Podcast

Analytics technology selection priorities: Innovation & customization

Are you the debating the advantages of Amazon Redshift vs. Google BigQuery? At Callahan, we’re making the move from Redshift to BigQuery as we release our Intelligence Platform 2.0.

Stick with us as we explain how this shift will greatly impact future marketing insights. We’re taking advantage of new technology that will allow us to customize, innovate and become more flexible with big data.

Learn how we select marketing analytics technology so that your process is better informed.

Listen here (or subscribe on iTunesStitcherGoogle PlayGoogle PodcastsPocket Casts or your favorite podcast service):

Welcome to Callahan’s Uncovering Aha! podcast. We talk about a range of topics for marketing decision-makers, with a special focus on how to uncover insights in data to drive brand strategy and inspire creativity. Featuring Jan-Eric Anderson and Zack Pike.

Jan-Eric:
Hi, I’m Jan-Eric Anderson, head of strategy at Callahan.

Zack:
And I’m Zack Pike, head of data at Callahan.

Jan-Eric:
Today on the podcast we’re gonna be revisiting a topic that we discussed on a previous episode, when we talked about the intelligence platform we’ve established at Callahan. We talked on this prior podcast about the technology that was used to develop it, and why, kind of what those components are, for any listeners who are interested in developing their own.

Today we’re gonna revisit that topic because as is the way of the world, things are changing and things are updated, and we’re making some changes. So before we get into some of the specifics of the change, of what the changes that we’re making are, Zack, it probably makes sense to just talk a little bit about what’s driving the need for change. Why change?

Zack:
Yeah, yeah, so it … gosh, I can’t remember when we talked about this last. It’s been several months, and we’ve had our current tech stack in place for about a year, roughly. And just as a refresher, it’s Amazon Redshift, it’s Tableau Software, there’s a partner we use to process data into Redshift as our core database, it’s a really nice system.

Jan-Eric:
Tableau for visualization.

Zack:
Right. Exactly. And it’s treated us really well, but as with anything, you want to innovate and you want to get better, and we’ve got some ideas for 2019 — and even further out — that we need a little bit different of a stack to that technology, but we also need some new items inside of there.

Zack:
And you know, a couple … so if we rewind six months ago, a big problem of ours was just processing all of the data that we had access to. All of the Facebook data, and all of the media data that we’re using, as well as all of our clients’ sales information, we had to have a place to store all that. And now we’re really good at it. And now it’s time to figure out how we use that data for better insights but also to predict what could happen in the future.

Jan-Eric:
So it’s a little bit of a new technology becoming available but also kind of changing need of what we’re trying to do.

Zack:
Right, right. So it probably makes sense just to kind of talk about some of the parts and pieces that are changing.

Jan-Eric:
Yeah, so what’s changing?

Zack:
Yeah, so the biggest thing is we are moving from Amazon Redshift as our core database to the Google Cloud platform. And the reason for that, and this is highly applicable to anyone who’s struggling between these two solutions, or many of the others, is that Google is doing a lot to make prediction, machine learning, and really just handling massive loads of data easier. Amazon is doing a lot here too, but from a marketer’s perspective, from the types of stuff that we’re doing, I feel better running on Google Cloud. So all of our data will be sitting in a platform called Google BigQuery. It’s a platform that will handle anything we can throw in it.

For example, we’ve got some weather data that are hundreds and hundreds of millions of rows of data, and you’ve got to have a good platform to do that in. Everything is out in the Cloud, safe, secure, everything we could ever hope for.

That affords us some additional benefits. So, from a visualization standpoint, nothing is changing. Tableau is still our solution, we’re still using Tableau Online for all of our clients and consumer reporting. But by us moving to Google BigQuery, we are taking over control of much of our data infrastructure that we didn’t have control of before because of the way things were organized and set up. And what that’s gonna allow us to do in 2019 is work on some of these more innovative approaches to the way we want to do analytics.

I think we’ve probably talked on this podcast before, this idea of a perpetual testing environment, right? The whole PTE thing, which is still in my opinion one of the most powerful things a brand or a partner to a brand can be doing. The more testing we do, the smarter we get. And the easier it is to predict what’s gonna come in the future.
Testing is hard, though. And measuring testing is hard. And it’s often very time intensive. But I think there’s some ways we can get around some of those time limitations, get into an environment where we’re perpetually testing, and build a platform that can help us understand what we should be testing. That can alert us to variances in the data, that would be good testing opportunities, or risks that we need to be aware of.

Jan-Eric:
Yeah, and we get into these conversations with clients, and it’s never hard to come up with ideas of things that we’ve love to test. Or, “Man I wish I knew that. We can test for that, right?” And it’s almost become clichéd. “Yeah, we should test that.”

It’s very easy to say we should test something, it’s much more difficult to put an actionable test in place, that will give … with a reporting plan on the back end, or measurement plan on the back end, that’s going to give you true, actual insights. And there’s time involved with that, and we’ve experienced that. There’s also, when you’ve got multiple clients, or multiple companies that want to test different types of things, everything almost has to be customized, so you need the flexibility, right? And in that sense, you are innovating within your testing, because you’re kind of creating a different thing every single time.

So by having a little bit … with these changes within the intelligence platform, essentially we are establishing a bit more control to innovate, and tweak, and customize, what we’re actually trying to do. And this idea of perpetual testing environment, is that the idea is that once you get some learning, you’re turning around and applying it again in a different way, to pressure test those insights, and to see if it’s scalable, or can expand, and really trying to do that in real time.

Zack:
Yup, yup, yeah, flexibility is so important, and we’ve really learned that. The environment that we were in previously wasn’t as flexible as we needed to be moving forward. Which was a bit of a saving grace then, ’cause we had so much work we were just trying to do, we couldn’t worry as much about the technology. The team’s gotten bigger now, we’ve got a little bit of breathing room, where we can think more strategically, and having that control’s gonna be really valuable.

Another big reason that we made this move, so we also moved partners for the ETL side of our stack, and that’s extract, transfer, and load — that’s movement of data. Out of the source system and into the database where we can actually use it. We moved to a partner called Fivetran. And Fivetran is laser focused on moving data, from a data perspective. A lot of the companies out there are focused on moving data for a marketing people, and that’s good, if you’re kind of a marketing analyst, at your cube, you don’t have a lot of resources. But when you’re trying to do things that are more advanced, you need deeper access to even the data in the source system itself.

Really good example of that, if anybody listening is using a tool like Marketo or Adobe Analytics, both of those tools hold an enormous amount of data. If you have a large lead database, or you’re supporting a sales team with leads through Marketo, and into Salesforce or something, the amount of data Marketo produces is insane, on what they can track. The problem is they don’t make it all available on the interface, so it’s hard to really get the value out of that data.

This partner, called Fivetran, specializes in that world. They specialize in getting raw data out to people like me who can then massage it, and tweak it, and organize it in the way that we need to use it for marketing. And some of the early results that we’ve seen, even just with Marketo and Adobe data, is pretty exciting. The level of insight you get into your actual lead pipeline is better than what I’ve seen out of other tools.

We’re getting out of the world of looking at tactical metrics, like opens and sends and clicks, to okay, well what’s happening with people? How many people opened my email last week, and where are those people at in the sales funnel? And am I increasing or decreasing velocity? Am I increasing or decreasing conversion rate to a customer? You get to be able to answer those questions. It’s hard to do if you’re not at that level of detail.

Jan-Eric:
Yeah, well it seems to me that the changes are really delivering two big benefits. One is we have expanded our ability to gather data.

Zack:
Mm-hmm (affirmative)

Jan-Eric:
Gather and store data. The second big benefit is that we’re able to create action, and measure actionable hypothesis that comes out of these insights. So, seems like out of the gates, the focus was really on, hey the appetite is for insight, I want to get data to gain insights that … into my business, what’s driving my business. That’s great for a while, until you start to get antsy about wishing that you could put that insight into action, and then what happened from that. So it seems like we’re kind of greasing the track, to be able to get more data and insight. So not sacrificing that, in essence, expanding that, but also then really putting ourselves in a position to put it into action, and monitor the new tests.

Zack:
Yeah, and like with almost anything … I mean, stuff gets stale. In the analytic space especially, the stuff a year ago that was really exciting to us isn’t as exciting anymore. It just … we know all of that information now. And in fact, we’re even questioning some of the analysis we did a year ago, and if it was the right way to do it, and if we could have done it better, and you know, you’ve always got to be thinking and moving and changing, and really innovating. And that’s the … you know, the analytics organizations that don’t do that, they get left behind. And they create processes, and we sometimes have to clean up those processes, even today, but they create processes that put you in a world where you don’t necessarily even know what is possible. And so yeah, to your point, all of that stuff is gonna be really exciting, and my personal hope is we’re doing … you know, the focus of our work in 2019 is on predicting results, rather than analyzing what had happened.

That analysis will enhance those predictions, but we should be helping people make big decisions that don’t have a clear answer, and the way you do that is by predicting what you think is gonna happen.

Jan-Eric:
Well, and we’re certainly already seeing that, aren’t we? I mean, we were, prior to sitting down to record this podcast, we’re working on a project for a client where predictive analytics is critical, and projecting the impact of planned tests is critical to their decision making process, so we’re actually already seeing it in action.

Zack:
Yeah, absolutely, yup.

Jan-Eric:
Excuse me. Well, any other thoughts before we wrap up, just about this, and what the future holds?

Zack:
No, I think for any analytics person listening, any CMO or marketing director or VP of marketing, what you were doing a year ago in analytics should be different than it is today. And a year from now, it should be different than even that. And this industry is moving extremely quickly, so don’t be afraid to try different technologies. The scary thing about analytics is there is no playbook for what we’re doing. No one knows what they’re doing, so everyone is trying new stuff, so don’t be afraid to jump into something that is kind of unknown, uncharted territory, and try it, and if it doesn’t work, you can always go back.

Jan-Eric:
Well said. Zack, thanks for your time today, and we’ll see you again on another episode.

 

Thanks for listening to our Uncovering Aha Podcast. Callahan provides data savvy strategy, and inspired creativity for national consumer brands. Visit us at Callahan.Agency to learn more.