139 – Monetizing SAAS Analytics and The Challenges of Designing a Successful Embedded BI Product (Promoted Episode)

Experiencing Data with Brian O'Neill (Designing for Analytics)
Experiencing Data with Brian T. O'Neill
139 - Monetizing SAAS Analytics and The Challenges of Designing a Successful Embedded BI Product (Promoted Episode)
Loading
/

Please note, this is a special Promoted Episode of the podcast.

This week on Experiencing Data, something new as promised at the beginning of the year. Today, I’m exploring the world of embedded analytics with Zalak Trivedi from Sigma Computing—and this is also the first approved Promoted Episode on the podcast. In today’s episode, Zalak shares his journey as the product lead for Sigma’s embedded analytics and reporting solution which seeks to accelerate and simplify the deployment of decision support dashboards to their SAAS companies’ customers. Right there, we have the first challenge that Zalak was willing to dig into with me: designing a platform UX when we have multiple stakeholder and user types. In Sigma’s case, this means Sigma’s buyers, the developers that work at these SAAS companies to integrate Sigma into their products, and then the actual customers of these SAAS companies who will be the final end users of the resulting dashboards.  also discuss the challenges of creating products that serve both beginners and experts and how AI is being used in the BI industry.

Highlights/ Skip to:

  • I introduce Zalak Trivedi from Sigma Computing onto the show (03:15)
  • Zalak shares his journey leading the vision for embedded analytics at Sigma and explains what Sigma looks like when implemented into a customer’s SAAS product . (03:54)
  • Zalak and I discuss the challenge of integrating Sigma's analytics into various companies' software, since they need to account for a variety of stakeholders. (09:53)
  • We explore Sigma's team approach to user experience with product management, design, and technical writing (15:14)
  • Zalak reveals how Sigma leverages telemetry to understand and improve user interactions with their products (19:54)
  • Zalak outlines why Sigma is a faster and more supportive alternative to building your own analytics (27:21)
  • We cover data monetization, specifically looking at how SAAS companies can monetize analytics and insights (32:05)
  • Zalak highlights how Sigma is integratingAI into their BI solution (36:15)
  • Zalak share his customers' current pain points and interests (40:25)
  • We wrap up with final thoughts and ways to connect with Zalak and learn more about Sigma (49:41)

Quotes from Today’s Episode

  • "Something I’m really excited about personally that we are working on is [moving] beyond analytics to help customers build entire data applications within Sigma. This is something we are really excited about as a company, and marching towards [achieving] this year." - Zalak Trivedi (04:04)
  • “The whole point of an embedded analytics application is that it should look and feel exactly like the application it’s embedded in, and the workflow should be seamless.” - Zalak Trivedi (09:29)
  • “We [at Sigma] had to switch the way that we were thinking about personas. It was not just about the analysts or the data teams; it was more about how do we give the right tools to the [SAAS] product managers and developers to embed Sigma into their product.” - Zalak Trivedi (11:30)
  • “You can’t not have a design, and you can’t not have a user experience. There’s always an experience with every tool, solution, product that we use, whether it emerged organically as a byproduct, or it was intentionally created through knowledge data... it was intentional” - Brian O’Neill (14:52)
  • “If we find that [in] certain user experiences,people are tripping up, and they’re not able to complete an entire workflow, we flag that, and then we work with the product managers, or [with] our customers essentially, and figure out how we can actually simplify these experiences.” - Zalak Trivedi (20:54)
  • “We were able to convince many small to medium businesses and startups to sign up with Sigma. The success they experienced after embedding Sigma was tremendous. Many of our customers managed to monetize their existing data within weeks, or at most, a couple of months, with lean development teams of two to three developers and a few business-side personnel, generating seven-figure income streams from that.” - Zalak Trivedi (32:05)
  • “At Sigma, our stance is, let’s not just add AI for the sake of adding AI. Let’s really identify [where] in the entire user journey does the intelligence really lie, and where are the different friction points, and let’s enhance those experiences.” - Zalak Trivedi (37:38)
  • “Every time [we at Sigma Computing] think about a new feature or functionality, we have to ensure it works for both the first-degree persona and the second-degree persona, and consider how it will be viewed by these different personas, because that is not the primary persona for which the foundation of the product was built." - Zalak Trivedi (48:08)

Links

Sigma Computing

Email: zalak@sigmacomputing.com

Zalak's LinkedIn

Sigma Computing Embedded

About Promoted Episodes on Experiencing Data

Transcript

Brian: Hey listeners. Before we get started today, just a heads-up that today’s episode with Zalak Trivedi from Sigma Computing is the first promoted episode on Experiencing Data, something new that I’m trying in 2024. So, what’s a promoted episode of Experiencing Data? Well, before you run for the hills and think that you’re going to hear some sort of ad or sales pitch for the next 40 minutes or so, let me give you some backstory. Every week, I get pitches in my inbox from PR firms, and companies, and book authors, typically in software, probably most of the time with some kind of data bend, or analytics and machine learning and AI, and frankly, a lot of those pitches are completely off the mark in terms of being content that I think will be helpful to you.

Sometimes they’re not even in those spaces, and they’re just completely outside of what I would bring to you. However, a small slice of these pitches are in the ballpark for listeners like you, and if I think there’s an angle that you’d find interesting, and if these companies pass my screening process—which basically lets me ask pretty much whatever I want, and control the interview—then my hope is that you’ll find these promoted episodes super valuable while keeping Experiencing Data ad free and at the high level of production quality that I’ve strived to bring since episode one.

So anyhow, when Sigma Computing first reached out about their embedded analytics platform for SaaS products, they were very flexible and open to me, asking if we could actually do a discussion about how they do product and design internally at Sigma, given the complex ecosystem of users and stakeholders involved. And I have to say, Zalak was super honest about the challenges they face, and just like my regular guests, he did not get any of my questions in advance. So, what you’re going to hear is just a very candid interview that’s just like the ones I strive for with my regular guests that I invite on the show. So, if you are curious to learn more about promoted episodes on Experiencing Data, visit designingforanalytics.com/promoted, and you can also email me from that page with your feedback about this episode.

This is the first of these promoted episodes, and I’d really love to hear what you think. If for any reason, you all feel like the show is not going in a good direction and these don’t feel good, I want to know because I’m going to shut it down. So, this is very much an experiment for me, and it’s easy to turn off if it doesn’t work, but I’m already really excited about sharing this episode just like my other ones. So, please send in your feedback. So, now without further ado, let’s go behind the scenes with Zalak Trivedi, the product lead for the embedded analytics solution at Sigma Computing.

Zalak: Hi, this is Zalak Trivedi, Product Manager at Sigma Computing, and you’re listening to Experiencing Data with Brian T. O’Neill.

Announcer: You’re now Experiencing Data with Brian O’Neill. Experiencing Data explores how product managers, analytics leaders, data scientists, and executives are looking at design and user experience as a way to make their custom enterprise data products and analytics applications more useful, usable, and valuable. And now here’s your host, the founder and principal of designing for analytics, Brian O’Neill.

Brian: Welcome back to Experiencing Data. This is Brian T. O’Neill. Today I have Zalak Trivedi on the line from Sigma Computing. How’s it going?

Zalak: It’s going great, Brian. How about you?

Brian: I’m doing really well, and I’m really looking forward to digging into a little bit of Sigma tools. You have, some might call it a data product; some might not. We can even get into it if you want. Correct me if I’m wrong, you are an embedded BI solution for, like, enterprise SaaS tools. Is that correct?

Zalak: We are both. We are a BI product, but we also allow embedding our BI product into other SaaS applications.

Brian: Got it. Got it. So, tell me a little bit about your role there. What are you up to at Sigma? What’s your purview?

Zalak: Yeah, so I joined Sigma a couple of years ago, and I’m now currently leading the product direction and the vision for the embedded analytics portion of our product. Something I’m really excited about personally that we are working on is, embedded Sigma can really go beyond the analytics to really help customers build entire data applications within Sigma. This is something we are really excited about as a company, and kind of marching towards this year.

Brian: And again, now knowing what you know about the audience listening to this show, who are mostly data product leaders and people that have been doing this kind of work for a while, I know one member’s already, probably—her wheels are probably spinning, and she’s like, “Oh, you can create these, like, embedded applications.” So, what does that mean exactly? Because a lot of times when we think of something like an embedded analytics thing, it’s like, go to this tab, and then you’re in the reporting world, but you have to stay on that island. Otherwise, you’d take a boat, you leave the island, and you’re completely out of there, you’re back into the application. So, is that kind of how Sigma is structured, or are there ways to bring insights out of the quote, “Reporting area,” and into the user experience at the right time that you might need it? Or tell me a little bit about that.

Zalak: So, I think it’s a little bit of both the world today that we are in at Sigma, I think it’s a lot about the first part you talked about where you click on this tab, and you are into this world of analytics. But where we are moving towards is the second piece where you can actually extract this information, and interact with the application where Sigma is embedded into. So, let me kind of step back and give you, like, an overview of what that looks like. Does that sound good?

Brian: Sure, sure.

Zalak: All right. So, let’s say you’re a product manager, right, and you have created this amazing SaaS product that provides a lot of value to your customers. But as your product matures, your customers are invariably going to ask for analytics and insights of the work that they are doing in your product. This is the same with if you are building internal tools, and if you have stakeholders internally, where you will be faced with the same question. So again, as a product manager, you will have to choose to either invest your own engineering resources to build these analytics solutions, or you can partner with some really powerful business intelligence tools like Sigma to provide these analytical capabilities in your product.

That’s the world of today where we are providing these capabilities into your product. But the tools and technology that we build that allows you to take these insights beyond this, right? So, a lot of times when someone looks at a dashboard, they get an insight, and kind of, the insight dies there, right? At Sigma, we are thinking about, okay, you’ve looked at this insight. So what, and what’s next? Right? So, and the so what and what’s next piece, most of the time is not within that analytics framework that you’ve embedded.

Which comes to the next iteration of things that we are working at Sigma which is, how do you go beyond this iframe that you’ve embedded? How do you interact with the host application? So, we’ve created lots of tools and technologies around actions and workflow frameworks that, let’s say, if you have identified, and if you analyze and figure out, this is the top 50 customers that you want to do, like, put them in a marketing campaign, you can immediately send this information out to, maybe, some other third-party tool that is handling your marketing campaigns, from within the Sigma’s embedded analytics application. Let’s say if you want to create this entire application which is managing, let’s say, your HR, which is things around managing performance reviews, and everything that you’ve done in this internal tool within Sigma, and if I want to extract that out to, let’s say, connect to Greenhouse or something else, you are potentially able to do that in the world of the future that we are building towards.

Brian: Got it. And that interaction is kicked off within Sigma, or it was kicked off in the third-party application that’s calling or requesting information. Or which way does it go?

Zalak: It is initiated from within Sigma because that’s where you’re looking at the insight. Because you’ve done all of your analysis, you created these charts, and again, we want to make sure that you have this entire closed-loop execution. You’re not really stopping at looking at this chart or an insight; you are able to immediately take an action on something that you see. So, it initiates from within Sigma, but it can go anywhere, right? It can go into your application because again, there’s a lot of intelligence in your application as well, and if you want to create a workflow which updates something in your application or triggers a workflow, you are potentially able to do that. So really, if something happens in Sigma, you are able to take an action anywhere in the world. Like, that’s the future we are going towards.

Brian: Got it. So, is this, like, a… you have a generic workflow builder that can interface with third-party applications? So, you would build, like, you know, if this analytic goes over a hundred or whatever, then do X, which has an API connection to some third-party application. And, like, create a marketing segment called ‘Top Users’ or something like—

Zalak: Exactly.

Brian: That, in Mailchimp or whatever, AWeber or something like that. So, it’ll create a segment of those customers and push that data over there. Something along those lines? Maybe not exactly that use case, but is that the idea?

Zalak: That is the idea, right. And so, there will be, like, a generic framework with APIs, but we will always have point-to-point integrations with certain tools that we want to partner with in the future. So, if we identify a text tag that a group of our customers are consistently using, we would potentially create [direct 00:09:23] integrations from our product, so people can configure it, and again, it should look seamless, right? The whole point of an embedded analytics application is that it should look and feel exactly like the application it’s embedded in, and the workflow should be seamless. So, if we are able to provide all of that capabilities through our embeds to our customers, it just really enhances the customer experience.

Brian: Yeah, this is, like, a really hard design challenge to me because, first of all—and maybe you can kind of break down the personas in your world because you have a buyer of Sigma, which is x person at SaaS company; I don’t know who that is. Then you have the developer or a product manager at the host company SaaS company, who has to—

Zalak: Yep.

Brian: Build out use cases specific to their customers. So, it’s really your customers’ customers who are the ones that are using all of this stuff at the end of the day. So, there’s an abstraction layer there, not just code-wise, but from a design standpoint, you have this, like, multi-tier experience, and I’m guessing you’re pretty far away, if not completely blocked from talking to actual end-users of your service. Is that right? That’s, like, a really hard design thing. Maybe you can unpack how do you approach that? Or are you only focused on kind of the tier one, like, your paying customers and their developers? Or how do you think about that?

Zalak: You are spot-on right. This is one of the biggest challenges that we faced when we were initially scoping out embedded analytics as a product. Because historically, Sigma had been a business intelligence tool for the data analysts. It was a completely different persona. So, when we actually went to market with embedded analytics as a product, we were again, you’re right, we were selling to the product managers who wanted to embed analytics into the product that they are creating.

We were also interfacing with engineering teams because they are the ones—like, the developers are the ones who are embedding it into the product; product managers are defining it. So, we had to—at a [first 00:11:29] level, we had to switch the way that we were thinking about personas. It was not just about the analysts or the data teams; it was more about how do we give the right tools to the product managers and developers to embed Sigma into their product. So, that was step one. But as people started embedding Sigma into the product, again, now we are exposing our product to hundreds of thousands of users, and these users can range from store managers, you know, delivery drivers, technicians, to very tech-savvy financial analysts to CEOs of different companies.

So, that was another challenge where the second-order persona was really not consistent because it changed depending on where the product was actually embedded. So, after a little bit of, you know, hidden trial, in the initial days, where we were working with a few select customers that we were going live with, we created a model of opt-in complexity. So, you start off with a very basic view-only dashboard where you don’t give a lot of bells and whistles to your end-users, but then, as you feel that your end-user is technically savvy and advanced enough to actually explore the data that they have been presented with, you can then add in your complexity. So, the product managers that we worked with really liked that because it gave them two main things that they wanted: one was they can incrementally add features and functionality to their product through Sigma because now you start off with a pre-canned dashboard as your first release, but then you can give them a little bit of exploration and bookmarking capabilities, and then you go into, oh, you can create your own dashboard, and so on; and the second thing is they can actually monetize this, right? So, they can almost say, in my basic plan, I’m just going to give you a pre-canned dashboard. If you want to explore this data I’m presenting to you, you can pay us more. You can go to the essentials or the premier license. And then the pro license, you can create your own dashboard.

So, we kind of worked with the product managers to figure out what was the exact mix, kind of, fell into these three buckets, and then created the technology in a way that the simplest version of an embedded Sigma was a pre-canned dashboard, and then our customers had the control, the complete control, over how they want to curate that experience for their end-users. In the initial days, I actually joined—a lot of our customers called when they were having it with their clients, they had—our customers had customer advisory boards, and they were taking feedback on their own product because they were also—they were, like, startups or mid-market companies who were going live with new products in the data space. So, I joined them and tried to figure out, okay, what is something that is complicated for the end-user versus what is something that people just get out-of-the-box? And then we kind of built in these tiers and worked with a few other customers to figure it out and iron out the plan. But yeah, it was… it was a challenge early on to figure out what do we launch. Because Sigma as a BI product has a lot of features and functionality for the analysts of the world, and when you’re embedding it, you’re not necessarily always exposing this product to the analyst. The persona varies widely.

Brian: Maybe you could even just step back a little bit and tell me about your team structure. So, as a designer, I don’t know if you’ve listened to the show before, but I kind of talk about this idea that you can’t not have a design, and you can’t not have a user experience. There’s always an experience with every tool, solution, product that we use, whether it emerged organically as a byproduct, or it was intentionally created through knowledge data—which might be qualitative or quantitative—but it was intentional. I’m wondering, who does that work, like, at Sigma? Do you have a design team that works with the product team? Is that de facto, like, you as a product lead? Who’s thinking about all this stuff, especially when there’s multiple tiers of users, and there’s a developer user experience, there’s your host application product manager’s experience, like, what are the features, functionality they need? Like, what’s value to your product manager customer look like? And then they’re thinking about their own customer. Just so much here. So, who does that work—

Zalak: Yeah.

Brian: —who’s thinking about that work? Who is responsible?

Zalak: That’s a great question. So, the way that Sigma’s engineering pods are created is that each engineering pod, essentially—or it could be a group of engineering pods—they have a product manager and a designer and a tech writer associated with them. So, I work very closely with the designer who is on my embedding pod, and we together—or my role is to figure out what are the experiences we want to curate at different tiers for the product manager that we are selling to, and my designer, her name is Sam Watts, she comes in and works with me, and we figure out what we want to present to the end-users when they embed a dashboard when it’s just a view-only mode. What do they get when they want to explore? What do they get when they actually want to have some administrative controls?

So, those experience, we work very closely on coming up with those and we iterate really quickly based on the product and the customer feedback that we receive on that. Because early, I would say mid to late-2022, we were in the process of figuring out what that correct mix is, so we iterated really fast over that six months. And then early in 2023 is when we kind of started getting really good traction among startups, mid-market, and now even enterprise.

Brian: Is there a method by which you measure whether or not the experience for the target user has been met in the way that you had envisioned it? Because I can see issues here where Financial Services Product Manager A thinks that everybody needs to drill down into Stat C, and your Delivery Manager Product Manager from Company Zeta could care less. Like, his drivers are never going to look at that stuff. They don’t care about how each ride that they deliver, like, each delivery broke down. No, they don’t care. They just want to know, what was my average route delivery time, and my whatever time, and did I get my bonus? Or I don’t—I’m just riffing here.

But how are you measuring that experience? Who gets to have the loudest say? And there’s the UX side, and there’s the business side, which is sales, but that’s not the same thing as delivering a good user experience, right, because the buyer—

Zalak: That is correct.

Brian: May not be the person that’s going to live with us all day. So, who decides that? Like, how do you decide that?

Zalak: Our sales and solution engineering team do have a lot of say in what we build and design because our solution engineering team specifically is, these people are coming as industry experts into Sigma. They have been at other companies who’ve been doing business intelligence for so long. So, I do value their opinion really high. But the designers and the product managers are the ones who actually have the loudest voice and decide what actually makes it into the product. To answer your question about how does the product manager at financial analysis versus product manager at Company Zeta create the experience for the end-users, we have account types, which has basically a roles and permissions-based model, and we have gated several things which we feel could add complexity to the end-user experience behind features permissions.

So, you go in with a very default account type, which gets you n number of features and functionality, but you as a product manager who’s, like, the admin of Sigma which is being embedded into your product, you can essentially, piece by piece, create an account type or an essentially an experience for your end-users which will be embedded. So, what me and Sam, we decide is, what are the different breakdowns or different points or different bits of permissions that we want to split out which creates the right experience for the financial analyst persona as well as the [unintelligible 00:19:31] technician persona. What we provide are the tools for the product managers for being able to create these experience, rather than dictate them. And as I said, the first six months was all about figuring it out, what are those permission bits? What are those experience we want to get behind something that can then unlock some complex user workflows for your end-users?

Brian: And do you have a way of measuring the usability or the utility of these solutions with, I guess it would be your paying customer PM, or the builder, the engineer on their developer integration person. I don’t know. I assume you don’t test with your customers’ customer. You probably don’t get a chance to do that. Or maybe you do. But that’s kind of the thing I’m interested in is how do you validate the designs were, quote, “Good,” the experience the user experience was good or not?

Zalak: Definitely. So, we capture a lot of telemetry. We obviously remove any PII information, but we have all the information about how our customers’ clients are using Sigma embeds. So, we have telemetry around what buttons they are clicking, what charts and graphs that they are spending most time on, what pages that they are spending most time on. And we collect that information, we provide that to our customers as well.

So, it’s a mix of both, right, where if we find that certain user experiences where people are tripping up, and they’re not able to complete the entire workflow, we flag that, and then we work with the product managers, or our customers essentially, and figure out how we can actually simplify these experiences. One of the examples I’ll give you, we had a menu bar, which showed up on the left, bottom left corner, which was essentially menus of, like, my documents, recents, et cetera. The button was just not discoverable. Because in our user studies internally that we had done, it made sense, but when you embed that into a different product, that button doesn’t make sense in the context of the application. So, things like that, we get feedback immediately from our customers, and we have a customer advisory board that we’ve now created who we always add into private beta that give us a little bit of feedback around anything we’re launching before we open it up to the rest of our customers.

Similarly, like, we just recently launched a functionality called exploration mode on embed. This existed internally on Sigma for quite some time, but we enabled it for our customers’ clients, and we immediately got feedback on how some of the things are really amazing, and some of the things would not just work in an embed context. So, we had to make sure that the things that we assumed that would work for our internal, like, data analyst persona, if you just enable it as-is on an embed, they would not be the right experience that the end-users expecting. So, just working—I think there’s a lot of iteration, just careful due diligence that we have to do every time we launch a new feature, in terms of design, in terms of security, in terms of the placement of different components on the iframe that we will embed, and what we allow to embed, and then we actually release things.

Brian: I like to theorize that, in general, product managers should be pretty good at explaining a problem space instead of a solution space just as a core skill, so I’m kind of wondering, when product managers are your customer [laugh] do they go back into customer mode, and you get a lot of, like, “I want a filter icon next to this field so that I can do X, Y, and Z,” and you have no idea what the context of uses, but they’re basically delivering feature requests to you in the form of solutions. Do they come in that way, or do you tend to get—

Zalak: A hundred percent.

Brian: Oh, so they just go right back into customer mode, don’t they[laugh]?

Zalak: Yes. It’s very solution-oriented. “Because I want to do this in my product, I want you to, Zalak, Product Manager at Sigma, to do this.” And the way that we approach these things is I wear my product manager hat, and I’m like, “But why? What’s the use case? What are you really trying to do? What are the options you tried out?” And we figure out, like, the real root cause of why they are asking for something. And if there are other workarounds, or again, [unintelligible 00:23:54] feature permissions that we can enable for the user to be able to do that, we take that approach.

We also see a lot of resistance from designers at our customers. Because, again, the way that Sigma’s layout is being rendered and embedded into their application, designers have a lot of opinions about it because it’s their product; they pride themselves on having the best experience, and having this other product being embedded into their product sometimes doesn’t sit well. So, we have to work really well with the designers as well to make sure that we provide all the right controls on the colors, [themes 00:24:31], gradients, essentially white-labeling experience so that they can make it look and feel exactly like that application that they’re embedding in. So, we do [unintelligible 00:24:40] and do a lot of, like, solutioning by our customers [laugh], but we always have to take a step back and say, “Okay, but what do you really want?”

Brian: Yeah, yeah. So, for all you data product people out there that are listening to the show, you’re just as bad of a customer as all the people that you complain about that give you solutions in search of a problem. You’re just as bad. I think we all probably do that, unless you, you know—like, how much detail do I really need to give, and you know [laugh]. But I can understand.

Like, they’re up against the gun, they probably want to, like, ship some feature out, and, like, they don’t want to take the time to give all that context, and I could understand that. I mean, I could even see, as a designer in that environment—I mean, you were talking about surface-level customization, colors, fonts, all this kind of stuff. The thing that would frustrate me in the past with these kinds of situations is, like, for example, you can’t carry context. If it is an iframe, you can’t carry page context.

And like, I was looking at a primary object which has an ID in the main application, and now I need some stats on it that come out of this third-party tool, and I want to link directly with that context. Let’s take something like a stock symbol for an equity position. So, Apple stock, or whatever it is. Well, I’m looking at current pricing chart for Apple. Now, I’m looking at the News tab.

Now, I want to get, like, how many users in the system are tracking Apple in their dashboards, which is a Sigma dashboard, but I have to go to Sigma. I lose Apple context. I then have to run a search for the Apple symbol again because I lost the context of Apple. Things like that would be, like, challenging, and so I imagine you probably end up getting a request for, like, “I would like a query string parameter called ‘stock symbol’ that I can carry into the”—[laugh].

Zalak: Yep. Yeah, you’re spot on, Brian. I think this came up a lot earlier, where we created a list of JavaScript events that we expose out to our customers’ host application that they can then consume, and then they can send it back through just like, you know, POST messages. So, we’re trying to create this interactive two-way stream between the host application and Sigma, so you, with the right skill sets, the developers are able to, like, figure out how to not lose context with the current set of tools that we have provided to our customers.

Just to kind of add to one of the previous points we were discussing… I know a lot of product managers would come in with a solution to some of these problems, but the good thing about really good product managers that I’ve worked with at our customers is, as soon as you ask them these questions, they realize what they’re doing. So, they are now almost your ally in trying to figure out what’s the best solution for Sigma as a whole, rather than trying to solve their own individual problem. And those people are the ones I have selected for our customer advisory board because now I can trust them to think objectively about Sigma as a product, rather than always thinking about that one… removing this one thing from the UI.

Brian: Yeah, that’s a nice little side benefit, if you can [laugh] figure out the ones that have flipped the switch, or you can see where they’re thinking that way. That’s a great—

Zalak: Yeah, I just ask the right four or five questions, and then they get into the mindset of, like, “Oh, yeah, I understand what I’m doing. Let’s think about it from this other way.” And some of the really good product managers have helped me write PRDs, as well. So, I mean, it has been a great benefit to me as well, with collaborating with some of these PMs.

Brian: Let’s step back to, like, the buying process, the marketing funnel. What triggers someone to come in? Is this like an, I’m a product manager, I’ve been told we can further monetize the product, and we think data and insights is the way to do that, so now it’s a buy/build decision, and I’m starting out with a greenfield environment, like we have nothing? Or are they starting out with a resident solution, and you’re trying to come in and say, “Well, we have a better version of what you’re doing today, your hacked together solution,” or I don’t know, there’s Tableau or some other tool that’s being used, and it’s, kind of like, a messy experience.

Like, where do you come in? Is it mostly greenfield, or you will kind of replacing a resident solution? I would think a lot of places have this, I guess already, but I’m sure there’s always new, you know, there’s always new SaaS companies popping up. But is this not a—

Zalak: Yeah—

Brian: —resident, or almost a given that most enterprise SaaS companies would already have some kind of insight solution in place?

Zalak: You’re right. I think enterprise SaaS companies, yes, but as you said, there are a lot of startups and mid-market companies that are still figuring out what they want to build, and they are trying to get to market really fast. In the early phase of embedded analytics launch, we really targeted this group, where it was greenfield, they did not have any other existing product that they have embedding, where we essentially go with the pitch around the benefits that they have, right? A lot of these times, as I said, you have to make the decision where you want to either spend your engineering resources away from your core competency versus, you know, spend some money, partner with someone like Sigma, and get every benefit that Sigma has, the power of Sigma into your product.

So, the way that we created this message was around, like, five different things where we would talk about speed, right? If you are going to build your own analytics engine, it will take forever, and time is money, really, in the startup space or the mid-market. And over time we, with our case studies and with our different blogs and stuff that we’ve written, we can prove them that a prebuilt BI product embedded into your product would actually be faster and cheaper. We talk also about ongoing maintenance and support, where if you’ve now created your own analytics engine and if your customer asks for the new chart type, or they didn’t like something, it’s every time, you’re spending developer resources into it. But if you are embedding something, you can almost offload this to your analyst or your product manager or your solution engineers.

There’s also a lack of expertise. These people may need to hire specific visualization developers, that again, taking away from the standard core business that they are trying to build. There’s also technological advancements, right? Like, if there’s new technology advancements, for example, generative AI comes into action, you want to focus your resources in building that for your core competencies, rather than adding AI to your analytics engine. And then lastly is like, data governance and compliance.

A lot of these companies are not data companies themselves, so they’ll have to figure out how to actually manage this data at scale. But with some of the embedded analytics products available, like Sigma, you have built-in governance and compliance readily available when you embed the product. So, we went in with this pitch, and we were able to convince a lot of small to medium businesses and a lot of startups who signed up with Sigma. And the success that they got, once they embedded Sigma was tremendous. Like, a lot of our customers were able to monetize the data that they already had within weeks, if not a couple of months at max, with really lean development teams, like, two to three developers, a couple of people on the business side, and generate, like, seven-figure income streams out of that.

Brian: Is that the primary goal is monetization? Is that what makes them pick the phone up is that they think they have enough… there’s enough insights that their customers would want that they would pay for to get? Is that the primary driver?

Zalak: Right, yeah. That is one of the biggest drivers. I think… the monetization of data is, I would say, probably about 50 to 55% of our customers, but the rest are more, “I have this product now, but customers want insights, and if I don’t provide them, they might go to my competitor. So, how do I get to this parity faster?” And here comes Sigma, we help them embed your analytic solution really fast.

And recently, though, we’ve been starting to see a lot of our enterprise customers in embedding space as well. They were either using some of the existing tools, and then they are not really happy with the either customer support or the engagement that they’re receiving from these companies, or the technology is not advancing fast enough or providing them the capabilities that they wanted. So, in that case, again, we go in with the benefits that Sigma provides, which is again, like, speed, security, self-service, and you know, sales, which is the monetization aspect of it. And we convert them really easily just because Sigma itself as a product is really powerful, whether it’s, like, spreadsheet type interface, and being able to write [C:Python 00:33:25]. So, with all of the operating complexity that we just talked about earlier, it becomes really easy to convince people to essentially buy Sigma over something else. So, you’re seeing a shift now on the enterprise side as well, whereas over the last year, early last year in 2022, we’ve been doing really well, like, dominating the mid-market space.

Brian: Can I export it to Excel though?

Zalak: [laugh].

Brian: Is Excel a friend or a foe [laugh]? How do you—I mean, I could see—you talked about the financial analysts. I’m kind of curious when you mentioned you have a spreadsheet-style GUI and experience, there’s always the latent oxygen in the room, which is Microsoft Excel. How does that fit into the experience and your product? Like, how do you see that? Is it an enemy? Is it a friend that we dance with? How does Sigma see that?

Zalak: Everybody’s a friend, right [laugh], in a way.

Brian: [laugh].

Zalak: We really ask the question about why do you want to export to Excel, right? So, if you see a spreadsheet, a lot of times, financial analysts want to export to Excel because they want to do pivot tables, and then that’s not typically available on BI products. But in Sigma, you can create pivot tables. So, if your end goal is to just create pivot tables, then why not just do it in Sigma? There is another push here where, sure you can export to Excel, but as soon as you download it, that data is stale, whereas if you are on a, you know, cloud BI tool, your data always stays current based on your cloud data warehouse.

So, there’s a lot of arguments against exporting to Excel, but Excel is still Excel, right, so if there are certain things that you are able to do only in Excel, then sure. But most of the time, people are just wanting to create pivot tables or drill down because, you know, this tool that you’re using is not providing the drill-down capabilities. With Sigma, you don’t really need to export because you can drill down, you can add filters yourself, add controls, and just do explorations, then create pivot tables. And then if you want to create a visualization, sure, you can do that as well. So, a lot of things that people would typically do in Excel are possible inside Sigma as well.

So, our argument is—typically to the admins or the data governance folks—is that, well, you can maybe just disable Export to Excel because you want to be compliant, and you want to have high levels of governance. In that scenario, you don’t need to export to Excel.

Brian: Yeah. I can understand the, you shouldn’t need to go do it, but there’s always the resident behavior, the kind of the status quo, and I know teams are struggling—not just you; this is a common user experience friction… place of friction that has to be dealt with a lot of times, so thanks for sharing a little bit about your perspective on that.

Zalak: Of course.

Brian: The topic of AI and BI, I’ve definitely been seeing this pop up a little bit. I know you had some comments you wanted to make. What’s your take on how AI is impacting the BI industry? Are we moving to the place of less chart bouncing and drilling down, and more interrogating with natural language and getting results back? I mean, a lot of these LLMs, for example, don’t do math, and I don’t think even the average, probably, user—like, I don’t know, maybe your client—your customers’ customers know this, but they don’t know ChatGPT doesn’t do any math, it just guesses the next word.

So, when you say, you know, “Summarize my sales”—and it doesn’t tell you it’s lying, right? Because it doesn’t know it’s lying to you because it didn’t do any math; it just guessed stuff. So, I’m kind of curious, like, are you guys moving to that direction of natural language interrogation of the data, or is that not how you think about this?

Zalak: No surprise, really, Brian, on this one. AI is pretty much everywhere, all over the news, every day new things are being launched. I think there was a time last year where there were new BI products—or spreadsheet products, really—that were powered by AI, launching every day. And if they were allowing free trials, I pretty much signed up for everybody [laugh] I could, and none of them actually work because they were just riding the hype, right? So, at Sigma, our stance is, let’s not just add AI for the sake of adding AI. Let’s really identify in the entire user journey, like, where does the intelligence really lie, and where are the different friction points, and let’s enhance those experiences.

A lot of times when you are using a new product, you don’t know where to start from because your analysts maybe have created lots of datasets and tables that are now available for you to do analysis on, but you don’t really know where to start. So, that’s where your point comes in, where natural language interrogation is going to be the entry point into what we see in the future of BI. And again, going back to the first thing I ever said was, we don’t want the insight to really die. So, we allow you to ask a question, and we provide you the answer, which a lot of other products do, but the whole point about working with the OpenAI LLMs is to have that interactive behavior. So, we want to not just stop at giving you the answer; we will explain to you how we actually ended up with this answer.

So, the way that we are thinking about is… is this is a great starting point for a user where if they are learning a new product, if they are trying to create a new dashboard, we want to augment their Sigma’s learning capabilities using AI. So again, like a lot of other BI tools, we do have a natural language interface now which has a chat-based interface, where you can ask a question, and we will find the right sources, create these charts for you, but you can interact with it. You can say, “Oh, actually, I don’t like this bar chart. Can you convert it to a pie chart?” It’ll go and convert it.

Or if you want to add just another column with a calculation, it would do that. You’ve also fed in all of our formulas, essentially. So, if you’re writing a formula which is wrong, and if it’s a really long formula [unintelligible 00:39:35] analysis, you can ask the AI to help you fix it because it knows the right formula. With a few inputs, you are able to create all of those things. So, we are working on how do we remove friction and enhance the current users’ experience rather than adding AI for the sake of adding AI.

Every point where you see that the current level of our generative AI LLMs is advancing, we will start adding a little bit more to the product and see how we can reduce more friction points. But again, we are starting to see a lot of increase in demand from our customers as well on incorporating some of the AI functionality within the product. So, working with a few other customers for getting what really exactly the problems that they’re trying to solve, and then incorporating them slowly into the product, like, that’s the plan we have.

Brian: What are some of the problems that they come to you with, assuming they’ve backed it up from solution back into a problem? Are there examples of that that you’ve gotten that are fairly concrete that you can share?

Zalak: Yeah, one of the biggest things that we’ve been seeing, at least recently with our embedded customers, is the pressure from their executive teams to have an AI strategy. So, because Sigma is a large part of their product, they come to us and ask us, “Hey, what is our strategy? How can we actually collaborate on figuring out what’s best for this product?” So, combining what we have in our intelligence with what intelligence that they are building, they can almost have a—like, we can create an API for our Copilot functionality, which is a natural language interface, and provide them with a—like, they can basically create their own UI with the API that we are providing, merge that intelligence, and have a separate AI intelligence which lives outside the iframe. So, that is the most crucial thing that we are hearing from a lot of our customers is, “Help us create an AI strategy for our product because you are a very big part of our product,” rather than just having a point solution.

We’re also hearing a lot of requests around summarizations. So, if I created this dashboard with a lot of different visualizations, help me summarize this dashboard, or help me summarize this chart. So, that’s coming up a lot these days, as well. And the third one is lot around usability. How can we make it simpler for any user to be able to be a data analyst or data expert? So, more and more commoditization of being able to analyze data.

Brian: You talked about, like, kind of this default dashboard and everybody can start with that. It feels to me like I guess if I was in your role and I heard that they wanted a summary of the dashboard, I would say the dashboard design has a problem because already we’re asking for a summary of the summary. Like, what happened [laugh]? Do you see that as, like, a, that’s a design problem with the dashboard to start with? And I’m not saying that, you know, again, these LLMs potentially could replace a lot of the eyeball analysis time that might need to happen, but at the same time, we’re also visual, and sometimes seeing the information is going to be much faster. Our brain is going to process that much faster than it is words about the visuals.

It also depends on how complicated the visuals are such that they even need text rendering, text explanation, and some kind of verbosity that goes beyond well-titled charts that explain what the meaning of the data is, not just what the lines are. “The red line means this. The green line means this.” That’s not really helping you deliver the insight. So, do you see that as a design problem, or that’s a—it’s not, like, when they’re asking for a summary of the summary?

Zalak: It depends on the context here. For example, if I am trying to create a summary of this chart and I want to present it to my manager, in that scenario, a written text is, along with the visual, kind of goes well, but I could definitely see an argument here where, “Well, let’s just look at the chart. What’s not clear about that?” Yeah, I think that’s an interesting way to look at it. I will definitely take this back to our product team, who’s working on this specific feature.

Brian: [laugh]. And just for the record, I mean, haven’t seen any of this, so I’m just talking in abstractions here because we’re not looking at the interface, so I’m not even passing a judgment on it. I’m just thinking, what I would smell if I heard that. Like, my spider sense would go off and say, “Well, wait a second, the core thing is wrong.” Like, we shouldn’t need to be summarizing the dashboard if it’s supposed—I mean, this also assumes that a dashboard is intended to be some kind of high-level summary, and again, context of use really matters. Is this an operational one where I need to check it to make sure nothing needs my attention, versus a—

Zalak: Right.

Brian: —how is my business doing? I look at it once a month because I have to report some numbers to somebody else, and it’s not doing any, kind of like, operational work or helping me run the ship, so to speak. There’s different contexts of use here. And that’s what’s the hard design problem you have, I would think, is just these multiple different industries, contexts of use, the user tasks are really different. It’s very hard to design a great abstracted UX around something like that. It’s just tough.

Zalak: I agree. I agree. I think it’s—the way that we are thinking about this, generally, in terms of building embedded analytics product is, we provide the right kind of tools and support for our for customers to create and curate the experiences that they want. And if we are able to do that consistently well and take feedback from them as we are [unintelligible 00:45:10] and updating and adding new things into the product, I think it works well for both the customer and their clients.

Brian: Do you identify as a data product manager? A software product manager? Like, does that word mean anything to you? I’m just kind of curious how you self-identify?

Zalak: That’s a great question. I feel like I still identify myself as a product manager, but everyone at Sigma is a data analyst, is how I see it. Everybody at Sigma logs into what we call Sigma on Sigma for anything and everything that they do on a daily basis. So, I feel like I’m a Sigma expert, but I’m not necessarily creating a data product. My customers are data product managers because they are taking the data that they have and then either monetizing it or presenting it to their stakeholders. So, I feel like my customers are all data product managers, but I still identify myself as a product manager who is a data expert [laugh]—

Brian: [laugh].

Zalak: If that makes sense [laugh].

Brian: No, I understand. I mean, these words mean different things to different people, and I don’t think there’s necessarily a right or wrong. And I would say that probably tracks with the majority of, like, the people in the DPLC, the Data Product Leadership Community that I started last year, I would say that probably tracks pretty well. I tend to think of software product managers like yourself that are building analytics tools and this, as DPMs as well because they’re kind of doing it at a meta level. You’re doing it at an abstraction level, in a way, and I kind of grandfather the meta, I guess you could say. But there’s no right or wrong. And anyways, it’s just it’s self-identif—it’s a label, you know [laugh]?

Zalak: Yeah. Yeah. I definitely consider myself as an expert on Sigma, so—

Brian: Right, right.

Zalak: And everything that I do from, like, tracking metrics on features, and consolidating Jira requests, and all of that happens in Sigma for me. But I do build the tools that people can then use to then embed Sigma into their products, so—

Brian: Sure, sure.

Zalak: —I could see it [laugh].

Brian: It’s been great learning about this tool, and I want to let people figure out how to get in touch with you, and if they want to get a demo, too, how to do that, but before that is there, like, one thing, that’s the biggest challenge for you as a product manager in this space? Like, one thing that stands above them all?

Zalak: The biggest challenge is to work with so many different personas at the same time where your core product has a different persona that you are catering it to. So, our core BI Sigma product caters to the analysts of the world and the business users of the world, and we don’t want to reduce the velocity of that product. We want to keep making innovations and updates to that product. The biggest challenge here in the role is how do you extend it to the product managers and developers in our customers’ clients? Every time you think about a new feature or functionality, you have to make sure that it kind of works in this first-degree persona and the second degree persona, and how it will be viewed by these different personas because that is not the primary persona that the foundation of the product was built on.

Brian: Right.

Zalak: So, I think that’s the biggest challenge that I see on a day-to-day basis where I want to make sure we keep releasing new things and cool things for our customers, but how do we extend it correctly so it doesn’t create security issues or UX issues for our customers and their users?

Brian: Cool. Again, this has been a great conversation. Any final words before we kind of talk about where people can get in touch with you, but any kind of closing thoughts for our audience that you want to share?

Zalak: A lot of people in your audience are data product managers, so I want to share that reach out to me [laugh]. There’s a lot of cool things that Sigma is doing, and I want to work with you and figure out how you’re building your products so that we can incorporate a lot of these new requests and solutions [laugh] that you might want to create into our product. 2024 is going to be a push towards how do we extend Sigma’s embedded capabilities beyond analytics into your applications. So, I’m really excited about that, and I want to learn a lot more from all the other data product managers who are trying to create these products today.

Brian: Cool. So, how do they reach you [laugh]?

Zalak: They can email me at zalac@sigmacomputing.com, or they can find me on LinkedIn at Zalak Trivedi.

Brian: Cool. And that’s Z-A-L-A-K, correct?

Zalak: That is correct.

Brian: Got it. If they want to just kind of check out the tool, maybe they don’t want to contact you, but they do—they’re curious about this kind of embedded analytics solution we’ve been talking about, where would they go to get that? What would be the URL?

Zalak: You can go to sigmacomputing.com/embedded.

Brian: Got it? Okay, cool. Excellent. Well, Zalak, this has been really great to chat with you. Thank you for sharing a little bit about how you make the sausage for the people who make the sausage for the people that eat the sausage. So [laugh].

Zalak: [laugh].

Brian: The double hop problem [laugh].

Zalak: It was a great experience. Yeah, I love having this conversation and discussions around embedded analytics, AI, and where the future is. I’m super excited about the future.

Brian: Cool. Well, take care and we’ll hopefully stay in touch.

Zalak: Will do.

Array
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Subscribe for Podcast Updates

Join my DFA Insights mailing list to get weekly insights on creating human-centered data products, special offers on my training courses and seminars, and one-page briefs about each new episode of #ExperiencingData.