Zing's CEO on the Monday Morning Data Chat with Ternary Data

Zing CEO and co-founder Zack Hendlin sits down with Joe Reis and Matt Housley of the Monday Morning Data Chat and Ternary Data on their podcast to talk about why the future of data is increasingly mobile, why he and Sabin started Zing Data, and whats next for business intelligence. They also cover robots with lasers, building a data team at a small company, and the importance of good design.

Joe and Matt wrote a bestselling book on data engineering and run a data consultancy.

Podcast links to Apple Podcasts, Google Podcasts, Spotify, and more are here.

Unedited, autogenerated transcript below

Matt Housley 0:01

Happy Monday. Happy Monday. Good to see you, Joe. It’s actually been a while since I’ve seen you in person. It is even though we live in the same town. We’re actually recording from Brooklyn, New York right now.

Joe Reis 0:13

Stranger that happens. I just got on the red eye. This morning. I flew from Salt Lake at midnight and got in at six, which actually means four in the morning for Salt Lake anyway, so it’s gonna be a fun day. Yeah, those red eyes are always super convenient and always super painful at taking the meds. Yeah, it’s fun. Enough about us. How are you doing, Zack? I’m doing great. I’m doing great. Nice to nice to be on your show. Yeah. Good to have you and have you on. For people who don’t know who you are. Do you want to give a quick intro? Sure. I’m Zack Hendlin, I’m co founder and CEO of zing data. And we are mobile first analytics. So you can query any major data source in a couple of seconds on your phone.

Zack Hendlin 3:10

you saw you saw that happened with email. So remember, like when the iPhone came out, and it was initially, you didn’t see sent from iPhone that much. But then occasionally, more and more, you’d see people would actually go from viewing an email on their phone, but responding maybe when they were back with their computer to actually sent on my iPhone, which maybe was sort of like a brag, but also a sign that you were now able to start doing more and more and more from your file. And what were the things that let that happen? Well, faster internet connections, LTE, 3g, and then LTE and 5g and autocorrect personalized language dictionaries, that meant that you didn’t have to re correct a miss type each each time. Auto Replies now with Gmail, where you literally can, can pre select responses. And so there were a set of things that took it from basically view mode on a phone to you actually can respond to a lot of emails on the phone. And we think that is sort of where email was call it 10 to 15 years ago, because if you look at the trend lines in terms of like how people actually use their phones and use their computers, in the US, there are a greater percentage of US adults who have a smartphone than have a laptop or desktop. It’s actually 85% of US adults. According to Pew Research in 2021 have a smartphone 77% have a desktop or laptop. So if you think about kind of accessibility.

More folks have a phone than a computer. And you also see that with how folks are increasingly getting work done. So 76% of slacks. Weekly active users use Slack on their phone in a given week. Think about that. And in fact

Joe Reis 5:00

You go, first you go a little deeper, you’re like, Well, maybe they’re just reading stuff turns out actually more than 20% of all actions. And this was as of three years ago. So it’s probably hiring now. More than 20% of all actions on Slack happen from someone sending them on a phone. And so this shift, QuickBooks is more than 5 million downloads on Android, Allah, the shift to actually getting work done yet, the shift to getting work done on your phone is happening. And it hadn’t happened broadly in data yet. And we looked at that and said, that’s a huge opportunity. If you’ve ever used Google, Google sheets on your phone or Excel on your phone, it’s not a great experience. And and that’s, that’s already got more than a billion downloads on Android alone. And so if you sort of look at that, you kind of it doesn’t take too much work to realize that the way people are getting work done is increasingly multidevice. And that if you’re limited to hey, I can only ask this question, write this query, create this dashboard. You know, if I’m in front of my computer, writing sequel, or Python, or whatever, that ends up being really, really limiting to how people work, people are on their phone more than four hours a day in the US, US adult is on their phone north of four hours day, growing 5% year on year. And so you look at that, and you’re like, Well, why can’t I do anything except basically viewing data? And that was sort of the realization that sort of now is the time in the same way that you know, for email, I think that was probably 1010 or so years ago, on mobile.

So So what’s different, right, because I think a lot of the responses would be Well, that’s nice. Sack. I like looking at charts on my phone, too. Not really. So what what are some of the use cases or the I would say the the user experiences that you envision with mobile that you really can achieve? With a desktop?

Zack Hendlin 7:00

Yeah, I’ll give, I’ll give a couple examples. So one is a major retailer. In the US, they sell pizza, and they actually want to go and audit franchises and figure out if a franchise is performing as well as other franchises. And that’s actually something you could create as a dashboard. But if you say, well, actually, I want to cut it by this product. Or actually, I want to view it for other things that are in the state, unless all of those filters and all those things have been pre specified in a dashboard, typically by someone at a computer on a data team, then you’re kind of out of luck when you’re going in and literally doing a store visit and trying to figure out what’s going on with the store and why they’re maybe not performing as well.

Or another example is, we have a company in the event space. And so they literally hold in person events, we actually had a film festival use us. And they were looking at the number of votes for different films as they’re coming in. These are people literally running between different showings. And they don’t have a big deep bi or data science team. But they need something more sophisticated than a Google sheet. And so they literally would look at votes by by movie, and could read off results and say, Okay, this one’s going to the finals, or this one’s winning this prize from wherever they are. So we’ve seen it sort of be used in two ways.

One is traditional companies and data teams and startups where execs or sales folks use this to monitor stuff.

The other though, is actually 80% of the global workforce, which is deskless - emergence capital has a really interesting paper on this. And 80% of the world’s workforce is not primarily at a desk. And in fact, Joe, when you were flying it out, hopefully you weren’t doing too much work in the on the red eye, or in transit from the airport. But if you were flying at a non red eye time, perhaps someone would ask you, you know, hey, how many of our users use this product yesterday? Or, hey, this prospect that I’m going to meet with in 20 minutes? How many active users do they have if I’m going into a renewal conversation. And as a salesperson, you don’t always have someone on the bi or data team to have pre created that for you. So we basically built this really easy to use kind of mobile interface that that lets you get to that, that answer without it being pre created. And it turns out that that’s this huge swath of folks that usually we don’t think about when we talk about data teams, or even traditional data consumers, because they’re folks who are fixing your utility lines. And they’re folks who are driving trucks and in the logistics supply chain, and their alternative is to call someone Yeah, all someone and ask them for a number. An example that came up recently with someone check to wanting to check into a hotel early, and the manager was walking Get around the lobby. And the guy asked the manager, Hey, can I check in? The manager says, oh, no, you can’t do early check in, we don’t have any rips. person walks up to the front desk, then they say, Hey, can you just can you just double check? Are there any rooms and the person double checks and says, Oh, I just refreshed the data? Yes, there’s actually room that just opened up. And so this person was able to check in. But right now they had to go through multiple people, someone had to be in a computer to get them that information. And if you imagine what the optimal experience is, it’s, you go up to the manager who’s walking around the lobby and say, Can I check in early? And he says, Let me check, yes, you can. Room 27, a go to check in and they’ll get you set up. And so these are like data use cases in the field that you probably don’t think about, if you’re thinking about like product manager, engineer, data scientist, salesperson, in a structured away,

Joe Reis 10:54

lets get technically nerdy for a second here. So a couple of months ago, we had Jordan Tigani. On our show. He’s working on a product called Mother Duck right now, which is duck DB using wasum. In the in the browser? Are you guys taking a similar approach? Or have you considered a similar approach of having like an on on device OLAP database, which would basically allow you to return results very quickly and crunching numbers on the device. So yeah,

Zack Hendlin 11:23

so so on, on device works really well, for small data sets. So the the types of use cases that we’ve seen, at least initially, are actually bigger data sets than you might expect. So if you want to know, let’s say, sales by day, well, that data actually comes streaming in maybe even by hour, depending on how you’ve set up your your data transfers and all that stuff. And so we think that they’ll probably be areas, we offer that in the future. But we haven’t done this sort of purely on device queering. Rather, we’ve said, we’re going to do a ton of optimizations, and we can talk about them if you want. But we’ve done a ton of stuff to actually make the thing work well on mobile. And an example two quick examples of that are, you don’t have a persistent internet connection, when you’re on a phone necessarily, you could go through a tunnel, you could be in airplane mode. And to make that work, well, you actually don’t want the client to query the server, you want the client to persist a connection to a server, which will persist the connection database, and the server and database will run that query, you cache the result. And then you send that cached result to the client when it is online. And so we do that we send a push notification, we background load, the cached result. So it sort of just works. But if you didn’t do that, what you’d have is you run a query, maybe you go through a tunnel, or you just have a congested mobile network, and the query will fail. And you’re like, Okay, this doesn’t work. Data doesn’t work on mobile. And in fact, I, in a previous life, tried to write SQL queries in a web interface on my phone, just to just get some work done. I know, it just, it was not the most efficient way.

Joe Reis 13:12

This is harder than just waiting.

But it was you could tell I was really into mobile data before it was a thing.

Right, just as a quick aside, imagining when the iPhone first came out, one of my friends wrote an entire blog post on the iPhone, and it’s such a novel things like I am writing a blog post on my iPhone one. Right? It’s sort of ridiculous. Like, why would you do such a thing is that this was back in back in the day. So anyway, carry on. I remember people writing blog posts on their iphones.

Zack Hendlin 13:50

But I think now, I think this is actually a great, great example. Because now you actually could dictation is not amazing, but it’s gotten enough better, where you could actually draft thoughts. Or you could kind of get the basics of a blog post down, maybe you’ll refine it later. Maybe that’s not the version you share out to the whole internet. But you can actually kind of get your thoughts down quickly, maybe just through dictation right now. And so I think a lot of it comes down to how the technology matures to make it work better and make it more usable. And we think that’s kind of where there’s this interesting inflection point, where if you think through what the mobile experience is first, and I think Robin Hood did a good job of this. I think Canva the design tool, they’ve done a good job of this. And basically they said, What is a good experience to trade stocks on the phone? They didn’t say, hey, let’s take what’s on web and kind of smoosh it down. And I think that’s actually a pretty key part of how you build a better interface. Because for instance, if I run a query and as I did when I was trying to write sequel in a browser window on my phone And many, many moons ago, it would just timeout, or it would give me a result set that was like a billion rows. Both of those are bad. And so then the question is like, Well, what do you do to handle that natural question like limit 10, limit 100. But then you’re actually getting into results that that’s, that’s pretty bad. Because you actually don’t necessarily just want 10 rounds, you actually want the biggest contributors, you actually want a windowing function, probably, or a ranking function in the background, based on the biggest contributors to the thing you care about. So a fancy data scientist who maybe reads the best selling book about data engineering, might have that level of familiarity. Yeah. But most folks don’t. And so we actually automatically apply windowing functions for large result sets where there’s more stuff to display on your phone than would make sense to do. And then we say, Oh, by the way, we limited results, click here, if you want to see him. Okay, it’s awesome. That’s like one small example of optimizing the experience to make

Joe Reis 16:04

some good comments here to like a question says, you know, I guess it might work for some predefined and more frequently use queries, I would love to see that this combined with natural language analytics to have a voice enabled analytic system. I think the more broader point to this question, which I think excites me about the mobile device is the fact that the mobile device has a lot of mobile native things like sensors and a camera. Right? GPS, imagine what you can do now with analytics, where you’re not just related, you know, the whole world of possibilities opens up where I think, you know, it’s it’s sort of like, as you say, email, or just the phone in general as an interface. Like it just opened up a whole new world of possibilities and how to interact with the world.

Zack Hendlin 16:47

Even screen sharing, so with your phone, you can do you can with a you can Chromecast if or use Apple TV. And so you think about like, Hey, I have a graph here. What is a way that this should look on a screen that’s larger? What can I do? Let’s say I’m searching for nearby locations, or searching for something in a certain radius. I could precede that query with my latitude and longitude if a user gets permission. So we don’t support that GPS enabled thing today. But that’s along the lines of how we’re thinking about what the right experience.

Joe Reis 17:24

The possibilities are endless. Yeah, frankly, it’s Yeah, I mean, it’s kind of cool. Like, back in the day, I got the an Apple watch, because nerding out on the accelerometer for weightlifting, to see like, if I could detect like a perfect lift. It’s actually a really hard problem to solve.

Matt Housley 17:41

That was like a week,

Joe Reis 17:43

probably better than we actually. Yeah. But, but the whole notion was okay, I could probably figure out analytics just using a lot of the sensors onboard the the watch or the phone, for example, right? Like, again, you have some you can collect so many more data points, then you could just offer a laptop.

So totally. Now, the question that came up about voice specifically as an interface in natural language, so I shipped Facebook’s first work in speech recognition, actually many, many moons ago. And the challenge with voice when it comes to query and you see this with Alexa or google assistant, or Siri is, there’s two challenges around how you interact. One is knowing what you can ask. And then the second is one of disambiguation. So a challenge with saying, I won’t even say it because they who shall not be named that have speakers and microphones will probably respond and mess up the mess up the podcast. But if you say, hey, insert name here. Give me revenue by quarter. Well, there’s a disambiguation problem with revenue. Is it net revenue? Is it gross revenue? Is it revenue underscore with underscore tax? And so you can ask a user via voice, do you want revenue underscore with tax revenue underscore gross revenue, right? But if you’ve ever gone through a interactive voice response system where you’re like pressing one for this or replying yes or no, it can be pretty frustrating.

Yeah, exactly.

Zack Hendlin 19:16

The fancy phone tree. And so that disambiguation part can be hard. And so actually, that’s not where we didn’t start with NLP. Because we think there’s a lot of challenges around. We’re the best word arrays, if you look at like Google’s benchmarks, they’re in the mid 90s, basically, and that’s on domains that you know, but if you’re literally using like terms in a database that are arbitrary and have underscores and all this sort of weird abbreviations, it’s probably quite a bit lower. So does the user know what they can ask? And then is it kind of easy enough to use with all the disambiguation and all that stuff? And so like, we think it’s gonna get there and we’re doing some some work on that ourselves. But we actually don’t think that that’s the sort of killer use case today. He, just because it’s actually a little bit more confusing than saying, oh, I want, you know, gross revenue by year, boom. And it’s that’s too tight with with us that’s like two tabs. Okay, I’m gonna stop sharing.

Joe Reis 20:15

I’m fine. It’s a good use case. So are there any other like kind of killer things today that is on your mind?

Zack Hendlin 20:28

I think a really interesting area that when you put in interface, and we buy, as contexts, we support iOS, Android, we also support the web, because our view is that you kind of need to work everywhere. So you still work at Facebook and LinkedIn. And if you ran an experiment, only on mobile, or you ran an experiment only on web, you actually wouldn’t capture the full network effects. And that, basically, is knowing that if I send you a message, you could reply quickly on whether you’re on web or on your phone, and everybody in the experiment. And we did some pretty fancy experimental designs, where we would select similar lookalike groups, but make sure there was very highly network density within a group we were turning it on to. So maybe I want to test a group of 100,000 people who are tightly connected, and I turn on the feature for that 100,000 people. And then I try to find a look alike, other 100,000 people with overall similar attributes, but who are not deeply connected to this, this test group. And so when you did all these fancy experiments, you’d find what is pretty common sense, which is, if it’s hard for someone to reply, because they can only reply on a computer, you’re gonna get fewer replies, and turns out the feature is going to test less well, because people are not able to kind of respond quickly when you add mention them. And so by opening up this sort of mobile mechanism to interact with folks, you actually can better realize the network effects. Because if I tag you, so we have like this collaboration, where if I tag you, Joe, and say, Hey, can you take a look at this? Do you know if there was a data outage Wednesday of last week, because I see a dip in the numbers? If you have to be back at your computer? To answer that question, I’m going to think that like, the system doesn’t work very well. And people aren’t very responsive. And it’s kind of a ghost town. And I don’t get answers to my questions. But if you can reply within 10 minutes, because you get a notification, and it’s on your phone, and you could reply in line and be like, yeah, Wednesday we had an outage, I would exclude that. Great. Now, this whole system works a lot better. And I’m much more likely to keep using it and and kind of be a sticky user, if you will. So that’s one of the things that a sort of available everywhere, layer opens up that web only or even mobile only wouldn’t, wouldn’t allow. And I’d seen that from Facebook and leave LinkedIn. But that was kind of one of the things we thought about when we were leaving your building saying

Matt Housley 23:08

this is cool. I mean, what I think I’m hearing is that instead of just trying to take the desktop data experience to mobile, you’re really looking for new things that can happen on mobile that maybe people aren’t doing right now, like more dynamic interactions as such. Let me ask you a fairly open ended question which you’ve you started to answer a bit. But maybe maybe you can give me a bigger sort of theory of how this works. I feel like the big transition to like successful smartphones away from like, the early smartphones that no one really used was UI and UX, right? It was all about like user interface and user experience. So you’ve told us a bit about like how you think about user interface, but you sort of have a grand unified theory of, say, user experience for data and how to solve those problems. Because I think that fundamentally, is the really hard problem, right? Like getting data to be something you can interact with on a phone, not through SQL, and not through really awkward controls. That’s tough.

Joe Reis 23:59

I think I’ve seen some other analytics companies that won’t be named, basically reporting their experience onto the mobile device. And it’s just, it’s just like a crappy version of their web interface. You know, I mean, I’d rather go use a web interface. So yeah, if possible, it’s just not seeing the art of the possible, right, like you’re using a device, which is like, the phone, the iPhone or the Android. I mean, this is a it’s a work of

Zack Hendlin 24:27

pinch, you can pinch to zoom, you can do these interactive, I don’t need to have a plus button than a minus button. You

Joe Reis 24:34

to test you do is you get kids, right to go on a laptop, the first thing they do is they touch the screen, at least my kids do, right? That’s the experience they’re used to. And so you’re designing not only for, you know, the possibilities of today’s devices, but also for the future generation that grew up on. They didn’t grow up on laptops, if you’re talking about the developing world. People don’t have computers there. They have phones. They sort of leapfrog the entire desktop Up in laptop world. So,

Zack Hendlin 25:03

yeah, we actually we have a company in Argentina, this guy runs a series of retail channels like, they sell hubs where goods, and they’re not a big name, you probably have never heard that I had an either about 1010 stores. And this guy, he’s only logged in from his Android device. And he literally added a Postgres database as a data source. And he checks each stores sales by day and inventory and stuff like that, as he’s running around, checking on various stores. And I think that’s probably a different usage pattern, that if you went to traditional BI vendors, most of them you actually can’t even add a data source from your phone. And this guy literally is running his whole business from his phone. And so I think when you go a notch or two outside of like, the traditional, what we think of as kind of primary data users, you find that there’s actually this huge number of folks who, if you make it easy, if you make it work, well, actually are very interested in this thing. And I think, Matt, to your point, it very much comes down to how you do it and how you do it, right. So I was looking at a step by step guide for a VA tool that shall not be named, and wasn’t as though an established large player. And it was literally seven steps of building a custom mobile friendly variant of an existing dashboard, and how you would test it on different screens, and reflow, all this stuff, and do all that manually in it. I went through all those steps. And it’s basically hours of work to make a thing look good on a phone, one, one dashboard. Oh, and then you have a clone of that dashboard. So yeah, it’s something here, and it doesn’t propagate. So you have to redesign it again. And

Joe Reis 26:53

well. It’s a lot of work. I mean, we get a lot of embedded analytics use cases in your iframe, hell at that point. Yep. Yeah, it’s I mean, I do appreciate the the efforts, but it is a lot of effort. And you do have to appreciate that effort. And so yeah, but

when you guys are when you guys are doing an embedded kind of analytics use case, how do you think about, okay, I need to go check this in an iframe, or I need to go resize this stuff like, how do you think about that? Or like, what’s your flow?

You mean, as a user, as somebody who’s building this? Yeah, I just think it feels like a lot of work. I’m like, it’s easier, right?

Matt Housley 27:34

I try to get other people to do that stuff in all seriousness. Because I’m not a UI guy. Right,

Joe Reis 27:41

right. And then you give it to it, then you give it to a developer, right? So adding this iframe, and then just like, can you go away, a lot of tickets have to be knocked out. So that’s the other disconnect is it feels like data and Dev developers a sort of a dichotomy there. If you give them the work that’s like, you know, front end Dev, for example, right now they try you know, or a mobile dev, you try and get them to build an analytics dashboard. It’s just they don’t understand,

Zack Hendlin 28:07

or you can use d3js, and it looks really cool. And you’ve built it like this very, very specific way for this one very, very specific graph or question. And then, oh, actually, we want to filter to do this, or we want to be able to zoom it zoom in, or we want it to work well on big screens and small screens and dynamically reflow. And like, is all that technically doable? Yes, it is. How many companies are going to put the time in to build that from scratch with d3 or native libraries and all that? Very, very few? And very, very few should? And so if you think we look at like figma. On the design side, they basically said, What is the thing that used to require like Adobe Illustrator, Adobe Photoshop, or sketch, like it was kind of a single player thing? And how do you make it like, you can start for free and you can use it without being a hardcore designer. And then oh, by the way, you actually can look at what your colleagues have done, because you see, like all those shared all the shared context. So you can piggyback off of what the designer on your team did. And like maybe you start tweaking it, kind of like a lot of folks learn SQL, maybe they modify someone’s SQL, and then they’re like, Oh, this is how it works. And you learn that way. And so we think there’s this whole element that is for data as a whole, somewhat nascent aside from like, hardcore, like machine learning models and checking in your code to GitHub and stuff of like, shared questions and shared understanding this sort of collaboration layer around data that I think is going to become much more exciting over the next couple of years where it’s not just I do analysis, I put it in a presentation but rather that is like a live analysis that everybody’s weighing in on and so I think that’s gonna be we’re gonna get to see this like shift from single player to multiplayer happened. For docs, right Microsoft Word to Google Docs happened for design, illustrator balsamic sketch to figma. And I think it’s starting to happen with that a lot of tools out there have some basic collaboration functionality, but then it goes back to that network effects thing of, can you access it from wherever you are? Is it easy and fast?