How BambooHR Scaled Analytics to 100K Users with Omni and Databricks

Summary

BambooHR solved a reporting problem that was both the top feature request and a top churn reason for its 30,000 business customers by partnering with Omni and using Databricks as the data foundation to launch a full embedded BI solution to 100,000 users in four months.
A semantic layer provides opinionated defaults while allowing deep per-customer customization, and the analytics platform is packaged across three product tiers to match different customer needs and price points.
Embedded analytics delivered a 75% retention boost for BambooHR's elite tier and a 31% customer satisfaction improvement, while AI integration through MCPs is enabling conversational analytics on the same data foundation.

How BambooHR Scaled Analytics to 100K Users with Omni and Databricks

BambooHR solved a critical retention problem by embedding powerful analytics directly into their HR platform. By partnering with Omni and leveraging Databricks as the data foundation, they launched a full BI solution to 100,000 users in just four months without diverting core engineering resources.
Discover how BambooHR used a semantic layer to provide opinionated defaults while enabling deep customization for each customer. Learn the key decisions around performance testing, packaging strategy, and AI integration through MCPs that turned analytics from a churn driver into a 75% retention boost for their elite tier. The session covers proof of concept, scale validation, and how embedded analytics became a strategic product lever.
🤝

Chapters

FAQs

Why did BambooHR decide to partner with Omni rather than build its own analytics?

BambooHR identified that reporting was both the top feature request and a top churn driver among its customers, and concluded that partnering with Omni would deliver value faster without diverting core engineering resources. This video describes six key criteria that led to selecting Omni, including performance at scale, semantic layer capabilities, and speed to market.

How did BambooHR validate that the Omni and Databricks solution could scale to 100,000 users?

Before committing to a general availability launch, BambooHR ran a proof of concept followed by scale load testing to confirm that the Omni and Databricks combination could handle the performance requirements across 100,000 users. This video describes the full timeline from POC to GA as four months.

What business impact did embedded analytics have at BambooHR?

BambooHR's embedded analytics drove a 75% retention boost for customers on its elite product tier and a 31% customer satisfaction improvement, demonstrating that reporting transformed from a churn driver into a retention asset. This video frames those outcomes as the business case for continuing to invest in the analytics platform.

How is BambooHR integrating AI into its analytics platform?

BambooHR is integrating AI through MCP endpoints that enable conversational analytics, allowing users to query their HR data in natural language on top of the Omni and Databricks data foundation. This video describes this as a key direction for the product, building on the semantic layer already in place.

Full transcript

[00:07] All right, what's up everybody? I think that's my cue to get started. Thank you all for coming. My name's Thomas. I'm a group product manager at BambooHR focus on our data and AI products. Um, these silent events are pretty funny. I feel like I'm just talking to myself up here. So hopefully y'all can hear me well.
[00:22] Um, but yeah, super excited to talk to you all today about how we partnered with Omni to build and launch an AI data platform to our customers at Bamboo. And um, take you through some of the learnings and things that we tackled here. Let's see if the clicker will work.
[00:39] There we go. Um, so really quickly, uh, I'm curious just raise your hands. Who all is familiar with BambooHR? Okay, a couple people. Is anyone a BambooHR customer? Does anyone use that at their company? Maybe? Okay. Um, yeah, a couple of you. So we are a, uh, HR platform that focuses
[00:57] on small businesses. And just a couple things about us. We serve about 30,000 companies um, through 190 countries. And we ran into this problem about a year and a half ago where reporting was not
[01:13] only a top feature request for our customers coming onto the platform. It was also a top churn reason for some of our customers. So our reporting platform um, just felt a little bit stale compared to some of the other areas of our product. So um, we jumped in. I have to use
[01:30] keyboard here. We jumped into that challenge and started to um, try to identify some of the gaps that we're having. The first one is people would start a report in our platform and then they would take it out into Excel and they would complete the rest of their
[01:46] reporting in a place where they were more familiar with the tools. So this made it really hard for us to do a couple of things. Uh, the lack of customization and flexibility in our product forced us to lose out on lots of insights on what types of things our customers wanted to accomplish. We didn't know what other
[02:03] types of data they were joining with our data. We didn't know, you know, what the end purpose of a report was. Is it compliance-related? Is it an executive dashboard? Um So, we we with those problems in mind, we started to think about, "Okay, do we try to build a solution to this, or is
[02:18] there a way we can partner to deliver value to our customers a little bit faster?" Um and that led us to our conversations with the Omni team. Um really quickly, just setting a little bit more context on on where we've gone with the Omni team. So, we have three
[02:34] product packages at BambooHR. We have our core package, our pro package, and our elite package. And one of the best things that this uh this partnership enabled us to do is really differentiate our reporting across different package tiers so that we could provide more value for customers. So, in our core
[02:50] package, you can see our very basic kind of table report. And then pro and elite is where the Omni partnership comes into play. Pro is a set of standard dashboards that they can access, and then elite is full BI flexibility embedded into our tool. So,
[03:06] let's talk a little bit about why we chose to build with Omni. Um first of all, we talked about people exporting reports into Excel and building on their own. The self-service of having a BI platform that's intelligent built into ours means
[03:22] that our users no longer have to go find the right answers to data by building their own spreadsheets and slicers and filters and all these things. They can just ask questions about their data, and they can get those answers really quickly and share those out to their company. Um AI alignment. We'll talk a little bit
[03:38] more about this in a in a moment, but we have a as I'm sure everyone here has a very ambitious AI strategy for Bamboo over the next year, and we needed a product that fit into that strategy really well. Next was speed to market. I I mentioned this already, but um we were in a little
[03:53] bit of a time crunch trying to fix reporting. We had lots of customers churning. We again, top feature request. And so we really were looking for a solution that we could get out quickly to customers and line it up with our packaging launch so that we could separate out those tiers and provide some kind of um strategic value across
[04:08] each of those. Then next one, and this is huge for us, customization and brand control. So we take Omni's business intelligence platform and we embed it in the BambooHR platform and we care a lot about the look and feel of our platform. Our customers are very particular. Um we
[04:25] don't typically do extended term contracts. We're typically month-to-month contracts. So we have to re-earn customers' business every month. So uh design experience really matters. With Omni, we were able to build this in a way that it looks and feels very native to everything else in our platform. So the embedded tool very
[04:41] flexible. Um and then these last two kind of go together, performance at scale and granular security. Um security matters a lot in general, but again, earning your customers' business every month, um you have to make sure that they really trust your platform. So we needed to make sure that we could apply
[04:57] our permission model through the Omni semantic layer and have it be performant enough that at launch we could go to potentially 100,000 users. We thought it was optimistic at the time and that's ultimately what ended up happening as we got it out to 100,000 users at the launch date. Um so we had to make sure this worked really well.
[05:15] Uh so obviously uh performance at scale and granular security, those are hard to identify and really understand at a glance, right? So when we started evaluating Omni, one of the biggest questions was like, can you guys actually do this? This is an embedded use case where we're getting this out to
[05:30] all of these customers. Um are you going to be able to meet our concurrency requests and requirements to get this out in a way that scales? And one of the best parts about this partnership is we essentially came and said, "Hey, we have some of these requirements, these these hard, you
[05:46] know, walls that we have to get over to be able to get this product out. And we're having a hard time figuring out how to test these in a way that makes sense so that we can actually understand if the scale is going to work. The Omni team jumped in with us and was able to actually go and start load testing against their own product in a
[06:02] way that we were able to very confidently say, "Okay, yes, this is the right product. This can perform at scale. We can do our we can map our permission model into Omni in a way that will grow with us."
[06:20] And with that, uh with the the uh scale and load testing that we were able to put in place, we then were able to take our data from Databricks through the Omni semantic layer and surface that in our product. And this is super important because this Omni semantic layer actually enables us to um create
[06:36] these default topics or models where customers can all get opinionated defaults in their account, but then they can go and customize from there. And again, permissions applied, it makes it really easy for us to be able to surface data insights in our product.
[06:56] So, let's talk a little bit more about what I just said and what that means. This core centralized model is again something where we at Bamboo we say, "Hey, there are probably things that you care about a lot as an HR business that you shouldn't have to think about building on your own." Things like turnover analysis, um
[07:12] compensation data, there's like benefits reports that you need to be able to show to your leadership team. And so, we're going to give you all of that packaged nicely in this core model that we've defined in Omni so that you can report on all that data ready to go. However, like I mentioned at the beginning, there are tons of cases where
[07:28] someone's like, "Hey, I want to measure like the the pounds of concrete poured per employee every month." And that is really important to our leadership team. We have to be able to show that in a report. Obviously not default HR data. It's a custom field, and the great part about the semantic layer and what we
[07:43] were able to build here is you can just map that to your data, and you can keep that in your platform, customize it as one individual customer. It doesn't have to impact anyone else's data model, and it's something that you can store and report on consistently.
[08:03] So, let's talk a little bit about our launch timeline. Um we obviously mentioned, you know, again, we were under a little bit of a time crunch. So, uh reporting's a big issue. We have this big packaging release coming up in a couple of months. So, we started evaluating a lot of BI partners and a lot of embedded BI partners.
[08:18] And ultimately, we landed on Omni and immediately started this proof of concept, which is where that load testing and architecture design came into place. For us, I don't think we uh we didn't communicate this with the Omni team cuz we wanted to keep it a little bit close,
[08:34] but this was like right away, the decision was clear that it was the right choice for us. It was a uh quick partnership to be able to uh again, do some of that load testing and understand what types of volume we would have for users at scale. From there,
[08:50] in month two, so again, two months after we said, "Hey, this is the path we're going to go down with the proof of concept." We were actually able to launch these Omni dashboards in our internal BambooHR account. So, we dog feed our own products. Um we again, use it for performance reviews, payroll,
[09:07] um benefits management internally at Bamboo. And we were able to ship this to our internal HR team to start getting some feedback on the look and feel of the product, the types of data they were reporting on, all of that information that they wanted in their account.
[09:26] Which led us to month three, feeling really confident that we could open up our beta program to a much broader base. So, three months into starting to build this, we were able to open up a fully embedded business intelligence platform within BambooHR to 30,000 users. Um this is
[09:41] awesome scale for a beta program. We were really excited about this and we started to get a lot of feedback. Um most of it related to just like look and feel of things. A lot of the data models were there, um but this helped us make some UX tweaks and start to again really refine this. So nobody using this feels
[09:58] like they're leaving the Bamboo platform. They feel like they're just getting added business intelligence capabilities directly within our tool already. So lots of great feedback and then in month four we were actually able to take this into GA and get this out to 100,000
[10:13] users. Again, this is a global product, so lots of people using this. Um and we've continued to get great feedback and I'll talk a little bit more about some of the results here. So results so far. Um again, four months from decision to GA for us, which was a
[10:29] huge add to our platform. Um we got it out to 100,000 users. What did it actually do for our satisfaction and retention? We were able to see a 31% satisfaction bump from our core and pro packages to the elite
[10:44] package with custom reporting and analytics. So a massive jump in terms of how people felt about our reporting product and our tool. The next one and this is uh probably my favorite stat, but customers on our elite package are now 75-ish
[10:59] percent uh less likely to churn than those on core and pro. And again, you could say there's like some conflating variables with larger companies typically being on the elite package, but with month-to-month contracts, I would argue that that's a a lower kind of conflating variable than you would normally see. This is
[11:16] definitely something where again, we're earning this business. Customers are much more satisfied with the reporting tools that we offer now. And then of course, this created new revenue streams for us. So um really thinking about how partnering for a BI tool is more of a strategy than a
[11:32] product add-on. Uh it's something where we don't have to be the ones that are continuously developing new BI capability. We get that through the partnership and can offer our customers a great product because of it.
[11:47] So, um I I have to share this quote. I love this quote. I'm just going to read it really quickly. This is from one of our customers who's jumping on this. But, um essentially he jumped in and said, "What I was doing before was extrapolating all of our employees and making a slicer in Excel." He Uh the customer's having to pull from all the employees and then
[12:03] filter out by 18 different slicers in there with what department they're in, what manager they report to, age, nationality, so on and so forth. Again, lots of standard HR reporting that they're taking outside of our platform. When we launched this this product, the response was, "This is way easier than
[12:19] what I'm doing. Way easier. I like easy. I'm lazy. That's some next level stuff that I wouldn't be capable of doing. Like, honestly, that's really cool." So, again, as a product person, things like this warm my heart. It's something where, um you know, we provide immediate value, but the other
[12:36] side of this coin is as we mentioned, we've now been able to get a lot more data sources into our product, and we've had our product teams' road maps start to change because of it. So, we see someone pulling in like hiring data for ATS, applicant tracking system, in our tool, and
[12:52] they're using it in a really unique way that's informed decisions to say, "Oh, actually, you know, maybe we should house that data in our platform locally so they don't have to import it from a different system." So, how can we start to think about changing some strategies and road maps based on what we've been able to deliver.
[13:11] So, again, awesome success getting that out to customers, launching that. Um I mentioned this at the beginning, our our AI alignment. Let's talk a little bit about what that means and how we're thinking about that. So, in Bamboo today, we have a chat tool where a an employee or an HR admin can
[13:27] go and ask a question about things like uh uh benefits policy, they can ask about like time off, all HR related data right now. We care a lot about our branding in that conversation. We care a lot about the voice, the tone it returns with, the way
[13:43] things look. And so what we were able to do with Omni, they didn't force us into using their chat experience. We actually were able to take their MCP, use an agent orchestrator to recognize things like is this an analytics question or is it a benefits question or an ETS question?
[13:59] And from there we can have customers ask questions in natural language and return analytical events in our conversational interface. So again, very smooth integrated experience for our customers. And this is actually launching today, I think, the 16th. Is it 16th 15th today?
[14:16] I don't know what day it is. But anyway, this is about to launch. It's going to be an awesome product. We're super excited. We've had a couple of betas going on with the conversational analytics piece and a lot of people are are, you know, some people really care about being able to build the report and use the BI tool. A lot of
[14:33] people just want quick answers to questions. So by enabling us with these MCPs and playing into our AI strategy, this feels a lot less like a bolt-on and a lot more like an integrated experience.
[14:48] Okay. So let's talk through, we've gone through a lot of this so far, but let's talk through a couple of learnings here. Um obviously we we have touched on each of these, but the first one is just that this semantic layer is so incredibly powerful for us to be able to scale
[15:05] without focusing on additional resources to build our own semantic layer. Um what I mean by that again, we have a core model that goes out to our customers that answers a lot of their HR questions, but they are able to self-serve and build into that model and
[15:20] get additional information from there. Um on top of that, for people at Bamboo actually contributing to this core model and thinking about what types of core reports people care about coming out of the box, uh that's also much easier for us to do. So, that's learning number one. Um
[15:35] that plays into learning number two. Those are kind of mapped together a little bit here, but uh the the custom fields that people are able to bring in have genuinely changed our product road map because we all of a sudden have a much broader lens into what people care about and what they're working on. So, just by adding this tool
[15:52] into our platform, we were able to get new insights into what people care about when it comes to HR, how they want to report on their data, where it needs to live, um you know, does it need to be real-time or can it be cached, all that information. And then the last one, embedded analytics is really a product strategy
[16:08] and not just a product feature. So, because of this tool, again, with our packaging launch, we were able to separate and offer real differentiation between our core, pro, and elite packages. And um start to get some information across what works and what doesn't in each of those. And again, we talk about things
[16:25] like retention and satisfaction in the elite package, um and it has made a huge difference for us. So, um these are kind of the three takeaway learnings we have coming out of this. Um it's been a a phenomenal product launch for us. It's enabled us to scale across a much broader uh industry and a
[16:41] much broader customer set. And again, it's helping us continue to learn on product teams as we keep building. So, um that is all I have today. We have uh maybe a couple minutes for questions, but again, thank you all for coming. Really appreciate you all coming out.

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.