How HP is building an AI-Ready, B2B GTM Data Infrastructure for Global Marketing

Summary

This video covers how HP's Global Martech team built a B2B customer data platform on Delta Lake and Unity Catalog, enabling first-party data unification across segmentation, activation, and measurement for global marketing use cases.
Luis Alonso explains how HP adopted a composable CDP architecture with Delta Sharing for partner data exchange, Genie Code for account clustering, and Genie Ontology for KPI definition, reducing audience mapping time to under an hour with CustomerLake.
The video outlines HP's roadmap for extending the marketing lakehouse with journey orchestration, cross-channel measurement, and CDP cost reduction, positioning the Databricks Data and AI platform as the foundation for AI-ready marketing infrastructure.

How HP is building an AI-Ready, B2B GTM Data Infrastructure for Global Marketing

As marketing teams look to deliver more tailored, measurable, and efficient customer experiences, a trusted Data Infrastructure is foundational to modern martech strategy. In this video, Luis Alonso, Head of Data Strategy & Engineering for Global Martech at HP, will share how HP is approaching customer data, identity, and analytics to support global marketing use cases across segmentation, activation, measurement, and AI. Attendees will learn what it takes to build a scalable customer data foundation that connects marketing ambition with governed, enterprise-ready data infrastructure.
Talk By: luis alonso, Customer Data Strategy & Engineering , HP Inc ;

Chapters

FAQs

What is a composable CDP and how is HP using it?

A composable customer data platform (CDP) assembles best-of-breed capabilities on top of a central data platform rather than purchasing a bundled, all-in-one CDP vendor product. This video explains that HP chose to build its B2B composable CDP on the Databricks Data and AI platform, using Delta Lake for unified storage and Unity Catalog for governance across global marketing data.

How is HP using Delta Sharing for marketing data exchange with partners?

HP uses Delta Sharing to provide marketing partners with direct, governed access to data without copying it into separate systems, reducing the friction and lead time previously associated with partner data delivery. This video shows how this approach sits within a broader marketing lakehouse architecture built on the Databricks Data and AI platform.

What is Databricks CustomerLake and what results did HP see in preview?

CustomerLake is a Databricks capability for customer data management that HP was previewing at the time of this video. The video reports that the CustomerLake preview enabled HP to complete audience mapping in under an hour, a significant acceleration over their previous process.

How is HP using Genie for B2B marketing analytics?

HP uses Genie Code for account clustering tasks and Genie Ontology to define and govern KPI definitions consistently across global marketing teams. This video positions these capabilities as part of HP's path toward AI-ready marketing infrastructure that non-technical marketing staff can query in natural language.

Full transcript

[00:09] Right.
[00:23] you registered for this topic. You know, one of the things when I come to the summit and similar events that I always try to walk away with is there one or two or three nuggets that I
[00:35] can take back to me with with my team. Perhaps it's a different lens to view a particular topic or a particular solution that we're looking to develop. I know it's information
[00:49] overload. There's a lot of sessions. Sometimes it feels like you're in Vegas for the weekend. There's so much going on. So, I'm hoping that you can walk away with with a few, you know, nuggets that based on our experience over the last
[01:03] five years is is what we're bringing to the table today to talk about. Uh let's see. To talk about
[01:16] Okay. Uh to talk about this. So, building an AI-ready B2B go-to-market data infrastructure. That is a mouthful. That's quite a bit there.
[01:28] It's also another way to say maybe it's fancier, maybe more complex to say what are we doing to build out CDP capabilities in our B2B world. And you know, CDPs, I think in some
[01:41] sense, it's a loaded terminology. I think the way technology is evolving today the way we define a CDP is changing considerably. Uh
[01:55] you know, just about 2 weeks ago, the same day I had one audience member in a different discussion we were having in a meeting mentioned, "Hey Luis, you know, we know
[02:07] we want to do a few things with orchestration and journeys, but we don't have have a CDP yet." And then similarly, in the same call a little bit later, somebody from another point of view in
[02:21] engineering in in data strategy came to me and said, "Hey Luis, I heard you're you've now brought in the new intent data to the CDP." And so again, I think depending on who
[02:33] you are, what lens you come from in the marketing organization, what a CDP means to you can vary and it's important to to define that. And
[02:45] that's what we try to do. I'll talk about that more in a second. In terms of the agenda, we're going to talk a little bit about HP history and CDP context, what we've done from a data architecture
[02:57] and enablement point of view. Then we're going to talk a little bit more about customer lake and databricks capabilities. I think as everybody's aware most of you are, databricks announced earlier today the release of
[03:09] customer lake. We have been in preview mode for several weeks now. We are essentially testing it out, previewing it.
[03:21] Very interesting results that I'll talk about in a few minutes, but that's going to be a core of what we're discussing later a few along with a few other CDP centric capabilities.
[03:33] Then we'll talk a little bit about the conclusion, some experiences that we have found, a few nuggets that I think will be helpful for you all and and some of the findings that that we're seeing.
[03:49] So, HP history and CDP context. I think hopefully as a lot of you know, HP Inc. we've been around for for a long time for 80 plus years. The company was founded in 1939 building audio oscillators.
[04:02] Essentially, equipment that translates currents, electrical currents, into frequency into speakers. Walt Disney was the first customer and they used it for Fantasia, the movie
[04:15] Fantasia. Over the years in the 50s, 60s, and 80s, the growth for HP continued in that space, measurement testing equipment, calculators. Mody, I know your boss, wherever you may
[04:27] be, Mody and Mody David Preston still carries his HP scientific calculator around. And then, you know, in the in the PC era in the 80s, 90s, we became a household name in that space.
[04:40] 2015, HP was a behemoth, one of the largest companies in the world, top five. Decision was made to split the company into two, HP Enterprises focusing on
[04:52] servers, enterprise solutions. HP Inc., who I represent, we're focused on PCs, printers for home, office, consumers. And, you know, since that time really,
[05:05] we've moved into emerging technologies in the industrial printing space, 3D printing, leading to where we are today, workforce experience, the future of work, really where our focus is. We
[05:19] love our consumer customers. We're going to continue to to play in that space, but we feel that our strategy, our growth opportunities, lie in the future of work, particularly digital
[05:33] experiences for the workforce. How teams collaborate. How can we make them more productive? And so, that's what we're trying to build from a B2B CDP capability. How do
[05:48] we enable our markets to go and find the right customers, find the right accounts to to move the future of work forward.
[06:03] In terms at a high level, you know, we say the future of work and when we say B-to-B go-to-market needs, and this goes back a little bit to how we define a CDP. A lot of you know, a lot of different ways to look at it, a lot of different lenses, but at Marketing what we focused
[06:17] on is we will define it based on the needs that our stakeholders are coming to us with. Obviously, data strategy, first party, third party, that is an absolute key component. Account
[06:29] profiling, whether it's ABM, mid-market, enterprise, critical piece and and you know, I'll talk about this a little bit later, but that starts to bring together the
[06:41] insights together with the data to bring better profiling, account clustering for for targeting purposes. Identity, critical component that we are about to
[06:55] undertake with with several partners, the ability not just to leverage it for identity resolution, for data quality, but for identifying buying committees.
[07:08] Buying committees are changing, it's not just the IT director anymore who's making decisions. We're also leveraging for B-to-C-to-B. Essentially,
[07:20] for for for for marketing, if we think of what we have on the B-to-B side, what we have on the B-to-C side, we have a much larger universe of B-to-C customers who are just sitting there and
[07:33] we're not really leveraging them in the B-to-B world, but what if we can in a consent and privacy preserving way understand what their business profiles are, and how can we leverage them upper
[07:45] funnel to talk to them, to engage with them, and then last mile turns their B2B profiles, and send that off to to the sales team. So, that's very important. Segmentation
[07:57] and activation, we'll get a little bit more into that. Measurement, key component obviously. Identity plays a really, really big role in measurement. And you automation,
[08:09] always on, self-service. Today, we do have our stack, but it's certainly not as automated as we need it to be. It's not a self-service as we need to be. So, these are the key
[08:21] components at a high level that we're looking to build out with the CDP. So, this is how we define a CDP in the B2B
[08:33] space. And this is driven by what our stakeholders are coming to us with, what their needs are, our experience in the space. The foundation to it all obviously being data strategy, data availability, first
[08:47] party, third party. Again, going back to the idea that we're in a different world and the way that we engage with our customers and with our accounts has
[08:59] changed. We need better data. Technographic data, installed data, that's something that we've recently brought in. Intent data, we're looking at a lot different types of intent than
[09:11] in the past. You know, we're looking at skill levels, we're looking at hiring trends, we're looking at location strategy. Really bringing a a lot more data
[09:23] to make those right decisions, to to cluster and profile accounts at the right level. Account profiling, again, I talked a little bit about that. Again, being able to bring what the insights
[09:36] team does, put that together with what our data is telling us, and leveraging that so marketers can use that as a playbook to go out to go to market. Critical component, identity and
[09:48] persona, already talked about that. Um but again, the B2C to B space uh really important component. Closed-loop measurement, right? Can we
[10:00] finally put together the end-to-end story, uh put together devices, put together individuals and households, and get a clear picture of attribution and how our journey fits with sales and put
[10:13] them put them together. So, identity really plays a big piece in a lot of this. Segmentation, um you know, key component. Uh I think a lot of this discussion is going to be about self-service,
[10:26] agent-based capabilities. Um you know, you Our experience has been it's it's engineering-heavy. Um uh data science-heavy. And while we do
[10:38] have our media team members and business analysts playing a larger and larger role in building audiences, we need to go to the next step, and we'll talk a lot about that. Onboarding and activation, um
[10:51] similar, you know, how do you build journeys through agents? Uh how can you enable that closed-loop measurement that identity offers to come in and form the propensity models and have a always-on
[11:03] process uh for your uh journeys that you're going to build out, very critical. Uh and then measurement, again, um closed-loop measurement, we need identity. Uh being able to funnel attribution
[11:16] through one uh through one funnel is very important. There's a lot of overhead in trying to build pipelines to that platform or that walled garden. Putting that together through one funnel
[11:28] is is a very important piece of what we're trying to do. Um and the other thing here from uh from uh from uh uh experience layer on your CDP, being able to see how your audiences are
[11:42] uh what their performances, how they're behaving back from the media platform, so you don't have a stale look at what your audiences look like, and the media users, business users, and the the the
[11:54] models themselves can leverage the performance data that comes back into that layer and uh leverage that for uh for continued uh improvement uh in the audiences.
[12:12] So, strategic drivers, um this is the lens from which we've been viewing this over the last 5 years. Uh you know, earlier in the decade, cookie deprecation, not sure where, you know, Google and then everybody else is
[12:24] is in that right now, doesn't really matter. We've moved to a first-party data strategy, uh but that really was one of the big drivers for us to move into that first-party data focus for marketing. Get away from cookies, move
[12:38] toward deterministic first-party data uh audiences, and from that really for marketing was born the extension of what other partners, digital transformation office, uh technology innovation office,
[12:51] they were already uh had started their journey with Databricks, um and we followed course with the uh standard that they set. We brought uh we onboarded uh Databricks at the federal
[13:04] uh at the federated level uh in HP, and that led to the first-party data strategy, obviously. You know, you we've gone through that process, we've consolidated our first-party data, we now have uh uh that data together, what
[13:16] are we going to do with it? Well, the you know, at that point in time, we went the route of a traditional CDP for consumer. Um we've been doing a lot with consumer with a CDP space, less so uh
[13:28] for B2B. But, at the time, that was our thought process. Great, we've built uh we've built uh Databricks, we've got uh our data consolidated. Uh now, what in terms of activation
[13:40] um for consumer? And And we went that route. Uh move forward a few years after we implemented our own workspace and and clearly where industry is going, um you know, the question becomes, we've
[13:53] invested so much in Databricks. We've invested so much in building our data models. Uh the overhead associated with building pipelines. Uh where architecture, right, composability came.
[14:05] Why are we doing this? And And And you know, for many of you here, this is uh necessarily revolutionary now. This is the big talking point about architecture, leveraging your data
[14:17] warehouse uh as your CDP. And so, that's the way we are thinking now. Uh composability, data-centric, uh compute-based data models. Uh there's a lot of different opportunities and and synergies to gain.
[14:32] As an extension to that thought process, you know, this is where where I think the the industry is today. And I can tell you, based on what we've seen with Customer Lake, uh I think it is spot-on. Obviously, you know, we were
[14:46] moving away from that monolithic approach, send everything you can think of to your CDP. Uh that next phase was that composable component. You have your data layer, which is your homegrown CDP,
[14:58] Databricks, whatever you may have. And then, you have that activation and experience layer, which is pushing down into uh that data layer. It's great, zero copy. Big fan of that,
[15:11] but it's missing a key piece. It's missing context. Um that experience layer doesn't have the context that your data layer has. And we recently experienced this, and we'll talk about
[15:23] that a little bit more. Um and then, you know, you move to where we are on the right, right? Composable plus lakehouse native, where that context lives. Very critical
[15:35] uh you know, for self-service and agent-based capabilities, that makes all the difference in the world. It also makes a big difference in engine for engineering purposes. It's not just about the customers. I'll talk about
[15:47] that uh in a few minutes, what the engineering teams experience has been in in connecting to customer lake, getting it up and running, uh makes makes a big difference.
[16:01] So, that was the context, a little bit of history, uh context. Let's talk about data architecture, Genie enablement, which is a big component if we're building out self-service capabilities. Um Really important uh thing here that I
[16:14] had in the previous slide, uh which my boss, uh who's sitting up here in front, you know, has called me out on multiple times. And and he's right. Being business ready or being technology ready versus being business ready. And from a
[16:28] data perspective, we're technology ready in what we're trying to do do to B2B. But we're still working on the business ready component, and we think Genie and and the agents for audiences, for
[16:41] journeys is a big piece of that to bridge that gap for the business folks, for the media folks, to what they can do with the data. So, we'll talk a little bit about that. Um
[16:53] Really quickly, you know, again, this is a a journey that HP as a whole has been in for for many years now. Uh took many hundreds of of data lakes and repositories. For the most part, we have a handful of Databricks lakehouses
[17:07] now uh at uh at HP. Um you know, it's central standards, but federated execution. That's the way we think about it. And from a data exchange point of view,
[17:19] this is one of the biggest selling points. And this almost sounds mundane, almost looks ho-hum, but I can tell you where we were 3 years ago to where we
[17:31] are today, to the fact that we are Delta sharing with our partners. The overhead is is very small. Self-service and data discovery capabilities that we
[17:44] have. You know, we federated domains and ownerships and stewardship. We want to make sure that the domain experts keep that knowledge and they continue to manage the data while we have that data
[17:56] exchange. And data is a product. So, these things sound mundane and and ho-hum, but again, where we were a few years ago makes a world of difference. The overhead that you've reduced is
[18:10] considerable. So, this is one area where we are you know, that technology ready aspect that I talked about. We are definitely there. Just a quick view at the marketing lakehouse. This is our workspace. As a
[18:23] result of the strategy, some of the key outcomes here. What we have in the marketing lakehouse today represents only 10% of what we previously had in our operating
[18:36] environment. That's a lot of new data, a lot of insights that we now have to work with. We've brought in media performance data in house, brand market share, technographics, new intent signals,
[18:50] things that were separated before. Our propensity models are now sitting in the lakehouse where before it was a different environment in the data lake as opposed to our data warehouses. Discover domains, I'll talk about that a
[19:02] little. Really key component to what we're trying to do. We leverage the medallion architecture like most of you probably do. We've moved 90% of our compute over to serverless for good or
[19:15] bad reasons. It was a little bit different in the previous environment. And now we're starting to explore with immediate connectors. That's a key component for TCO purposes
[19:28] leveraging our data flows through the connectors that Databricks offers to the Kappys of the world. Somewhere something we are taking a look at.
[19:41] So, Delta sharing. I was talking about this a few minutes ago. This is just a high-level view of what marketing is doing today with Delta sharing. The vast majority of these assets are
[19:54] business ready data products that as little as two years ago, we did not have access to. So, it it it truly makes a massive difference in that federated data model approach in having access to
[20:09] assets that you know, you previously did not to. Obviously, the reduced overhead and just a little you know, a little story about this in
[20:21] with our sales partners, we were looking to share some data for SMB purposes. Over the course of that summer, true story, three months, about 15 email
[20:34] exchanges, three escalations, and a few other you know, friendly gestures toward each other.
[20:46] You know, the outcome was very very questionable. Today, the same individual and CEO sends me a message, says, "Hey Louise, do you have this? Can we get access to this?" Within 48 hours, once with some
[20:58] governance in place, of course, there's is a Delta shared asset ready in the Delta share for this organization. Again, some of this sounds mundane and ho-hum, but the difference
[21:11] in in in in the way we operate in in the way we can turn around critical measurement reporting discovery that we need to do. It just makes a huge huge
[21:23] difference. I can't repeat that enough to the point that today it just you know, we almost don't talk about it. It just seems so so everyday. So Delta sharing it's also critical for what we found in customer lake. Again, the
[21:36] ability to federated query lakehouse Federation. Can you take we have the federated data model as I had earlier mentioned. Can we take data from these different repositories to build an
[21:49] audience. And I can tell you at least for the preview mode that we're in we're leveraging Delta share to get some of that data into our sandbox or data brick sandbox, you know,
[22:01] without having to physically move that data and build audiences. So um really really critical and and and enabled ready capabilities here. A little bit more in the foundations
[22:13] particularly around Genie how we're looking to leverage Genie and the customer lake agents. I think to our data bricks partners here.
[22:27] I know you've been working a lot on how we and what the terminology is for the for the Genie family. I know we changed Genie to Genie one on Tuesday again, making sure we have the
[22:39] hierarchy of Genie Genie spaces Genie code. We will continue to work with you guys. But anyways, this is how we're looking. How can this help our B2B ops, right? And this is what what allows us to go
[22:51] from technology ready to business ready for decision and planning for the exacts for the business folks insight and activation for the data scientists and analyst and then build and scale for the
[23:05] engineering teams. But some of the use cases we have today again planning account profiling will take a little bit closer look at that shortly. Audience segmentation and building with what customer like offers
[23:19] that's that's a big one and then we're leveraging Genie for pipelines for our intent data. We've gone from 75 topics to about 20,000
[23:31] topics that we are ingesting today to support B2B go to market and the we need to cluster at scale. We need to take those topics cluster them
[23:43] profiling so we're using Genie code to help plan that out. How do we support Genie ontology right
[23:55] everything that I talked about a few slides ago context the agent self-service. This is what composable in and of itself does not provide it's that context and semantic layer. So how are we supporting
[24:08] Genie ontology? Discover domains. I think they're in beta right now, but we are building those out for starting with our our B2B domains providing that context that wiki like
[24:21] view of our data assets metric views another critical semantic layer for our kpis. We're looking to centralize that hey let's get that out of a data science
[24:35] scientist notebook. Let's get that out of an application. Can we centralize our kpis in one layer as a source of truth build business continuity. Oh and also
[24:47] to provide that context to Genie. So when a business user goes to Genie and ask a a about metrics and kpis. we have that
[24:59] living in that semantic layer. So, that's a key component. That's just one example we have up and running today for our media metric view starting with our performance media view. We've built out the media metrics,
[25:12] covers 10 26 dimensions, 10 measures, and that is is is leveraged for dashboards when it comes to media spend and similar KPIs in the media world. So, really critical component here.
[25:27] Equally as important, I don't know if Mark, I don't know if you're here or not. I know you're here in in at the summit. You know, Genie, how do we make sure that, you know, whether it's B2B or
[25:40] anything else, how are we instructing and and and steering Genie in the right direction? So, we've built out on our workspace this published best practice guidance. We have an instruction set and
[25:54] a skill set applicable on Genie. You know, things like, "Hey Genie, rather than building first, plan. What is the plan to build this pipeline out? What is the plan for
[26:07] whichever topic we're bringing to it?" Very critical. What we get back is a plan. And we've told Genie, "Don't execute unless you have our permission." So, before it executes any component in
[26:19] the plan that it puts in front of us, it's going to ask us. And we have other things embedded in the instruction set for Genie, how it should behave, the type of documentation it should provide for us so we can be
[26:32] confident in what it's producing. Now, by the way, I know everybody loves documentation, right? I'm sure all the teams here do a great job at documentation. We are now leveraging Genie as our documentation assistant.
[26:44] And that's a very, very good use case that we're really excited about. So, again, these are foundational items, how we're leveraging our data strategy, Genie, to be able to
[26:56] then leverage what comes next, segmentation, activation, measurement, and really provide our business users and executives the right experience. Part of this will be to
[27:09] to build out Genie one, I think it is Katie now. I know it was Genie a week ago, today it's Genie one, it was Databricks one a month ago. But yeah, we need all of this, right? All that that the semantic layer, the
[27:24] the the instruction sets to make sure that we are ready when our executives and business users are are ready to go on to Genie and leverage it for their purposes.
[27:36] So, GTM CDP capabilities. Let's talk about now that I we we've set the stage for the data strategy, very much ready, we're always adding new new data.
[27:48] What and how we're thinking about Genie and enabling our users to Genie, how does that help the rest of our capabilities that I talked about in one of the earlier slides, the you know, the six uh
[28:00] capabilities that we're focused on. Account profiling. Hey, guess what? We talked to Genie code and said, "Help us out here." About 5 months ago we had a discussion with our marketers, and since
[28:13] then they've taken a particular route for account profiling, but this is an activity where you know, think Hoovers, think PowerPoint PDFs that talk about account insights. How can we leverage
[28:27] that with our data to help inform and support our marketers in building strategy or a playbook for B2B go-to-market. So, you know, we went to Genie, we talked to Genie, it put this
[28:40] plan in front of us. Number one, it built the volume itself for the for the unstructured data. It created the account feature store
[28:52] with the right data sets, technographics, firmographics, really the core commercial data sets that that feature store is going to need.
[29:04] From there, it went to strategic scoring. I forget which of the models it used for that, but that was the next step. From there, clustering of the accounts.
[29:16] And from there are the outcomes. Ready for Genie profiling, a cluster playbook dashboard for viewing what those clusters look like and then cutting that data in many different
[29:28] ways. The integration with our account level data. What do those clusters look like and how are they integrating into the account level data set? And refresh,
[29:42] right? Genie has suggested, "Hey, this is what we we should do from a refresh point of view as you're feeding data into it." So, this is something that we've POC'd, but we want to take back to our data science team and the marketers
[29:54] and put it in front of them. Several of the scenarios they put in front of us, it takes them 7 to 10 hours in the old world to do this. What if we can build this framework for account
[30:08] profiling and just make it an always-on exercise. So, this is something that's very interesting work currently working on.
[30:22] Identity and people data. One of the ways we're looking to leverage identity in the new world as opposed to the old world is we want to bring that in to our warehouse into data bricks into our lake house.
[30:34] Historically, what you've seen is you send your data out, identity identity lives over here external to your environment, why? Why aren't we bringing this back into our lake house? So, we currently uh
[30:47] you know, we have some new uh providers and partners for identity that we are going to be up and running. And the idea here is we are going to bring that identity back into our uh our environment. Um
[30:59] the the you know, the universal scores or or identities, so to speak, bring that back into our our Databricks environment. Obviously, from a measurement point of view, make sure that uh our data flows in through that
[31:12] uh as it goes out to the to the media uh platforms, comes back through the same identity service, so we get that holistic look, that closed-loop measurement, back into our models and refresh and do it all over again. So,
[31:25] identities uh very critical. Um again, extensibility for B2C2B. Uh HP has hundreds of millions of consumer
[31:38] customers. Uh we have much much much much smaller set of that, obviously, for contacts. What if in I think what we've have found based on some initial studies and and and views
[31:50] is, hey, even if we get 10% of that consumer world, all consent uh preserved, and understand what their profiles are, what their
[32:02] business profiles are, what their you know, job titles, what do they do, and include that in our journeys for commercial. That opens up a big uh opportunity and and again, we need identity to do that. Um
[32:19] So, uh you know, key key area here we're looking to to move forward with. This is one of the next steps we're currently on. Uh customer link. Uh here we go. Um
[32:33] So, we've been in the CDP space uh as we all kind of know it today, for about 5-6 years post post cookies. Every time we were doing an RFP or we
[32:45] have a discussion with CDP provider or we moved forward with some with some implementation, tell me if this sounds familiar. Months of back and forth meetings.
[32:58] Documentation up the wazoo. Uh you know, getting things ready to make sure we're ready for that go to date. You know, when are we going to go live or when are we going to configure the
[33:10] data models? Are the pipes pipelines ready? Um cyber, privacy, just anything and everything. I think we've all been there, right? So, we've now been on customer lake for
[33:23] 3-4 weeks approximately. Um without a doubt, and going into the first meeting where we were going to put everything together, I was a little bit nervous cuz I said to myself, we have not done
[33:36] enough here. We've only had like one meeting. I got four bullet points from Cindy. I don't know if Cindy's here. Um but I got four bullet points. This is what we need. And then I remember going into that day, something's not right
[33:49] here. Um we're going to go into this meeting and look like fools. Anyways, we went into that meeting. Within an hour, we had our data models in place. We had attributes published.
[34:01] Um we had we were ready to build audiences. And of course, it's that native experience on Unity Catalog. Our data models, we double click into your
[34:14] existing schemas, double click into your tables, double click into your definitions. It was really eye-opening in how easy it was. I mean, it felt like an everyday exercise.
[34:26] Time to to market, time to turn around audiences. Can't say enough of of of the productivity and the reduction in overhead that that represented. Um,
[34:39] key other a few other things very important here. The ability to leverage more than one data model, uh, very critical. We've experienced that. Uh, more than one data model, you're going to be forced to,
[34:52] uh, to stand up a new instance. That comes with cost. Being able to bring, you know, different systems, uh, different workspaces, doesn't even have to be Databricks, could be Snowflake, right? And, uh, multiple data
[35:05] models into the same customer lake instance to build your audience, and this speaks to B2C to B. I need my consumer data model, I need my customer or my commercial data model, identity is going to bridge them, but I need to make
[35:18] sure that my instance for segmentation can accommodate them both. Um, and, uh, that's a really, really big win for us. Uh, certainly reduces cost and, uh, complexity.
[35:31] Attributes, uh, same idea here. Um, you know, in terms of the agents, uh, there's different paths to to build. You can build it through the agent, there's a manual approach. Um, again,
[35:44] within an hour we had, uh, mapped the data model, we had, uh, published the attributes that we wanted. Um, the native experience, that context, feel like you're home. You're building out the audiences, you're putting your your
[35:56] your your data model together, and frankly, it felt like home, it felt like just an everyday exercise. It was not a big deal. So, so far, so good on these things. There's a lot more where we're planning on,
[36:09] um, you know, where we're obviously at this point we're looking to build a few, uh, audiences and export them. Um, but my takeaway from this has been lakehouse, uh,
[36:22] the fact that it's na- that the fact that customer lake is is on lakehouse, the fact that you are working with your homegrown data model, which you look at every single day. The fact that you can go double click directly into your
[36:34] tables, into your attributes, massive massive productivity gain. Um, I you know, I think some of this would have taken us frankly 2 months, 3 months in a previous world where we did it in 1 hour.
[36:47] Uh, ma- ma- massive um, uh, productivity gains. So, some of our next steps, we're going to look to sync to the media platforms. Um, we're going to export the audiences, build the campaigns. And once uh, the
[36:59] engineering team and some of our uh, data scientists has um, uh, built that that MVP, we're going to start exposing it to our to our users, to the individuals we want to leverage this, to the media teams. Uh, business
[37:11] teams. Um, and that's going to be followed by journey orchestration. That is in product development. Um, we're working with a team to build the right uh, use cases. We're very excited about that.
[37:24] Um, We only have 2 minutes left, so I'm going to walk uh, a little bit fast through this. So, we have a value framework we're building for B2B. And uh, you know, this is a select uh, set of components that
[37:36] um, that what we're seeing from Databricks, whether it's on the data side, whether it's on the customer lake side, uh, help us get there for insights, media planning and execution, reporting and measurement. Certainly,
[37:48] you know, license fees, those types of cost, uh, being able to leverage things like AI, BI for users. Um, uh, you know, the propensity models, how
[38:00] fast we can turn those around with Databricks. So, there's a wide variety of different uh, value components for outcomes, productivity, TCO that we're looking at. Um, this is an example of a
[38:13] a potential savings in the best case scenario. Um, we spend a lot of money on licenses and platform fees. And and frankly, it's it's it's more than a lot more than compute. Um, potentially
[38:26] there is a big opportunity here and and we're surely taking a look at that. Um, and you know, the conclusion, the way we look at things, it's the CDP landscape and architecture has evolved considerably. Define it for your
[38:38] organization. I think it's getting harder to define given some of the the the new um, you know, the evolution in architecture and capabilities. Uh, make sure your partners work hand-in-hand to deliver those capabilities especially in a composable
[38:51] world. Uh, Unity catalog, that context that Unity catalog offers makes a huge huge difference. Um, can't speak enough to uh, time to production, uh, the context it provides. Composability is
[39:05] the new norm, but adding that contextual extensibility from that data layer it makes all the difference in the world. Uh, technology ready versus business ready, making sure that that's in place and Genie and and other components lend to that obviously.
[39:19] Um, but that's a big uh, component which we feel some of these new capabilities help us get there and not just be technology ready. Uh, and the TCO opportunities. Um, those are some of the takeaways uh, that that that I have,
[39:31] that we have. Uh, we have 35 seconds left, so I am going to stop there. Um, thank you very much and I'm happy to take any questions or comments that uh, anybody may have. >>

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.