxonomy or metadata, they will probably accept and then just not come.
Um but it is critical to the foundation of how your marketing team needs to run. you are probably targeted as much as we are on all these amazing tools on LinkedIn that say they're going to solve all of your marketing problems, but at the end of the day, they are going to be messy and ineffective if your data is messy and ineffective. So like tackling that kind of data foundation first is really critical to making the most out of the tech stack investment you are already spending.
Next is the business context and is is the compass you need to organize behind. If you're building all these capabilities that the business is not actually asking for, then your team's going to be frustrated and the business team is going to be frustrated. So really co-developing and helping the business lead the data work and the data exploration is really critical to adoption and making sure we're building something that the marketers will actually use at the end of the day.
Third is software tastes like chicken but change does not. A lot of the software as a nuke said is very modular. You could be using something one year and then they come back and ask for 50% more and your team goes never mind we'll swap for another tool. This happens you know year after year but change is not that easy. Change is very hard. So while you know it's easy to kind of swap in and out of these tools with our kind of lakehouse architecture, getting on top of change management and helping your marketer continue to understand the tools at their fingertips is really hard and will be an ongoing process that I think teams especially our data teams will invest more and more in as technology becomes more of a core of a marketer's day-to-day.
And finally, sponsorship dictates speed and success. A lot of these things actually require some air cover from your leadership team. It's not that, you know, again, we're going to take the plane down and say no marketing for six months while we rebuild our architecture. Everyone in this room knows that's not an option. You have to fix the plane while you keep it flying. But when that means that you keep the plane flying, it means that you know you cannot give that same level of support because you need to use those resources in that transformation you're driving.
So an executive sponsor who understands where we're going and in especially a resource constrained world like many of us are in today, their air cover is going to help your teams be much more comfortable and confident to move fast on a lot of these uh changes. And finally, where are we going? Um I think this is a really fun, you know, what we start with is probably not what we're going to end the year with on our priorities.
Uh but we're really excited about kind of the potential in the ad tech space. We think we haven't done enough here. Uh, and we really want to spend some time this year advancing our ad tech ecosystem from cross channel personalization, improved activation, fullfunnel measurement scale, using clean rooms and mediumix modeling to really kind of give us a new lever in how we're spending our digital marketing dollars. Next, we really want to do more with our agentic orchestration.
Uh, you know, AI today is a really an assistant to our marketers. We haven't done too too much yet on letting that the agent make the call and pull the levers. So, we're doing more this year on triggered journeys and campaign optimization to allow these agents to act on live context in real time. So, we're very excited about that. And finally, this theme that we mentioned is bringing the data to the marketer. We're really kind of trying to continue to reimagine how analytics can become more conversational because we believe it's be if it's more conversational, it'll be more used.
So with assistant first analytics, I think everyone will become fluent in our data without ever having to write SQL. So that's really how we're thinking about moving from insight to action to automation. So thank you. Uh really appreciate the time today. Anup and I will be up here if anyone has any questions and wants to chat. Uh but if you're interested in learning more about Marge and the what AIBI and these dashboards plus kind of the assistantled analytics are, we actually have another session from on our team who will tell you how Genie can really transform that day-to-day for your organization.
Uh and that will be on Wednesday. So thank you guys.
Learn more about the Databricks Data and AI platform.
The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.