PAW 2019 Podcast Series

Ep. 223 – PAW 2019 Conference Series – Allison Swihart – Syndetic

Welcome to the 2019 Predictive Analytics World (PAW) Conference Series. Recorded live in Las Vegas, this series is bringing interviews straight to you from exhibitors and speakers at this year’s event. In this interview, host Jamin Brazil interviews Allison Swihart, CEO and Co-founder of Syndetic.

Find Allison Online:





Hi, everyone.  I’m Jamin Brazil, and you’re listening to the Happy Market Research Podcast.  We are live today at Predictive Analytics World. I am standing with Allison, the CEO of an up and coming company Syndetic.  You can find information on them at Allison, thanks for being on the Happy Market Research Podcast.


Thanks for having me.


Well, let’s start with what do you think?  I mean this is Day 2 or 3.


Day 2.


Day 2 of the event.  What do you think about the event so far?


I think it’s been wonderful.  It feels like there’s a really good energy here compared to some of the other conferences that I’ve been to recently.  The audience is slightly more technical, which is great for us. And I am getting a lot of good feedback on the talks. And I think that the decision to kind of combine the different PAWs together into one Mega PAW.   


Eight PAWs


Were there eight?  There have to four PAWs.  What kind of animal is an eight-pawed animal?  I mean… There should be four PAWs. Anyway, yeah, I think it was a good decision because everyone who came for the marketing side versus the kind of deep machine learning side is mingling together, and you’re getting lots of great conversation.    


It is actually interesting seeing that convergence of…  One of the terms that a previous CEO I had on the podcast said is “diversified data” or “data diversification,” meaning…  You see this in executive teams. So executive teams that have diverse gender, ethnicity, whatever, they outperform non-diverse groups.  It was interesting having him creating that connection with respect to different data. And I hadn’t actually thought of it in the way that you just articulated it, but that’s actually pretty relevant also where you’ve got broad disciplines that are now being combined into a single event, which is kind of cool.


Yeah, yeah.  And having people of different titles and at different levels in the organizations, I think, is really important for conferences.  I know there’s a lot of summits where people kind of focus on just the CIO level, but I think at this event, in particular, we’ve talked to a bunch of people who are analysts, who are doing the actual work, which is really important to bring their perspective of the individual contributor, not just the manager.  The manager might be making the decision to purchase one piece of software over another, but you need the perspective of the kind of “boots-on-the-ground” in order to, I think, make the best decision. So, I think they’ve done really well.     


So, you have a relatively new company.


Brand-new.  We just launched this year.


Congratulations.  What month?




Awesome.  So literally brand new.


Literally brand new.


Tell us what your company does.


So, we are a virtual data warehouse, which means that we allow you to leave all of the data where it lives:  in databases, spreadsheets, third-part services that you access via API, cloud-hosted services like say you use SalesForce.  You have an HR system; you have an accounting system plus you have a non-premise Oracle database. Leave all that data where it is; we virtualize it into a data warehouse as if it were in the same physical location.  So you can query against it, using plain SQL, and define the right slice of data for every project you want to do, and then you can syndicate out the results in different ways. So you can take a slice of data for say you’re building an e-commerce website, and you want to expose a slice of data to Shopify, but you have data in different underlying systems.  Instead of having to copy the data all into one place, you can through Syndetic virtualize it, syndicate it out, without having to do the copying.    


That’s pretty cool.  


Yeah.  It’s actually been around…  So, data virtualization is not new.  It’s actually been around for a long time.  It used to be called data federation, but what we bring that’s different is that we’re the first web-based, cloud-native virtualization system, which is really important because it allows the business side and the IT side to collaborate inside of the same system to define that slice of data,  So, if you have, for instance, metadata that is living in a spreadsheet somewhere and it’s important that it live in that spreadsheet. Like we take a very business-friendly approach. We love spreadsheets. We think that there’s a reason that Excel continues to exist and everyone uses it. If there’s metadata living in that spreadsheet that you need for a specific purpose, you shouldn’t have to copy that into a data warehouse.  You should be able to use the workflow that’s appropriate for your business and still get that project done. So we cut time to deployment in half or more. And your analytics are not dependent on the last batch job that you ran to copy the data. There’s also a whole host of other benefits: security, abstraction…   


So, you’re not actually copying that data and moving it into…




You’re just querying it based on where it resides.


And the queries get pushed down to the underlying data sets.  You can, if you want… We give you the option to materialize the view, it’s called, or cache for performance reasons.  But what we say is that should be project specific.


That was my next question.  There’s a lot of overhead.


Yes.  So, well, actually the performance is very good even in the virtual, non-cached version.  But, if you’re running a query on a 850-million-row database, you might want to cache that, and you can just with the checkbox.  So we make it very easy. 


That’s awesome.


Yeah, thank you.


Did you found the company?


Yes, I did.  Myself and one co-founder, the two of us.


Two of you.  It’s terrifying, right?  


Yeah, it’s terrifying although it’s very exciting.  And the two of us ran a previous tech company together.


So good relationship.


Yes.  COO and CTO.  So it doesn’t feel entirely foreign.   


You’ve ridden the ride before.


Yes, yes, the crazy ride.


Who is an ideal customer?


An ideal customer for us is a company between 500 and 1500 people, big enough to have data in lots of different places.  Oftentimes these companies are going through M&A. And, when you do that, you’re acquiring not just operational overhead, maybe a revenue stream, but you’re also acquiring data sets.  And IT integration is a nightmare. So we provide a really good option for those types of companies. We’re industry-agnostic. Based on our personal backgrounds, most of the companies that we’re working with now come from finance or we also have some customers in the insurance and higher education space.  But really any company can use Syndetic.


It’s funny you say that, M&A.  So, the last company I was CEO of, there was eight acquisitions.  It was like this big kind of rollup sort of typical… And the data management on our side – because there was literally eight distinct data – and ironically all of them were data companies. 


Oh, that’s interesting.


But it made it even more exciting.  So, having a unified view of performance and invoicing and customer usage, all that kind of stuff, would have been…  Like we spent a year and hundreds of thousands of dollars, trying to do the integration (actually, it was over a million dollars), trying to do the integrations, creating that sort of like – BI is not the right term – but overarching view, and it was a pain in the ass.


Yeah, there’s a phrase that gets thrown around very casually, is the “single source of truth,” right?  You want a “single source of truth” for your data. Our hypothesis or our opinion on that is that there’s never going to be a single source of truth.  It should be project specific. And you may want to use, for instance, if you saw a physical good… The name of your product that you want on your e-commerce website may be different than the name of the product that you need to use for regulatory filing or that you need to submit to a trade partner. And that might be for business reasons like you want to market your product in a different way than you care about putting onto a report, but it might also be for technical reasons.  You might be sending data to a trade partner or inputting data into a system and there might be character limits, right? So you by virtue of the way that you’re interacting with your other partners in other systems, you can’t use the same single source of truth. And so, we say that’s OK; you shouldn’t have to. There might be a product name that’s living in one system that’s different than a product name in the other system. And differentially, you want to use different versions of the truth at different times, and no one is better than the other.  


So, we’ve seen a lot of growth in data, I mean, volumes of data, and it’s only expanding.  What do you see as some trends that are materializing right now like the last maybe three or four years?  And then, where do you think the industry’s headed?     


So, I think a really big trend and also on the people side…  A big trend is that companies are starting to hire data engineers and data scientists, and they are throwing them into or giving them a mandate of “bring predictive analytics to our organization,” but they’re not really preparing them with the tools that they need.  And so, I think they’re putting the cart before the horse a little bit in that they are so excited about the kind of end-state objective: “We’re going to spin up a bunch of Hadoop clusters; we’re going to have Big Data; we’re going to use machine learning; we’re going to do AI.”  And you can’t really do that without having all your ducks in a row on the integration side. You should buy the systems that are appropriate for the amount of data that you have is another big, core tenet of ours. So don’t buy a system that is meant for pedabytes of data if you don’t have pedabytes of data.  It will take you much longer to deploy; it’s going to be much more expensive; and you can do that later. Buy a system if you have terabytes of data or gigabytes of data. If all of your data is structured, don’t worry about fancy AI on unstructured data.   


Not helpful.


Do the simple thing first.  And so, we think Syndetic is really well-positioned for that kind of incremental growth.  


…and supporting it.  As you look forward, what do you think, where do you think the industry’s headed?


So, I think, for this conference in particular, talking about predictive analytics, I think that we can separate out a little bit the predictions and the analytics.  I think right now we’re still on the analytics phase where people are figuring out what to do with the data they already have. And it’s a little bit backward-looking.  “Let’s take data. Let’s figure out how to apply some statistical models on it to make better decisions.” I think we’ll get to the predictive part over the next four to five years where we can actually do things like decide how to allocate resources based on who we think our future customers are going to be, not necessarily based on who our customers are now.


Oh, that’s very interesting.  If somebody wants to get in contact with you, Allison, or somebody else on your sales team, how would they do that?


So, please go to www.Syndetic, and you can schedule a demo there or you can write to us at


Perfect.  Thanks so much for being on the Happy Market Research Podcast.


Thank you so much.


Everybody else, if you found value in this episode, please take time to screen capture, share on social media.  I really appreciate it and hope you have a wonderful rest of your day.