Drill to Detail - Drill to Detail Ep.65 'Bootstrapping, Growth Hacking and Supermetrics Data Pipelines for Digital Marketers' with Special Guest Mikael Thuneberg

Episode Date: May 13, 2019

Mark Rittman is joined by CEO and Founder of Supermetrics, Mikael Thuneberg, to tell the story of how a mention in the official Google Analytics Blog and a prize of a t-shirt led to him founding and bootstrapping a €2M ARR marketing analytics business that’s probably the most important software vendor the Drill to Detail audience has never heard of, and who recently moved into the data pipelines-as-a-service market in collaboration with Google Cloud Platform and the Google BigQuery team.Announcing Supermetrics for BigQuery: Get a marketing data warehouse up and running in minutes (Supermetrics blog)Supermetrics, Google BigQuery and Data Pipelines for Digital Marketers (Rittman Analytics blog)

So hello and welcome to Drill to Detail, the podcast series about modern cloud analytics and the people and products making the news in this industry. So I'm your host, Mark Rittman, and I'm pleased to be joined in this episode by Mikael Thunberg, CEO and founder of Supermetrics. So Mikael, thank you very much for coming on the show, and it's great to have you here. Thanks. Great being here. So Supermetrics, I've often said that you're the BI company, the biggest one people have never heard of, really, certainly within my world or the world I came from, because you kind of work really with kind of marketing users, marketing personas and so on. So maybe just tell us a little bit about who Supermetrics are

and I suppose the route you had into actually founding that company. Yeah, so many years ago in 2009, I was working at a gaming company in the marketing department. And part of my job was to create reports for various parts of the company. And a large part of my time went into gathering data for these reports. So mainly copy pasting data from sources such as Google Analytics into Excel.

And that was really, really time consuming and really boring. And I figured there's got to be a better way to do this. So I went to look for ways to automate that data gathering process so I could do my own work more efficiently. And in 2009, the Google Analytics API just came out. And I figured that there's got to be a solution for getting data out of that API into Excel. And I did some Googling and I found a forum post by a guy at Google called Nick Mihailovsky, saying that the first person to link that new API with Excel will get a free Google T-shirt.

Actually, there was one guy, a Swedish guy, who had already responded and said that they have a solution but Nick couldn't get that to work and neither could I so I figured that I need to create my own solution I was actually not a coder at all so I had some limited experience of doing Excel macros that was that was pretty much it but I decided that that can't be that difficult to get that working. So I spent a few hours coding and I made a simple VBA script that goes to that API, gets that data into Excel. And it was a pretty good solution. So I responded to Nick Michalowski on that forum and said that I have this solution here and can I get my t-shirt now and then Nick tried it out and was happy with that and he did send me the t-shirt

and Google toothbrush as well but more importantly he then blogged about the solution I had done in the Google Analytics blog and that gathered or generated a lot of interest towards my script and a lot of people from all around the world started contacting me saying that okay you have made this script that is awesome can you do something on top of that that script and I started considering like I had never really thought about becoming an entrepreneur, but the opportunity just felt really, really good. So after a few months, I left the job and I formed my own company to work on Google Analytics API solutions to start with.

And then gradually I then expanded to other data sources. Okay. One of the things that certainly when I moved into this world and from more kind of, I suppose, enterprise analytics before that, I was surprised at how big a deal Google Analytics is to marketers. I mean, it is such an important data source. And as you say, you know, it's not always been the easiest to get data out of there. I mean, so maybe just kind of paint a picture of how important Google Analytics is to people

and why getting data out, as you did, was such a kind of big deal. Well, yeah, obviously, Google Analytics is used in millions and millions of websites all over the world. And it's a really good tool. So you can analyze pretty much anything you want on your site with Google Analytics. But I think there are still limits on the reporting UI that Google provides out of the box. So for a lot of advanced use cases, you would like to get the data out of GA and put it somewhere else where you can really analyze that data yourself and combine that data also with other sources.

So there's a big market for any API integrations in the GA that make that possible. Okay. And another thing that I was surprised about was it's not always easy to get data out at the most granular level of detail. It's sampled and so on. Is that still the case now? I mean, how does sampling work with that? And again, how did your solution work in that sort of area? Yes, so with GA, we work with aggregated data. So, of course, GA does have a hit-level export into BigQuery as well for premium customers. But we work with the API, which is aggregated data. That is sufficient for the vast majority of use cases. Google does sample the data,

but there are certain ways to get around the sampling, and that's been one of our strengths on how we can avoid this data sampling that Google performs. Exactly. So I suppose the other thing really is, again, having worked with your product, and again, another reason I was interested to speak to you was it's, you know, Supermetrics is a product that I see in almost every customer site I go to, really, where they're kind of evolving from maybe being a startup. And that's the first data source they would use.

And they'd use Supermetrics to bring that in to then using your products to bring in, say, Facebook ads and so on. So maybe again, just paint a picture of what, I know there's quite a few products that you have, but what broadly is your coverage in terms of kind of data sources and I suppose this sort of jobs to be done that you support as well? Yeah, so we cover a wide variety of marketing data sources. So advertising platforms such as Facebook ads, Google ads, Bing ads, Twitter, LinkedIn, and many others. Then we have web analytics tools, GA obviously, we have Adobe. Then we have databases of various types. We have Salesforce, Stripe, PayPal, and various other sources, but mainly with a marketing focus. Okay. And who would be the typical, I suppose, persona of someone who'd use your products? Who would be the kind of person you mainly cater for or envisages your customer that you work with?

Yeah, so it's gradually evolving, but traditionally it's been, the tool has been, so if I start with where the tool has started out as. So we've been obviously focused on spreadsheets for a long time. So the company started out with getting data into Excel. But since then, we created a Google Sheets add-on that has become the main product for the company and product of which we are best known all around. So getting data into spreadsheets, which we see that a lot of marketers want to do because nothing really beats a spreadsheet when you want to analyze data and combine it in a very flexible ways.

So that's been a really, really good niche for us, getting the data in the spreadsheet. And in that market, it's usually an analyst or the marketer himself or herself that is purchasing the tool and then using the tool. Okay. So how was, I mean, if you're building a product for that kind of audience how how do you think about i mean you say you deliver it as an add-in into

i know there's quite a few products you do and it'd be quite good to talk about some of the other ones later on but when you when you're thinking i mean the way i came across super metrics was i was looking in the the g suite sort of uh marketplace and i saw it listed there amongst the other kind of analytics tools and analytics sources um you know so is that is that your primary kind of I suppose channel for selling or how do you go about you know selling and building products for marketers really? Yeah so obviously when the company started out like I said Google blogged about the solution many years ago that gave me the initial traction. And since then, Google has been really, really helpful. So we have a tight relationship with the Google Analytics, Google Data Studio, and also the

BigQuery teams nowadays. So they provide a lot of visibility for us. And that's been important for getting people to use our products, both through their social media their blogs and their application galleries but on top of that we do a lot of content and influencer marketing as well okay okay and i suppose in terms of the product it needs to be something that is i wouldn't say non-technical because i mean most marketers i i know are are competent with the tools that you use there but presumably it has to be a fairly kind of turnkey solution really and something that is focused on like immediate value I guess really. Yeah definitely so myself I'm not by background a technical person so how I try to design the product is for a person like me that is working with a lot of marketing data

uh it's not really a coder uh it's somewhat somewhat like technical skills can can do a lot of stuff in a spreadsheet for example but but uh but still not not really a really a developer um and that's always been been my aim with the products to make them simple for normal marketers to use without the help of any IT guys. Okay, okay. And you've managed to grow. I mean, I was looking on Crunchbase actually before speaking to you, and I think you're one of the most successful Finnish startups and so on.

What kind of size have you got to really, and how big is the team and that sort of thing, the number of customers and that sort of thing yeah so we are approaching 50 people in the team uh we now have offices in in finland where we are where we are we are coming from and we recently opened a development office in lithuania and we are looking into opening in the US later this year. We recently passed 10 million in annual recurring revenue. We have thousands and thousands of companies all around the world using the products. Half of them are marketing agencies

that use them to analyze the campaigns they do for their clients and optimize those campaigns. And then the other half are direct customers such as Warner Bros., Dyson. Actually, Google itself is a client as well. And many other big names. Okay. That's fantastic. I mean, so your products have evolved over time but so you started off with you say excel and it's in google sheets now and that so on i mean and then you've i think there's some support you've got for google data studio as well i mean

again you know i think google data studio is often the kind of bi tool that is the biggest one that no one's heard of within the kind of the the old school world um uh why why did you focus on on maybe to explain what Google Data Studio is and what did you do with that product, you know, to bring Supermetrics to that? Yeah, so Google Data Studio is Google's BI dashboarding tool, pretty new one. So I think they released it or made it generally available like two years ago or something. So it's a pretty new thing.

Feature-wise, it's not quite on par yet with some of the established players, but what we see is people really love the ease of use and all of the sharing

options there. So it is gaining, I would say, a lot of users and maybe at the expense of some of the existing players on the market, because it's very easy to use. And I think also one aspect has to be that it's free. So that's definitely part of it, but it's not the only thing. It's also a really, really good tool to use. And we also use it here at Supermetrics ourselves.

And yeah, so a couple of years ago, then Google, they had already released Data Studio and then they contacted us and said that, you know, we are going to open up a data studio for data integration companies like Supermetrics. So I asked if we would like to be the first company to develop connectors for bringing non-Google data into Google Data Studio. And, yeah, of course, we took that opportunity and started working with Google on that and then released that functionality in August 2017.

We were their launch partner with the connectors, along with a few other companies. But I think we had the vast majority of the connectors along with a few other companies but uh i think we had the vast majority of the connectors available and i think we still have a uh the strongest uh strongest uh suit of connectors uh for that product and okay okay so so so how i mean just i mean i know you obviously you're not you're not the technical person behind it and so on but how do that i mean in broad in broad terms how do the connectors work what do you i mean they connect via an api for example to to the various data sources and and they come back is

it via batch i mean in broad terms how does that how do your connectors work really yeah so actually i am i am the technical connectors i have to say uh so that was still, I would say, the last product where I was really, really hands-on with the coding. Obviously, now we have many more talented people in our team to do that. So Data Studio has this certain framework for connectors. They let us specify a config to show to the users what kind of selections the users need to make in order to get that connector working.

Then we tell Google Data Studio what is the schema for the connector. So we have on our side a list of fields that the connector supports. We give that to Data Studio. And then Data Studio lets users, when they do reporting, they pick fields out of that schema. And they give us a get data request that contains these fields and various other parameters, such as the date range that the user has selected. Then we make a request with these parameters to the source API, Facebook ads, for example, get the data,

then process that data in our backend into a standard format, and then return that data for Data Studio to display for the user. Presumably you've got tens of thousands or more of customers doing this at the same time. How do you handle doing this at the scale that you do it at? Yeah, it can be a challenge. Obviously many of these data sources have rate limits, quotas, various things that make it difficult to scale

these connectors when you start getting a lot of users and we have put a lot of effort on handling these massive amounts of users so we do a lot of caching on various levels to reduce the usage of the

source APIs to avoid going over these rate limits. And we also have, like, we are trying to stagger the queries so that not all of the queries are fired at the same time and things like that. And obviously, we do have a very good monitoring of our connectors in place. So if there are any issues, we can react immediately. Okay. And again, with the basic connectors and Google's Data Studio ones, how do people go about maybe trying to stitching together maybe sort of data from, say, Facebook ads

and from, I suppose, AdWords and so on? Is that something they do themselves or is there anything you can do to help with that at all to try and create a kind of, I suppose, a unified view of campaigns? So there are several ways how people can do that. Data Studio nowadays has a data blending option, so you can make one query with our Google Ads connector, one with our Facebook Ads connector,

and then use Google Data Studio's data blending option to merge that data together. That data blending option does have its limits. So what many people actually do is they use our Google Sheets connector to get all of that data into Google Sheets.

They're merged that in whatever way they want. Maybe they do other processing on that as well maybe they even have imports from sources that are not super metrics and merge that data as well and then in google data studio they use the google sheets import to get that data from google sheets into data studio okay okay um so so that that really is a good lead into i suppose the thing that got me again interested

in speaking to you which was um i i yeah i heard i noticed that there was um that actually uh your products or your connectors have been featured uh are now featured in the uh in the uh google cloud marketplace uh actually now fully fledged connectors for loading data into into bigquery which is which is another kind of stage on really from what we've been talking about. I mean, just maybe just tell us what that's all about and broadly, what's the kind of value proposition really with those connectors? So this is a very, very interesting new area for us.

So like I said, the company started from spreadsheets, then we expanded into Google Data Studio, and we have connectors for Tableau and Power BI and the other BI tools as well. But now for the first time we are also bringing this data into a data warehouse, into Google BigQuery. So we were Google's launch partner

with a new data transfer service product that they launched at Google Cloud Next just a few weeks ago in San Francisco. What this service does is that it allows for very easy import of data from various sources into Google BigQuery. So with just a few clicks, you can set up a transfer from, say, Facebook ads.

And then that transfer will run every day to get your latest data into Google BigQuery. So you can get a marketing data warehouse up and running in just a few minutes. And that's pretty cool. Okay. Okay. just a few minutes and that's that's pretty pretty cool okay okay so so the actual i mean the technology behind that is that is is it the same same technology you're using for for bringing data into say google sheets or or have you kind of hooked into a kind of google pipeline for doing this i mean is it any different at all so the underlying connectors to the source apis all of our products use the same same uh backend that has

the uh that has these high quality connectors that we have developed so that is the same uh but then obviously the technology of getting that data then into big query that that is uh completely new and it uses this new google data transfer service that they opened up now for third-party developers such as us. Okay, okay. So, I had a quick look at the marketplace again recently, and there's connectors for yourselves, there's connectors from Fivetran as well, and so on. And in a lot of cases, well, no, in some cases, there's some, not duplication, but

both companies, other companies will cover those as well and but it struck me that you're with your connectors you're maybe again serving a slightly different type of user or a slightly different type of market to to um the ones that say other companies connectors are aimed at i mean maybe just tell us about that really and how that's influenced the design of the product what What we did here, so there are a few things I would say that differentiate us from some of the other companies. Obviously, there are other companies like Fivetran who have been in the data warehousing, marketing data warehousing market for many years,

and we are like newcomers to the market. And these companies, they are good companies, and I think they serve certain purposes very well. But I think there's still an opening there for someone like us to do some things a bit differently to make things easier for a certain segment of users. So first of all, I would say the ease of setup. So when you use our connectors with BigQuery, you can handle the setup fully inside the BigQuery UI. You don't need to go to external tool to register or configure anything. You can set up everything inside BigQuery. And it literally is just a few clicks to do that.

So you can get the Facebook ads transfer running in two minutes. Then we have the really high quality connectors to all of these marketing data sources. So we are coming from the marketing world and we really know this data well, and we are known for our high quality connectors. So anyone can say they have a Facebook ads connector if they bring five metrics and split by campaign. But we really aim to get all of the data that is available in these sources. So we really aim to create high quality connectors and also update them whenever the APIs update to make all the recent new features available to our users. And then some things on how we actually bring the data into BigQuery.

So what we wanted to do here is make it really easy to do reporting on that data. And what we see many other companies do is they come maybe from a more traditional data warehousing world, and they aim for these normalized schemas. And that's not really the easiest thing for an analyst to use in a reporting tool such as Google Data Studio.

For instance, in Google Data Studio, if you use BigQuery data there, you would generally select one table for reporting and then Data Studio would show you all the columns in that table as fields. But with a normalized database schema, that doesn't really work very well because most of the data is not in one table.

It's spread out all over and you need to do joins across those tables to be able to really do good reporting. So those normal schemas are not really well suited for these BI tools, or at least not for marketers who don't know SQL. Of course, if you know how to write SQL, then you can do the joins, and then that kind of schema will work for you. But most marketers, they don't know. So what we wanted to do is we make the normal schemas, huge tables which contain a lot of data.

And then it's really easy to use that data in the reporting tools. So the marketers don't need to join between tables. They don't need to write SQL and they don't need to bother developers on helping out with reporting. Yeah, definitely. I mean, yeah, totally agree. I mean, so my experience with Data Studio was, as you say, you point it towards a specific table. Now, you could obviously pre-join dimension tables to that big table, and then you could therefore report against that. But then you hit kind of issues where you're joining to, you know, or you're bringing a lot more data in than you would expect to really.

And do you create a view for every potential combination of kind of dimensions and facts? And it gets complicated a lot quicker than you think it's going to be. And I think the other thing that struck me was I brought in Google Search Console data through your connector. And I brought it through actually Console data through your connector, and I brought it through actually a Stitch one I tried it with as well. And granted, the Stitch one was through the Singer project, so it was an open-source one that was community-supported. But it was interesting seeing the different, I suppose,

different ways that the connectors bring data in. So something like Stitch would often bring the data in, and you'd have to then go and use things like unnest commands to actually unnest nested columns into something you can kind of work with, which if you know SQL and if you know BigQuery, it just takes you two seconds to do. But it's a kind of a level of complexity that will confuse people. And then you get, say, maybe sort of like, I don't know,

the five trends of this world where they would have dealt with the unnesting for you, but it would be a normalized schema. And your one is you go for the big wide schema, really, which suits Google Data Studio. And it also suits BigQuery, which is typically, you know, best at doing aggregating and projecting queries. And that's the main thing it aims for. So it struck me it was a well kind of aligned solution with big query and it's

a well-aligned solution with the market you were aiming for really yeah definitely and as you said this is the way that google says big query should be used with the normal schemas not doing joins across tables and that's what we went for so it's both easiest to use for the on the reporting side and it's the most performant. Yeah, and I think the last thing really I noticed was the quality of, I suppose, the table metadata as well. So something, again, something that if you're in news digital marketing, or if you're even just not as absorbed in it as kind of you would be or maybe I would be, understanding the meaning of various metrics is important.

And again, you've gone quite a way haven't you i think with actually putting a lot of very meaningful and and um and useful table comments and metadata in there as well yeah yeah we really uh have tried to we don't have descriptions for all of the columns but we are really uh trying to to cover all of them uh so to make it very easy to understand the data. And I think the client feedback we have gotten so far has been really, really positive, saying that or really appreciating the way it's made it easy to work with the data.

So do you think this is going to, are you still in the end aiming at the same types of users and customers or is this kind of broadening a little bit the market you think you're going to be serving? Well, we hope that it is broadening it definitely so obviously there's part of our current client base that is doing things in google sheets that really should not be done in spreadsheets so they are really stretching the limits of a spreadsheet and previously we didn't really have a better alternative for them. But now we do. So we can say, okay, use Google Sheets, no, Google BigQuery to get the data into Google BigQuery.

And actually, one interesting thing that Google also launched at Cloud Next is connected sheets. So you can actually have the data in BigQuery, but you can still use the data in a spreadsheet in Google Sheets. So you can put the formulas and visualizations and whatever in a spreadsheet while the data is still in BigQuery. So that's pretty interesting. Interesting. Another thing that was launched was the BI Engine as well. And I think there's maybe kind of very happy confluence of things there because you've got Google Data Studio, which is the kind of first front end for bi engine and potentially bringing your data in as well i mean is is that something that you think is is interesting and potentially

could be interesting to customers to use alongside your data as well yeah definitely so what's also interesting with our solution uh for for big query is that because we are originally from the reporting side, also with BigQuery, we didn't only focus on getting the data into BigQuery. We also thought about the reporting side, and we actually created a Data Studio connector that is dedicated to working with the schemas that we import into BigQuery. So that's, I think, a pretty interesting connector. So what it does is you select BigQuery dataset, and it reads through that dataset. It sees all the tables there. Then it, because we understand that table naming convention and everything, it sees which data sources have been uploaded to that data set.

And then what it can do is, in Data Studio, when you then look at the schema, you will actually get not the column names from BigQuery, but we fetch the friendly names of these columns from our backend. So you would have nice looking names for all of the columns. Then we can set the data types for all the fields such as, okay, this is country, this is city, this is date and all of that stuff.

So this kind of metadata layer is handled for you. And even more, we can add calculated metrics to the connector. So we see, okay, in BigQuery, you have clicks and impressions in this table. So we'll add a calculated metric for click-through rate. So you don't need to define all of these various calculated metrics yourself but we bring them through the connector automatically and that's quite a big benefit because in many of these data sources you would have a really large large number of calculated metrics actually that you would not have directly in the in the

database but you would need to directly in the database, but you would need to define in the reporting side. So we do all of that for the user. And I think even more interesting for a person who doesn't write SQL is that our connector can automatically join or union between these different tables across data sources. So there are some fields that are marked as join fields. So we see that this field called campaign name, for example, is available in both Google ads and Facebook ads. So then if you select campaign name and then you select a metric

that is also available in both of these, our connector will automatically do a union query to merge that data from these different data sources together. So you get really easy cross-platform reporting in Data Studio without needing to understand SQL at all. That's fantastic. That's really good. That's really good. So, I mean, just to kind of round things off then, how would people, how would potentially interested people get hold of these connectors, your products, you know, what's the first place they'd go to to find out more details, really? Yeah, so

I think our website, so supermetrics.com, there you can find products, and there you see Supermetrics for BigQuery, if that is the product you're interested in. Okay, okay. And, I i mean you know is is it in terms of next things to be solved and other areas you're looking at is there anything else you can share about what you're looking at in the future as well in this kind of area yep so we are of course we are all the time building new connectors so that's a very obvious one, I would say. There are literally like thousands of

different marketing data sources that people are asking us to connect to. And we are prioritizing that list and going through and building these connectors all the time. So that's definitely one. We are also looking at expanding more into connectors beyond marketing, so into financial accounting data, HR, customer service, and various other data sources like that. And then in the data warehousing field, obviously now we have a product for Google BigQuery, but we are looking at Redshift and Snowflake and others as well in the future. Fantastic. That's really good. Well, Mikael, it's been fantastic speaking to you. And it's been great to speak to, I suppose, the person behind the product I see so many times in

customer sites I visit, really. And it's great news about the Connected to BigQuery as well. So thank you very much for coming on and telling us about it really and telling us a bit of the story behind you and the company. Yeah, thanks for having me. It was really nice being here. Thank you.