How Ad Tech Runs on Databricks

Summary

• Ad Tech Scale: Leading ad tech organizations leverage Databricks to harness vast advertising data and power the marketing ecosystem at scale.
• Data-Driven Decisions: This video showcases how companies overcome the challenge of processing infinite data to optimize advertising effectiveness and drive measurable business outcomes.
• Industry Showcase: Trade Desk, Integral Ad Science, Adobe, and other major players demonstrate their real-world implementations on the Databricks Data and AI platform.

How Ad Tech Runs on Databricks

Talk by Abhishek Yadav, ML and Data Platform Team Engineer, LG Ads ; anj yamsani, SVP, Engineering and Product, Influential (Publicis) ; Atul Saurabh, Global Product Director, MiQ ; Cyrus Mohammadian, Head of Data, Influential ; Kayvon Raphael, Data Engineering, Magnite ; Scott Simony, VP, Sales , Databricks
Discover how leading ad tech and mar tech organizations including The Trade Desk, Integral Ad Science and Adobe leverage Databricks to power the advertising and marketing ecosystem.
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

FAQs

How do ad tech companies use Databricks?

This video features panels from leading ad tech organizations who share how they leverage Databricks to process and analyze advertising data at scale. Companies like The Trade Desk, Integral Ad Science, and Adobe demonstrate practical implementations to optimize marketing effectiveness and drive business results.

Why is data important in advertising technology?

This video emphasizes that modern advertising faces enormous volumes of data. The presenters note that companies must process near-infinite data streams to understand what advertising approaches work and drive client outcomes in a data-driven landscape.

What are real-world ad tech use cases on Databricks?

This video showcases practical examples from MiQ and other industry leaders building new products on Databricks to tackle data fragmentation and evolving digital landscapes. Attendees learn how the platform enables faster innovation in advertising technology.

Which ad tech companies are featured in this video?

This video presents industry leaders including The Trade Desk, Integral Ad Science, Adobe, Influential, MiQ, and Magnite. Each company shares their journey of leveraging the Databricks Data and AI platform to optimize advertising strategies and solutions.

Full transcript

Good afternoon everybody. Sorry for getting started a little bit late. Some technical difficulties that I think we've overcome. So I wanted to start today um with a quick story. Once upon a time there was a merchant who lived in Philadelphia. This was back in the 1800s and this merchant was widely credited with starting the first department store which is kind of a precursor to a mall. And this merchant's name was who is widely credited with a quote that we still use today.
He said, "Half the money I spend in advertising is wasted. I just don't know which half." Now the problem a hundred years ago was that there was very little data to be checking out what was working and what was not. The problem today and over the last 20 years has been near infinite data that's out there that we are all trying to process and make sense of how we make better ads that ultimately drive the outcomes that our clients want.
Uh I work here at data bicks and I lead our adtec business and adtech is one of our largest and fastest growing segments at the company and it's not uh a it shouldn't be a surprise because there is just so much data that all of you in the audience are looking to harness and turn into um some real insights and drive business results. Today we're going to be joined uh by a couple of different sessions inside this 40-minute uh uh session.
I'm going to have a panel of esteemed uh colleagues come up here in a couple of minutes and we're going to talk through some of the challenges and opportunities that they're facing on the platform. But before we do that, I want to welcome my friend Atul Sorab who is the global director of product at MIQ to walk through a demo of what they're building and I will let you take it away tool. Am I audible? Am I audible? Yeah.
Awesome. Thank you guys. Um really exciting for me to stand here and present some present in front of like some of the sharpest bright brightest minds in our advertising and technology industry. I'm I'm really excited to present some of the new product innovation that we've been working at MIQ over the last few years um to help us take a big leap in this era of fragmentation and proliferation u as we continue to grow and as the digital media landscape continues to evolve.
Before I go into like the product and demos, I do want to, you know, just quickly walk you through what MIQ does and why did we build this product in the first place. So we are MIQ and we've been in the programmatic business for last 15 years. Uh we started as a performance-based managing partner like programmatic media partner and then as the digital media landscape continued to evolve, we've evolved our business into trading CTV campaigns.
Oh wow. Wow. Sorry guys. This is my this is my laptop and there's going to be some issu there may be some issues with the Wi-Fi by the way. That's what we were dealing with before we start the session. Yeah, we've been in the business. We continue to evolve as the digital media landscape has been evolving and we've grown into CTV, YouTube, social and now also taking care of verticals like political, healthcare, um and gaming.
So whilst we've been doing this um in order for us to be able to add value for our clients, we've really had to you know partner with a lot of vendors um and now our partners in this ecosystem and that means like we've we've had to get their data sets into our ecosystem to plan our campaigns better gotten integr integrated with a lot of DSPs to be able to activate on these campaigns and then also partner with measurement companies to be able to report on KPIs that our clients want to see.
And this has been a bit of challenge as we've been growing now. We've grown into you know 28 global offices with 1300 plus employees. And the challenge for someone like me in the product team is how do we create something that is extensible that is scalable across all regions for all kinds of channels that we want to go into in the future. And that's why we built something that we start calling MIQ Sigma. This is the product that I'm going to demo today.
But before I go into the product, I want you all to hear about the why of building this product from three perspectives. Unifying our data vendors and partners that you saw sample of that was there on the slide. Unifying a consumer view for our marketers um across you know what they're watching, what they're browsing and what they're buying. And then also trying to create a connected and a transparent workflow for our clients who can see how the campaigns are planned, how they are activated upon and they can access the reports in a self self-s served capacity when they come to MIQ.
So all those partners that you saw, we've connected them through Sigma irrespective of what kind of formats they share their data in, irrespective of the velocity in which the data comes into MIQ. We help we use these data sets to help us plan across any audience activate across any unique supply path uh that's available and then also report back on the KPIs that our clients want to see. So this is the first perspective of like you know why did we want to unify things uh more from a partnership's perspective and then from a marketer's perspective every marketer wants to see what their consumers are watching browsing and buying and using our partnerships with you know our TV partners with our SSP and DSP partners and also uh you know commerce partners we've been able to create a unified view for a marketer to be able to see what their consumers really do and then last but not the least u you know to be able to reestablish a connected workflow.
We we started working on this flywheel of we need to plan campaigns. We need to push them into uh you know our audiences, plan creatives, run campaigns and then report back to our clients. And this all needs to be connected. Now for this we came up with a vision which was like really bold and ambitious a unified experience a world-class programmatic across any platform, channel or KPI. Now you know for someone who's in a product uh capacity and who knows that this job is not easy with the amount of fragmentation we have across different partnerships.
The first thing that came into our minds was we need to rearchitect our backend and create data and rearchitect in a way that you know we can harmonize data sets across different formats and velocities that are coming in through our different partners but also create something that is extensible enough. And an example of that is you know someone from Australia in MIQ would call me and say hey we want to get into social uh product we want to start start selling social campaigns can we extend our offering that we have in US to Australia someone in US would come to me and say we are partnering with a completely different you know data data data vendor in the in the US can you also ingest that data and have it flow through our whole whole ecosystem and that's that's a challenge and that's why we wanted to really build a solid backend in order for us to uh really scale this technology and that's where our partners data bricks come in.
They really helped us build this new architecture on top of uh you know all the data partners that we have all the initations that we have across DSPs um and pivoted in a way that could allow us to really release a product that is faster more reliable and extensive. Now I'm going to go into the demo very quickly but uh before I go into it there there are two parts of the demo. The first part of the demo is how a marketer can come to this platform, plan their campaigns, push their uh segments into DSPs and then the second part of the demo is going to talk about how we using some of the agent AI features that are uh that are based on top of data bricks features.
Right? So let's get in the demo and hopefully see the Wi-Fi work. So here's the platform. Uh you see intelligence, audience, creatives, campaigns together. It's not just in the UI. I'll also give you an example of how we've unified these together. and then streaming all the buying watching and browsing data on the homepage for me for marketer who comes in can go into intelligence they can see TV context and commerce which is the watching browsing buying story that I've been talking about in this case I'll take an example of TV uh you know use case where where a petare client fresh Pet is wanting to look at how they can advertise across social uh total video so they have linear OTDT and YouTube insights they can come in they can see what is happening with their competitors what's really working for them from ads perspective and this is really cool place where they can see how their reach and frequency has been across you know different audiences that they have reached and the fun part is you know they they've spent like close to 60% uh of ads for top 20% users and that's why I'm creating a strategy to target underexposed users directly from this and it'll be pushed to our audience section that I'll come back to when I go into audience and I'll show you how we activate more now they can continue to browse across different sites this is our latest addition where we've added YouTube in the mix of linear and OTT.
So anyone can come and plan across like you know all these three platforms and the coolest part is you can compare it with your competitors and you see what your competitors are doing on YouTube. You can learn from them what kind of formats what kind of ad lens they're running activate upon them. You can also look at their reach and frequency and probably build a contesting strategy to target their underexposed users for your brand.
I'm going to go into the audience and show you how I can find this segment that I just built here using insights and push them into uh DSP. So you see some of the overexposed underexposed segments here. They're already present in the audience section. I come here, I add them, I push them into our automat automated campaign setup templates and this will directly go into Xander and that's how the segments will get activated.
Now I'm going to come back to you know this this part a bit more but there's there's a genai persona generator which also gives a view of who the audience is before you start planning your campaign. So I'll come back to this in the second half of the demo. What I really want to do is quickly summarize how data bricks have helped us build what you saw. So the intelligence platform that you saw has been built on top of like multiple jobs.
Um there you know for everyone who's on on the product tech side here and has worked on data engine side they know uh they know we we do like a lot of batch processing before u you know of big data before we really build these kind of dashboards. So um earlier uh before we start partnering with data bricks we had all of this based on a map produced compute stack. I won't name the vendor but I think every every person would know what I'm who I'm talking about.
And then we moved into data bricks uh serverless and that helps that helps us immediately reduce the processing time from 12 to 5 hours which is massive but this is not just it there's more on top of this and you'll see of course because of the reduced time we had reduced costs um and in this industry as was mentioning the data continues to grow u brands in our intelligence platforms continue to grow um and the data became really huge um the order of data here is you know at a daily level intens of terabytes and at a monthly level like in pabytes.
Um and then last but not the least uh you saw the dashboards loading pretty quickly. Again all of that is u using data bricks technology serverless unity catalog and then I also want to call out you know some of our partners in the team uh u they've really been great consultants help us transform this tool um to where it is today. I'll come back to the second part of the demo. We'll start with where we left. So here's the audience part where you can look at more segments not just the segments that I push from intelligence but you could learn from this set up your campaign and here's a trading agent that we built which helps you manage these campaigns.
So I'm I'm asking it which of the fresh campaigns have I set up and it's going to come back to me and tell me uh you know what are the campaigns that are live today across all the DSPs that we're using and we have 12 DSP integrations at this point in time and within 12 DSPs we have 30 plus seats integrated. I'm also asking it more about like you know read more things about how is it pacing and it's going to give me an information about how the campaigns are pacing.
Um I can also write back the write back in individation is also there. So I'm asking it can I change the pacing although it's not going to allow me to change because I'm a product person and it's only allowed for traders to go and change things but um it gave me a logic of why changing pacing is not right at this point. This example is very much for you know your commercial teams where we've used the genie feature of uh data bricks.
We used we've created like a big data cube of historical data sets for our sellers to come and you know query about like different kinds of formats different kinds of channels. Here I'm trying to search about you know CTR sort of benchmarks for petare clients in Canada and this is all based on historical data. Um this is using the genie APIs and any commercial person can come write natural language query here can fetch data sets query and get get you know the answers out of it.
Um so this is completely built out of uh AIB genie. And then lastly I just want to say like as we continue to grow u you know uh a tech stack on top of data bricks we are continuing to build more on top of AIB genie. And then in the future we want to use what we saw earlier today in the keynote like lakebased um and more agentic operational features within um within data bricks to be able to evolve not just trading agent but we also have ambitions to build like sales agent a service agent within the company um and create a solid architecture around it.
So um that's it that's what it is. Um if if anyone really wants to get a real demo of this tool online please scan the QR code and I'd be happy to assist you guys. Thank you so much. Thank you, Atul. Awesome job. All righty. I want to ask my esteemed panelists to come up here. , and an please join me on stage. I will we'll get set up here for the panel. Alrighty. Um, welcome to the second part of our session today. Please take a seat.
I want to introduce you to this panel. I'm have every one of these uh folks actually introduce themselves starting here. Go ahead, Obby. Hey everyone, BB Engineering at LG Ads. LG Ads is a global CTV company with access to 200 billion plus screens globally from LG Electronics. Hi, I lead product and development at Influential. We're an influencer marketing software company. Um, we recently got acquired by Publicist. to talk more about what it is and some of you may not know that stuff so I'll go through that.
Cool. Thanks. Hey, I lead our core platform data engineering group at Magnite. Magnite has a programmatic video advertising platform as well as a large display SSP. We service both buy and sellside clients across CTV, online video, display, and audio. If you're wondering what that actually means, it means we help power the ads you see anywhere from the apps on your phone to the website you visit for news to the billboard you see while driving to the airport to on the airplane itself to while you're listening to Spotify to while you're streaming video at home on your couch.
We currently reach about 99% of US streaming supply and every day across all those formats I mentioned currently run about two trillion with a t auctions and serve a couple billion impressions. So that's idea of our data scale. Thank you. I mean these are uh three three big companies doing a lot of big data processing. Let's kick us off actually with a question for all three panelists. You guys are all at different stages of business, different stages of your journey.
Talk a little bit more about what either happened on the business side or on the technical side that made you decide to move to data bricks in the first place. Um so actually this is my second data bricks migration. You can imagine I'm a firm believer of the platform. Yeah. Um and I'll start with this. You know how many of us have had calls or a nasty slack message or an email about someone not getting the report on time or their jobs are failing or you know someone isn't able to get a complete data set right I think that's where we were when we started this journey less than a year ago and I think at this point we feel very comfortable within our scope of migration that we've achieved as well as the business uh opportunities we've unlocked you know we've got anecdotes coming from our uh business partners where hey this job ran for 72 hours and I had to run a software on the computer so it doesn't fail because we're running on prem there's a lot of restrictions to now hey this finishes in eight and a half minutes and they're counting every second of you know uh extra time that they have now uh you know that's that's on the business side on on the technical side we focus heavily on the architecture we're global we're we have a massive scale and it opens up to a lot of responsibilities both from a consumer as well as compliance side.
Uh we we focus on frameworks, we focus on standardization and that is the only way we're going to scale. We want to get to a point where where we don't have stability problems. We're able to scale across globally. Um, and we're able to do this to facilitate the growth of our business as well as follow the compliance and regulatory laws that come in uh with all these uh things that we try to uh focus on from a business side.
Uh the medallion architecture has been great for us. Uh you know, we're able to using data bricks platform, we're able to consolidate all of our data into a common platform. uh we're able to build frameworks that help us accelerate the onboarding of data. or we go make a new partnership with a DSP or we go enable an SSP you know there's the first thing comes in mind is okay how is the data going to be here right and if it's going to take us two months to onboard that data isn't really helpful right so with data bricks we're able to onboard all this data we're able to apply custom transformations we're able to leverage the different variations of compute that allow us to uh do things for uh for different sizes and as this data gets mature Third, we utilize Unity catalog heavily to make sure the right people have the access to the right data.
Uh we can since we're global, we can apply localization techniques. uh we can adhere to GDPR, PIA, all kinds of compliance rules whether it is for a company specific or domain specific or something that is uh pushed by our parent company LG which is you know uh similar global scale and have a lot of hardware uh things that we need to keep in mind and once you know this data goes through the process we put it in a very clean staging layer where now you know you can have reports be built off of it you can build segments you can build audiences start pushing it into our own ad stack or a third party ad stack and then you know eventually get to a point where we can run machine learning or data science and have all this data ready into AI agents that we're uh sort of uh encompassing within our ecosystem.
So uh you know we're we're very happy with the journey so far. We get very good feedback from our business partners and now we're going to facilitate this to take this to our customers. So we're able to partner with them and you know use the data bricks platform whether it's clean rooms or whether it's doing integrations or creating agents but we're mutually benefit from. Thanks for your trust a second time. An tell us more about your journey at Publus is influential.
Right. And you know on the background on influencer marketing I mean how many of you know Mr. Beast or maybe your kids know right? So he's the number one influencer and what's happening in the market is that the traditional ads are not performing well and you cannot scale them with so many varieties. What's the creators are authentic and they're producing this original content and the ads that we create through the creators are performing 2x to 3x than the traditional ads and that's where media companies like publicist now started allocating higher you know budgets towards the influencer marketing just like how it went through CTV it went through digital out of home now it's a time for influencer marketing with that influential company was on the rocket ship right so we were growing very fast but our technology we were having I don't want to name but we were having technical difficulties like just get our simple ML jobs working partly because we were only four member team right and we didn't have like devops access and all that so just um less than a year and two weeks after the last data summit we signed up with the database and within six months with the four member team we were able to build industry's largest AI search database with the multi-billion vectors um for influencers you know marketing software and you know the team is very like they helped us handhold us through the slack through everywhere and the platform our team calls like data bricks is our invisible team member with the superpowers right and it's become very easy for us um we had like very difficult challenges with others right so especially we took care of like you know we leverage the unity catal ml flow primarily because we run lot of machine learning algorithms to enrich the creative profiles, right?
We see some text on the posts, so we know, okay, they're talking about organic food, they're talking about cats, they're talking about babies, stuff like that. And that's very important for finding the good suitable creator, right? So that's where we help brands to find the suitable creator who can give like 3x, 4x, you know, one. And so that software is you know industry's largest and it only happened primarily because of you know data bricks help.
Thank you an feels good to be part of your team. So thanks for thanks for including us. uh tell us more about your journey at Magnite. Yeah we ended up on data bricks in the beginning uh a little bit out of necessity. We inherited a legacy pipeline and a kind of disjointed data ecosystem in that world and we needed to rebuild it. Uh we had a crew of people in New Spark. That's what they were great at and we turned to data bricks.
Uh we reimplemented the pipeline in data bricks. Um as an also said great crew shout out to our data bicks account team and all our essays and everyone. They were killer in working with us uh as we pushed the limits of Spark streaming. Um very fun nights. Thank you, Satish. We um then as we started building our foundational pipeline, we moved further down the data platform data bricks to warehousing. So moving uh off of a legacy reporting product into data bicks's SQL warehousing to power client-f facing reporting.
We then opened up uh giving our internal tech ops teams and account managers and things like that direct SQL access to the data uh that they use to make their decisions and troubleshoot keep our clients happy. Uh we brought our data science crew onto the platform which was amazing. They're like oh I can spin up a spark cluster and access all our log level data and just like type something out and get an answer. Fantastic.
Great. Uh started moving them in. Everyone's happy. Uh we're continuing from there to build a fuller data platform picture. Um as I mentioned we have a lot of data. One of our platforms itself generates about three and a half terabytes of data each hour for log level data. Uh so we're working on moving that to iceberg leveraging data bricks uh up and cominging iceberg functionality to get that efficient cost optimized queryable usable.
Um so we started from like a very foundational place of necessity and slowly expanded to leverage all the capabilities of an endto-end data platform. uh and obviously all the things that come along with that like the governance, the quality, the access controls, things like that. Um very excited to move to the step of incorporating that in the broader ecosystem and using some of the openness and interconnectedness and data sharing that I think we'll touch on a little later.
Awesome. Thanks. So let me actually start with an individual question and I'll start with you Obby. Um you mentioned at the end of your introduction a lot more work going on the data science side things that were echoed by an by KBO as well. Um talk to us about some specific examples of how you guys are using AI machine learning maybe even actually into the agentic side how that's playing a role in what you guys are doing with customers um and helping them.
Of course but I'll go back to the architecture uh that I you know talked about earlier. uh we focus heavily on building scalable architectures and having the right components in that architecture. Uh we're we're d we're doubling down on our AI agent uh program. Uh we're we're going to we're going to use a lot of large language models that are able to plan and reason. And with the evolution of newer protocols and knowledge graphs, these models are going to have access to information that is contextually relevant to the user.
It is going to have memory to store interactions and it's going to have access to large amounts of vector databases to fetch information and act as racks. And we're we're incorporating the MCP protocol that's that's allowing allowing the LLM to do things that historically hasn't been. Historically, LLMs can you give it a prompt, you ask it to type an email, and then it kind of ends it there. With these newer protocols, we're able to take that to a next level and apply actions to it.
But it's fetching data, applying PII, transporting it to a PDF, and stuff like that. And we're going to take this architecture and apply it both internally as well as externally for our customers. uh our our internal goal for this year which we only have six months left is to have 40 live AI agents within the ecosystem doing different tasks uh that are focusing on operational efficiencies along with uh propagating the media planning forecasting uh feasibility and uh once once you get a media plan then get on the delivery side and as we're delivering media how do we optimize it with flight and when things are are optimized and delivered.
Then how do we close the loop with reporting and and show our customer value? And that's where the customer side of it comes. If we can build AI agents that are real time give given our customer insight into all the great work that we've been doing for them, then it it helps them it helps us gain their trust as well as helps them spend more money with us. And if we can do this in a very fast iterative way, then they can spend that money with us even faster.
Uh so that's kind of how we're thinking of leveraging AI to kind of propagate our business, generate more revenue, keep our customers happy and and do that over and over. As a former TV buyer, I want to say thank you because you know a lot of the tools that unfortunately a lot of the buy side uses right now to get insights from publisher side or supply side is just not adequate. Takes forever. Yeah, it takes forever and it's very rigid.
Right. So it doesn't match at the time. Let's let's keep on talking about right. Um Kavon I want to switch over to you. You know you mentioned um uh something at the end of your introduction a lot more about access to like direct access to logs and audience information. Talk to us more about how Magnight's thinking about exposing that and kind of making that more useful for your co for your customers. Yeah, we'd be happy to.
So this is something I touched on just the end of my previous segment of looking forward to the future of like expanding with integrations and other partners and really leveraging um data bricks as a platform that connects. Uh Obby also touched on this. Uh we're now starting to explore and plan direct integrations with our partners and our clients to bring uh audience data. So like segment data about people available out there to serve ads to directly onto our platform and activate that activate that in current uh current situation to do that you have to you know sometimes send data through intermediaries third parties you're exporting a puddle of data to an S3 bucket for someone uh you're sending that back out to them you're they're you know sending all all these different wires different routes uh The advantage of being able to leverage something like direct data sharing or clean rooms through data bricks is it brings a lot of that down to zero.
Um you get increased data governance for example. So uh if OB's going to onboard his audience data audience data segments onto our platform he can do it directly through that connection maintaining governance of it. uh knowing who's seeing the data and where it is instead of sending some data out to somewhere and maybe you know something happens to it along the way you're not fully in control of it which is great. Uh same for us sending that data back those exposure logs back out to them.
We know we have control over the data we're sending out uh directly to them and also cutting out a lot of that time in that cycle. So uh you have it directly through a platform instead of again waiting for an export checking a bucket downloading it etc etc. So that's something that we're really excited about actively planning and implementing with partners and clients via data bricks um as we all get to that place. Thanks.
It's exciting. Looking forward to helping support you guys get that in place uh in the coming weeks. Um, an you know you you your your company at influential has come through a lot over the last year uh with a small and mighty team. How do you guys figure out a way to organize the world of influencer marketing which is inevitably a very large problem of how to connect the right brand to the right kind of influencers.
How do you do that? How do you use platform like ours to help? So I look at this as like two pieces. One is finding the right creator and the second part is that in your demo you had like respect social right so those are the type of campaigns we execution is second like analytics of those uh campaigns and performance predictions in case of searching for a right creator what we do is that we make around 50 million API calls into the networks through API integration we get up all of that data and data bricks helps us to analyze that data run through ML models and categorize them, contextualize them, vectorize them, embed them and put into AI vector database.
And with that now I can go into the vector database and I'm looking for a creator who is lives in Texas and who is in like 30 to 40 age and who has two kids or maybe kids more right and I'm trying to sell like baby parks and so with this I'm able to pick the right creators and display them using all the MLflow and you know uh the stuff that you The second part is the analytics. Now that we have connected to the network, we need to get all the metrics back and see how it's working and change right if we need to send it to LG or send it to somebody else, we need to decide based on the performance that campaign is going up and that is time sensitive and data bricks helps us to scale right if you need to scale up like 20 instances whatever to process it right and that's been very handy for us to with the very little DevOps now we can process on demand as much data we want and it's been you know incredible and we're looking forward to already like a lot of the LLM are integrated we call LLM through the data bricks environment um a lot of the AI stuff is implemented at every step in the first phase one of the thing that we don't notice is that brand safeties is very important for us.
We bring the creator, but what if they posted some pictures, they're like smoking or alcohol or something. So, we have to mine through all of their posts and images to validate that the creator is safe. So, data bricks platform, all these operations help us to do that job. Love it. An thank you. Um, we're going to have time for maybe two questions at the at the end. So, I just want to encourage you guys to start thinking about that.
We've got a microphone over here on the left side looking at the stage. Um question I want to ask all of you and I'll ask actually Kon to start first. Talk to me Kon about some of you know if you could wave a magic wand right if you could solve some data issue in the ad space. It's hurting your hurting your you know ability to do something. What what is actually that one thing you'd love to be able to solve for? Just just one only one.
We only have time for one. Uh I think I mean there's there's a lot of challenges that come with data at our scale. I think one of the like core fundamental ones that is trickier to solve is data quality. I think that is the one I would pick. U you know like sanitizing normalizing your data. Uh my brother who also works in data engineering likes to quote a computer science tenant of garbage in garbage out which I firmly believe in and that is a very hard problem to solve especially when you have a lot of different data streams coming in different data pipelines even when leveraging shared libraries components logic to do that it's still um it's still a hard job to make sure you write and that is the starting point of everything else downstream.
So if you don't do that right then the data you're giving to clients to your internal people to the people the pacing application that determines how much money you should spend etc the audience people like all of that will be wrong. Um so that is that is what I would pick. Awesome. Uh same question to you. So right now the influencer marketing is not as standard as the regular ad business. In the mixed model marketing like we want to make sure that we can predict and model and you know create the streamless pipeline integrated into the regular adtech the influencer ads program.
So that's what I want to do. If I have a magic wand, there are a lot of benchmarks. Every company has its own benchmarks. And I want to standardize and see that okay, we can all agree if a creator is performing well or not. So, we'll see in two, three years. Awesome. And then, uh, Obby, tell us what's what's the one thing keeping you up at night? What's the one data problem you could solve in the ad space? What would you do?
I have a short answer and a long answer. I'll save the long answer for the happy hour. My short answer is standardization or ask the industry. You know all of us kind of sort of in certain ways whether it's smaller magnitude or larger magnitude suffer from similar problems whether it's interfacing systems joining data comingling data and taking it out in in our worlds uh or creating agents that we can all rely on to do certain things collaboratively whether it it is one organization owning it or us collectively owning it.
So, I think there's a lot of work we can do and it would free up tons of our times so we can focus on on revenue generating things and not have to grind through these problems that can uh that you know sort of keep us up and aren't really uh hard to solve but it's just there's a lack of things uh being able to talk to each other because of standardization. Thank you for the questions. Just to pull it all together, I want to say thank you to these panelists for joining me on stage today.
I want to say something that I think I heard from them as well as actually another important client who said something to me today like there's so much on our platform that not everyone's actually always aware but I feel like as you are solving different data problems uh if you have not asked your account manager if you've not asked data bricks to take a look at it with you and ask for some help please do we've actually got new products new features new tools that are being created every single day and um it's our responsibility to come to you and actually make sure you're aware of those but Um, if you're not, please don't be shy.
Uh, we'd love to actually help you guys grow your business. Thank you to these three folks and the tool. Uh, have a great rest of your summit. Thank you. Thank you. Thank you.

Learn more about the Databricks Data and AI platform.