Skip to main content

Sponsored by: Fivetran | Raw Data to Real-Time Insights: How Dropbox Revolutionized Data Ingestion

Summary

  • Data pipeline acceleration: Fivetran automated data ingestion enables Dropbox to reduce pipeline setup time from 8 weeks to 30 minutes.
  • Real-time reliability: Automated data movement from tools like Zendesk, Google Ads and MySQL provides real-time visibility into customer sentiment and ad performance.
  • Unified analytics: Consolidating fragmented data sources into Databricks reduces redundancy and creates a single source of truth for enterprise analytics.

Sponsored by: Fivetran | Raw Data to Real-Time Insights: How Dropbox Revolutionized Data Ingestion

Watch: Sponsored by: Fivetran | Raw Data to Real-Time Insights: How Dropbox Revolutionized Data Ingestion
Talk by Chris Neat, Engineering Manager, Dropbox (Sponsor Speaker); Kelly Kohlleffel, Sr Global Director, Partner Sales Eng, Fivetran (Sponsor Speaker)
Dropbox, a leading cloud storage platform, is on a mission to accelerate data insights to better understand customers’ needs and elevate the overall customer experience. By leveraging Fivetran’s data movement platform, Dropbox gained real-time visibility into customer sentiment, marketing ROI, and ad performance-empowering teams to optimize spend, improve operational efficiency, and deliver greater business outcomes. Join this video to learn how Dropbox: - Cut data pipeline time from 8 weeks to 30 minutes by automating ingestion and streamlining reporting workflows. - Enable real-time, reliable data movement across tools like Zendesk Chat, Google Ads, MySQL, and more — at global operations scale. - Unify fragmented data sources into the Databricks Data and AI platform to reduce redundancy, improve accessibility, and support scalable analytics.
Here’s more to explore:
Production ready data pipelines for analytics and AI: https://www.databricks.com/solutions/data-engineering
The Big Book of Data Engineering: https://www.databricks.com/resources/ebook/big-book-data-engineering-2nd-edition
See all the product announcements from Data + AI Summit: https://www.databricks.com/events/dataaisummit-2025-announcements

FAQs

How did Fivetran help Dropbox accelerate pipelines?

By automating data ingestion and streamlining reporting workflows, Fivetran reduced Dropbox's pipeline setup time from 8 weeks to 30 minutes. This enables faster time-to-insight and lets teams focus on analytics rather than manual data movement.

What data sources does Fivetran connect?

This video shows Fivetran moving data from tools like Zendesk Chat for customer support, Google Ads for performance marketing, and MySQL databases. Fivetran has 700+ pre-built connectors enabling reliable, real-time data movement from most enterprise applications.

What business outcomes did Dropbox achieve?

With real-time data visibility from Zendesk and Google Ads, Dropbox can now optimize marketing spend, understand customer sentiment and improve operational efficiency faster. Consolidating fragmented data sources into Databricks provides the unified foundation for these capabilities.

How does this support global operations at scale?

This video demonstrates how Fivetran and Databricks work together at Dropbox's massive scale with a data lake of 80-90 petabytes. Automated, reliable data movement enables consistent analytics across global operations and multiple teams.

Full transcript

Hey, I'm with Fiverr. with Dropbox. Um, really happy to have you here. It's Dropbox has an incredible story about moving very, very fast with Fiverr and data bricks to business outcomes. So, we're going to hear story real quick. We want to do some intro. So I look after partner sales engineering for Fiverr. Been here and I were talking been here for 3 years and uh my story is pretty simple why I'm here. I was data consulting for the six years before that and I recommended Fiverr probably 125 times to clients over that six-year period.
So this just became very easy for me. , you want to do a quick intro? Yeah, I am an engineering manager on the data platform at Dropbox. And so what that means is essentially the suite of tools Dropboxers use to solve their data problems. Been there about 3 years. Um our data lake 80 to 90 pabytes. Lots of different tools and different things there. Um and so yeah that's me. And I know you said you love the company too.
Oh my gosh. That's so important. Yeah. I was at Amazon and Apple before Dropbox and the Dropbox culture um is being like engineering le um like from the core and then people are just genuinely kind which is the thing that I appreciate the most. So that's what I love. So I've known Dropbox a long time. I was I don't I don't think I was an original user but uh I was looking with my wife over the weekend. I've got some really old files in Dropbox.
You guys have been around for a long time. A lot of people know Dropbox. We're going to get the story of Dropbox in a minute. Who knows Fiveret? Who's heard of Fiveret? All right. Not bad. Not bad. Okay. So, we're not brand new, but if you're not in this uh sort of data space where we operate, uh fully automated data pipelines and data movement, been around since 2012. I consider us late venturebacked. Um we have over 6,000 getting up close to 7,000 customers now all over the world.
Tremendous data bricks customers. been a data bricks partner for a long time. In fact, have to mention data bricks data integration partner of the year. Thank you data bricks for that. Very very nice. Um would love for you to come see us. That's just a little bit about Fiverr. Again, automated data movement 700 sources into the data bricks data intelligence platform and customers all over the place all industries all geos all verticals. a little bit about Dropbox.
Yeah, Dropbox, one of the premier brands in technology from way back in the day. I think a lot of people will be surprised at the scale. So, 700 million people use it. Millions and millions of paying customers, obviously, a a load of files outside of just our data lake. Um, you know, for us, we are a company that has and competes on the scale of, you know, some of the bigger brands you might see, but with a smaller engineering team, right? in like a a nice combination of a small company but with like global scale that you might find at a Google or Facebook or something like that.
So, and I know we have success up there. There's been a lot of success together with data bricks, but I'd like to have you talk a little bit about some of the challenges that you've you've faced and these have been happening over the last few years. You've addressed some of those. I'm sure you've got more to come, but what what kind of drove you from a challenge standpoint to say, "Hey, here's where Fiverr could maybe help us."
Yeah, I mean, all I think you got to do is walk around this center and see all the different tools that are getting introduced every day. So, you know, every single day or every quarter, we get requests to add new data sets into our data lake, a new tool that we need to integrate with. And keeping up with that pace of new tools is just cumbersome for a team that's constrained but also operating at scale like Dropbox.
And so we had to introduce something to keep up with our business. So I when I when I was working with a lot of clients on the consulting side, I I heard four things repeated over and over. And one of the things on this really stands out to me, I've got to operate fast because if I don't operate fast, I become irrelevant in my organization. I need simplicity. I need, you know, operational reliability and sustainability. this note on here.
I think about weeks in terms of Anybody else think of weeks in terms of sprints? I think it's my consulting like I see four sprints right there down to 30 minutes. That's incredible. Yeah. I think the big thing for us is the last thing we want to be is a blocker for our business to make decisions and things are operating so quickly. The tools that they're adding and the data sets they want to bring in, the last thing that we need to be is getting in their way when they're trying to make a decision.
And so what we've been able to do by introducing Fiverr is really refocus a lot of our engineering efforts on building the value per se from the data marts that we want to create. We were able to create a driver data mart that you know are leaders within the company are able to make decisions and instead of focusing on the 700 plus integrations that Fiverr automates right and so that's the key thing that we were able to do.
I got to mention too, not not that this should be a Fiverr commercial, but we also have a connector SDK. So, if you have source 701, you can build that very very easily with Fiverr. So, I want to I'm going to skip ahead to this next uh this next slide. And I'd love for you to discuss a little bit just anything that you can bring out, uh different groups, different areas, different use cases, different outcomes because to me, it's all about what happens downstream.
You got to have high quality trusted usable data flowing into data bricks data intelligence platform so that I can deliver all those datadriven marketing outcomes. You know um our data engineering team really moved to like a self source self-s serve addition model with fiverr. So if I'm a business person and I want to add a data source into the data lakeink, they were able to self-s serve and do that. And then our data engineering folks could then focus on the boxes that you see after, right?
And you see some of the things that they were able to focus on going and doing. But um really we we pivoted it to where our business folks could get data into the data lake without having to file a ticket, have an engineer go look at the data source and a lot of those sort of things. that created um knock-on knock-on problems in terms of governing the data once it gets into the data lake and things like that, but we were able to stop being the bottleneck for ingesting things into data bricks.
So, and I I like the fact too another thing I and correct me if I'm wrong, but you probably didn't have to do with fiverr and data bricks, no infrastructure like you're not dealing with infrastructure anymore and infrastructure when you are dealing with it can really slow you down, right? Yep. And you know, Fiver Train makes it easy to see what's failing, what's not, um, health checks and those sort of things. So, from that standpoint, um, we really don't have to think about it.
Um, we're coming back to that problem, you know, cuz as more folks add ungoverned assets into the data lake, you know, you sort of need to be able to monitor what's going on. But in terms of accelerating velocity to decisions, Fiverr was uh, very helpful for that problem. How has it been when I look at some of these datadriven marketing outcomes and use cases you guys have, you know, working with the downstream business owners? um if I'm a downstream business owner and I'm not getting to what I want with the data I feel I feel a lot of pressure and how how has that experience been for your downstream business owners working with the data engineering team working with the product team to say hey let's do this in a really short time frame with a combo of fiverr and data bricks and let's do it in a way that's giving you huge value yeah I think for us last year we kicked off we called it the driver model data mart but essentially We wanted to reinstment how we measure the e efficacy of our business, right?
And as you can imagine, Dropbox, we monitor shares and different engagement metrics with the platform. We essentially were able to do that with Fiverr and reinstrument that data mart within 3 to four months. So yeah, really nice. I gota ask too uh as it relates to the data bricks platform I mean there's so much I know I was in fact on I had a LinkedIn post this morning and I asked who went to maybe one of the original Spark Summits uh way back when you know when I think now to where the data bricks data intelligence platform is you've got all of these different capabilities and workloads that are there but you need high quality trusted usable data flowing in to facilitate each and every one of those workloads. anything stand out to you right now in terms of that kind of harmony between fivetra and data bricks that you're seeing?
I just think it's ease of use, right, and point and click to enable a new connection um is really the thing that drives a ton of value for Fiverr. It really does reduce the time to ingest data um you know and there's trade-offs to that you'll have to manage as a platform, but we wanted to get out of being the bottleneck for getting something into the data warehouse. And so yeah, I that's what I'd say is like the key driver.
Yeah, very very good. Well, let's transition over here and uh again as it relates to overall business outcomes, I mean we talked about a couple of different areas. I don't know if there's anything that you know we talk a lot about transformative uh activities today data transformations business transformations anything there especially as it relates to your your marketing team and what your team has been able to add that uh stands out to you yeah I the one of the key use cases that I'll call out there's a bunch uh we've enabled a ton of connections and ingestion here uh one of the key use cases is return on ad spend so we were able to ingest Marquetto data decorate at um some of our data engineering teams focused on that decoration and then generate a data mart that allowed us to lift our oas um as a marketing team.
So you know um and we're able to tie that back to revenue and time to market things that Fiverr helped enable. I had uh I had not called on a lot of marketing organizations probably I don't know five six seven years ago and I remember sitting down I won't mention the name of the organiza great organization by the way healthcare organization but I sat down with the CMO and the CMO said hey I need you to help me spend more money and uh I said wait I've never heard this thing before what what what kind of a group is this you know it's kind of to your point additional budget I want to deploy this in the right way without the right data products to do that it becomes really really difficult.
Right. Right. Yeah. And then I think what we experienced at Dropbox is that like as I mentioned in the beginning the proliferation of marketing tools Marquettle is a big brand that everybody sort of knows but with all these different AI tools coming I think we're going to see an acceleration of the amount of integrations that we need to perform to be able to activate the data for the business. Looking forward to the announcements tomorrow too.
I think we've got some good stuff on the horizon, right? Um, speaking of horizon, what about looking forward? We got a little bit of 2025 left. We've got uh, you know, who knows what's look at look at what's happened in two and a half years with Gen AI, right? Um, what do you have on the horizon for you and your team and then ultimately what you're being asked to deliver over the next say 12 to 18 months? Yeah, I think the big thing and it's going to be internal for us is really a governance platform on top of all of this data.
We just have data from a any number of sources pouring into our data lake. It's growing at a reasonable amount and we really need to put a coherence management layer on top of that allows the business to activate highquality metrics. So that's kind of the focus over the next few years for Dropbox is really how do I know our rorowaz number is the source of truth, right? Um those are the type of things that we're going to focus on in the next few years and making sure that each metric that our business depends on is high quality.
We sort of we've gotten really a lot better at solving the ingestion problem. Now we need to solve it's in the data lakeink. How do we make sure it's high quality? How do we make sure it's governed? How do we make sure it's compliant? you know, when you unlock a lot of the integrations with Fiverr, how do you know someone's not ingesting uh something that could be HIPPA HIPPA relevant or things like that for compliance?
We're going to need to introduce governance that allows us to monitor that in a coherent way. Really good. Any uh any particular approaches or thoughts around how you might want to try to tackle that you can share or um we're we're in the early phases of that sort of thing. I think for us, we are we're a code first company. So, we're going to codify a lot of our key metrics and then create interfaces for the business to be able to search and discover our metrics.
Um, and you know what we've experienced with Fiverr is, you know, someone can load a G sheet into your data lake and then a really important metric can now be driven by this really not engineering CI/CD process, right? Right. And so what we want to build is something that can discover those sort of things and introduce engineering quality on top of it, right? And so um we're exploring some of the tools that you'll see here to do those sort of things.
But um yeah, more to come on that. So I I've got a one of my sons is a data engineer. He's always talking to me about automation and ensuring that and it's a really lean team. I'm hearing that same thing from you like if we can automate it, we will automate it as a data team. Yeah. I think you know resource constraints and the ability to operate at scale with a smaller team you know I think is something that allows us to innovate but it's been the thing that's allowed us to introduce third party tools right we've gotten to the point and big data has gotten to the point where the amount of third party tools that are actually effective at scale um has has actually reached a point where they're they can solve the problem that Dropbox had you know 10 to 12 years ago all of Drop Dropbox's technology to solve these problems was internally built.
Homebuilt code, homeuilt orchestration systems, all of these sort of things. And we're sort of going through this genesis where now that the tooling in the market is catching up, we're adopting a lot of those things and you know implementing it at scale. And the driver for that of making this conversion over y'all have been going through it over the last couple of years. Main dimensions being the time aspect, the cost aspect, simplicity.
What were the what stood out to you? I think time to market, you know, which we've referenced in this presentation multiple times. And then I think it's just the quality of the tools to be able to handle scale, right? Like when you're ingesting and operating on pabytes of data, lots of queries per second, a lot of the tools back in the early days couldn't handle Dropbox scale, so we had to do it ourselves, right? And so now a lot of these vendors are catching up to be able to solve our level of big data problem.
Cool. Anything else that uh you wanted to share? I guess we have a little bit of time, just a couple of minutes for questions, but anything you didn't talk about that you'd like to share with the with the group? Um, no, I don't think there's anything that jumps to mind. You know, I think bringing data together is one of the like core problems of the industry and so I think Fiverr is in like a really nice position to assist with that.
I think Dropbox also sees differentiation for different vendors and implementing our own integrations, right? And so we balance those things depending on how we feel about the value of the data set as it relates to our products and those sort of things. Fiverr really unlocks um those situations where we have time to market is imperative. there's a business team that needs something and we allow them to self-s serve to unlock themselves, right?
And then we we clean we make sure that we govern it after the fact. Self- service is elusive, too. That's a that's a tough one. I've been asked about many many uh years and I think we are getting really really close. Um we have do have like couple of minutes for Q&A.

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.