Episode 204 Featuring Alex Bond

Pragmatic Data Science with Matt Ranta

Pragmatic Data Science with Matt Ranta

Matt Ranta is a Partner and the Head of Practice for Nimble Gravity, which provides data driven digital strategies for eCommerce brands. On today's episode, we discuss pragmatic data science, implementing data driven solutions, competitive intelligence tools, and much more.

 

What is Nimble Gravity

Matt Ranta: So Nimble Gravity is a consultancy and we're passionate about the power of data, and we apply that across several different areas. So those would be data science, as you mentioned, e-commerce and digital transformation, which I'm the head of practice for at Nimble Gravity.

And then we also have analytics and BI, global engineering and strategy. And global engineering, just to clarify for folks is what we like to think of as nearshoring or offshoring as people have traditionally framed that up. 

But instead of focusing on whether that's in one part of the world or another, we focus on where the talent for that job actually is, and then try to geographically locate that as close to the client as possible. So that's what we mean when we say global engineering. 

Pragmatic data science

Alex Bond: So your website defines the company as using pragmatic data science. What is that? Do you mind explaining what that means to anyone that might not know? 

Matt Ranta: Well, pragmatic marketing, pragmatic data science pragmatism in general is just an approach that's sensible and measured, and that is tracked and built upon as you go along.

Right. So it's not dynamic, it's not crazy. It's not off the wall. It's not inconsistent complex. It's measured and it's a very sequential process that one would go through and building on experimentation and building on testing and understanding learnings and where you have failed and succeeded, and then following those successes and expanding on those, right.

And making sure that you are doing more testing in that area, that you're doing more marketing, that looks in that fashion, that you are taking that successful data science experiment and turning it into actual insights that are being used by the business, whatever it might be. 

Alex Bond: No, that's great. So you're finding different ways to be, make that stuff useful, you know, so you help extract. Relevant business data via testing and pilot projects that are typically, I think your website says completed in 90 days or less. 

Can you peel back the curtain and tell us kind of what that process looks like? What's kind of the data you're looking for? What are you trying to learn from it?

Matt Ranta: Yeah. It can be highly varied in what we're actually looking for, and it depends upon the type of engagement that we're looking at. So we might be doing something like due diligence for a merger or an acquisition where we're looking at a business from the outside trying to understand how they're operating things, and that's typically a very fast process.

And we'll look at the totality of their e-commerce operations. We might look at how they're, you know, doing their social media marketing, their paid search any number of things, and then present opportunities to the business that's looking to acquire or merge with somebody as a target where they can drive significant growth and impact when they come in and build a model that, you know, it takes a couple weeks to do something like that, you then, I also have somebody that you're engaged with deeply as a client, right?

And say for instance, you're running AB testing inside of their organization. In that instance, you're gonna come up with a series of hypotheses that are gonna become tests. Maybe we should use this type of picture. Maybe we should have a different call to action. Maybe we should put a call to action button in a different place than it is.

Maybe we should have a social proof statement about customers who have also bought this and how they actually feel about that. And then you start to iterate on that and you go, Actually execute those things and put them into play inside of an AB testing environment and prove that out for client.

Right? It could be something that is even more sophisticated looking at launching an entire website redoing an entire platform and having a digital transformation moving from one technology set to another. And sometimes those will extend quite a bit longer. Typically for, you know, a lot of our engagements, they are done in sprints like you would do for a development cycle, right?

When you're writing software or code, and we're looking at creating an iterative process where you're constantly building on those things one after another. So you're aligned with development, you're aligned with those teams that are rolling out code changes to websites, and. Doing that from a testing perspective as well and syncing up with them and working on those iterative cycles.

Alex Bond: So do you guys help with implementing these recommended programs and services or is your team really more responsible for the strategy part of things? 

Matt Ranta: Yeah, I think more often than not, we wind up being fairly responsible for the strategy. Okay. Versus the implementation. But we do both, and it really depends on what the client is looking for.

In a lot of instances, we'll have top tier private equity companies that are looking for recommendations about a business, or will have people who are just launching their first business, they're founders and they're looking for, you know, growth plans. How they can look at the future couple years of their coming product.

Then if you're talking about somebody who's looking at, they're an actual business, they're operating, they're running AB testing. In those kinds of scenarios, we are actually doing executional work and have staff embedded in the systems that those companies are using on a day in, day out basis. 

Alex Bond: And that would make sense. You'd hate to give people these wonderful strategies that you've developed based on the all this research that you've done, and then they kind of dropped the ball. After all that time and effort. So I could imagine you wanting to see that to completion.

Matt Ranta: Yeah. We really do like to build things out to an initial stage. We're not necessarily gonna be operators of a solution for a very long term period. Years on end. But we love nothing more than to build a strategy, come in and help the customer set that up, execute it, get it up and running, and then really kind of train them how to do what we've done and, and hand over the keys to their new car, so to speak.

Implementing AI generated recommendations

Alex Bond: And part of these recommendations that you've explained in the past is how you implement AI generated recommendations directly into the line of business for better and faster decision making, does that take away the ability of creative problem solving or, you know, how does that exactly work properly?

Matt Ranta: Well when you're talking about AI solutions you're not necessarily. Degrading the creative process in any way, shape or form. I think you could look at what's going on right now with the, you know, prevalence of information about chatGPT and how everybody's talking about that and thinks wonderfully of it.

And it's an amazing technology and it totally is right. But if you and I go and ask it the same question, if a hundred people go and ask it the same question. Is it giving a hundred unique answers or is it giving a hundred pretty derivative answers of one another? And are they're highly similar?

And so is that really the creative part of the process or is the creative part of the process thinking up how you're going to phrase an instruction set, how you're gonna phrase a question, how you're going to tune training, how you're going to set up these models? And then do those inputs that get into it and make sure that those are really where the creative part of the process is coming and, and going with that along, along the way.

So I would argue that when you're starting to put, you know, AI into the conversation there's still a high value of creativity upfront. And maybe just shifts where the creativity is happen. Versus more traditional methodologies. 

Alex Bond: And to build on kind of what you're saying, Matt, is a little bit of what I'm hearing is when you get the take that extra time instead of having to delicately figure out all of these different solutions, and instead you have an AI doing that, you can spend your time more properly to develop creative solutions in an in another way instead of having to. Deal with, say, copywriting or something like that.

Matt Ranta: Sure. Yeah. You can do that. You know, I would caution people against utilizing the direct outputs of say, copy generating AI. Like chatGPT or Jasper or copy.ai for the reasons I just mentioned. You're still always gonna want to go and look at those and say, is this in my voice? Is this phrase the way that I would want it to be done?

Does this actually make sense? Would I make these recommendations? Right. You still have to review it. I don't think that you're ever going to find a scenario where you should just, and I'm using that word, should very specifically, where you should just release that without those kinds of checks and balances.

Obviously you could make that choice, but I personally don't think that that's ultimately gonna be the best use of either those systems or the best practice for people as we move forward in time. 

Toxic backlinks

Matt Ranta: So toxic back links are like bad friends, right? So your website has a reputation and your reputation is based upon the people that are referring links to you. And thereby referring traffic to you, right? 

And so if you have a poor reputation as a website, you're a link farm, you're a spammer you don't have any value, your content is duplicative, if not downright plagiarized. And you're a bad domain, so to speak, but you're referring a link and thereby traffic to somebody else's e-commerce operation, a good domain.

Then what happens is the search engines look at all of that and they say, oh, well that website's kind of hanging out with some bad actors, and we want that. We don't trust that. And so what you can do is you can go into Google Search Console or other webmaster tools and utilize those to then disavow the connection between your two sites.

There's plenty of tutorials out there. You can look up how to disavow back links and Google Search console and you'll, you'll find a hundred. You know how to step throughs on that. But it's a very simple process. You basically just go and say, we don't want, you know, that link to be credited into our profile essentially. And thereby you're kind of removing that bad reputational connection from your SEO profile. 

And I'd highly recommend that people do this on the regular. It's just kind of one of those administrative tasks that you could, you know, sign up on your calendar for hey, every Monday morning, I'm gonna go check out what bad back links I have and disavow 'em and be done with it. And then tools like semrush and others will actually even help you with that process, and they can automatically kind of point those out to you and you can go and find them and take care of 'em. 

Alex Bond: Is that something that you found in doing, you know, some data research and insight is that a lot of companies didn't know that they had these, didn't know how to solve it or, you know, the importance of it. Is that something that you have found personally in Nimble Gravity? 

Matt Ranta: Yeah. Sorry to stomp on your question there by talking over the top of you, but yeah, it absolutely is. I've seen this across hundreds of sites so I'm constantly doing research and, and looking at competitive sets.

When we're doing diligence work, and I'm seeing that along with many other things as being very common areas where people could be improving their performance. So the thing that you have to remember about SEO ranking factors is that there's. 200 or more factors that go into SEO ranking.

They include things like your authority score, the way you know keywords are used on the page, the speed of your website, any number of things. And one of those things is the toxicity of your domain and then how that kind of reputational rank goes. And when you're playing the game of how do I ranked better than my competitor.

You want to try to tackle any of those 200 things that you can, right? Some of them are more impactful than others, but if you and your competitor are all doing the same top 20, but they've given up after that, and you keep going and you do five more, you do 10 more, you do 50 more, your rankings are going to.

Go above theirs, right? And you're gonna get more traffic. So the top three SEO rankings on a webpage search result page, those are gonna get the most traffic, and that's where you want to be. And you're only gonna get there is. If you start executing across the board, not only on domain toxicity, but on things like core web, vitals, the speed of your site, all these kinds of things.

Using data in eCommerce

Matt Ranta: There's quite a few things, right? So first and foremost, you have to track what you're doing, right? So hopefully everybody that is in your audience is using at least a free analytics platform, free Google Analytics or something of that nature that gives them at least basic information about things like what pages did somebody exit my website from?

How long did people stay on this page? What's the conversion rate? What's the average order value? Right? Just the very basics of any e-commerce operation. You can start looking at that data and utilizing it to improve performance overall. Right? So let's take pages that people exit from. Okay. What are they? Why are you sending people there?

Are they in the middle of your shopping cart process, your checkout process that people are exiting from? Or are they articles that you wrote that people are finding and, and then ejecting from your website from, so, okay. Say it's an article. Are they spending time reading the article? Are they on that page for three or four minutes?

Maybe five minutes plus. Okay, great. They're ingesting that information. Maybe it's not a loss that they're exiting. Are they coming back at a later time? Okay. Perhaps they are great. You've given them some knowledge and they're coming back to you for either more or to checkout or whatever it might be on, on your website, and you have to kind of look at all of those things if it's in the middle of your checkout funnel.

My gosh. Yeah. Immediately go address what's going on with that page and, and figure out why people are ejecting there and, and what are the problems. Is it your shipping options page and all you have is put it on a slow boat from a far away country. Okay. Problem, right. Like you can kind of figure these things out and start to go after them. 

As you're getting deeper and deeper into those kinds of things, though, you can start to take all of that data that you've collected over the years and you can start to do modeling on top of that, whatever that modeling might be. You could look at causal modeling.

You could look at, you know, just doing regression analysis against data sets that exist for things like fuel prices or housing starts, or any number of things like that. And see if there are impactful data sets that exist outside of your operation that have high correlation and or potential causal impacts on what your business is doing?

So I've done that in the past. I've looked at both housing starts and fuel prices as they related to the major appliance market and what that did for the major appliance market and how that predicated. For a consumer electronics and appliance retailer in the state of Montana. So it's absolutely things that you can do, but the base level get in and start tracking and, and playing around and digging into these things.

Alex Bond
Host
Alex Bond

Meet Alex Bond—a seasoned multimedia producer with experience in television, music, podcasts, music videos, and advertising. Alex is a creative problem solver with a track record of overseeing high-quality media productions. He's a co-founder of the music production company Too Indecent, and he also hosted the podcast "Get in the Herd," which was voted "Best Local Podcast of 2020" by the Richmond Times-Dispatch in Virginia, USA.

Share post

Latest episodes