Web Analytics x Marketing, with Joshua Lauer | Sponsored by SearchMaster

Speaker A: Hello and welcome to the Marketing Times Analytics Podcast. I'm your host, Alex Sofranis, and today we're on with Joshua Lauer. Joshua, would you like to introduce yourself?

Speaker B: Yeah, how's it going? I'm Joshua Lauer. I am, um, CEO of Lauer Creations Incorporated, a marketing intelligence consulting company.

Speaker A: Interesting. Okay, yeah, tell us more about that.

Speaker B: Yeah, so, uh, yeah, basically I what my kind of main thing is any data related to anything you do with marketing. And if you've been around for any amount of time, you're probably spending on a variety of different platforms, you probably got data about your website, the web analytics, and you may or may not spend a lot of time sort of chasing your tail trying to pull that data together each month to understand how last month performed. And that's one part of it, right? Understanding how it performs, being able to adjust your strategy according to that. But what I do is I take all that data for you, put it into a data warehouse, and then essentially automate the reporting side of it so that each month when you're looking back, instead of scrambling to figure out where you spent all the money on all the various different campaigns, you're able to instead see a report of that and then dig into those details and find out really like more of what happened and less of. Let me collect all the details. So that allows you to focus more on the decisions, more on the strategy, and less on just the sort of stuff that we don't think is fun. Diving into Excel and all that other stuff.

Speaker A: Yeah. And do you use particular tools to aggregate this data?

Speaker B: Yeah, yeah, I specialize in Google Analytics. And with Google Analytics, there's a sort of one click integration with BigQuery so you can install it on your website. Get that running into BigQuery. And then there's a variety of different other third party solutions for ETL that can connect to things like Meta, Google, TikTok and all those other things to get those platforms also putting the data into BigQuery and then from there writing a handful of queries to aggregate the data the right way. And then we present it using Looker Data Studio. And within that we can do all sorts of things, make interactive dashboards, make reports that get emailed to you on a certain regular basis. Whatever it is, if you want to see it today, you want to log in and you want to see what happened yesterday, or you want to see what happened this week up until yesterday, or whatever, we can give you preset date ranges so that all the analyses are basically the snapshots that you want to look at?

Speaker A: Yeah. And do you do additional types of modeling work or is it mainly reporting?

Speaker B: There's some media mix modeling on some of the accounts that I've worked on. There's also a variety of other tools that because you have the data in Google Cloud platform, it opens up the table there for any kind of integrations with the, uh, LLMs most of the time. Media mix modeling some more. I think a lot of these tools today that marketers are using have all sorts of AI built in to an extent as well. But yeah, the opportunity for that is really opened up by the fact that you have all this other data in this data warehouse and it's easy then to finagle that into a format that machine learning can consume, if that makes sense.

Speaker A: Yeah. What would you say are some of the most important types of metrics that businesses are looking to report on in marketing right now?

Speaker B: I think it depends on your business. But if we're thinking, okay, you want to grow, so you're going to want to see leading indicators that you have more potential customers coming. And you're probably looking at the web sessions, the volume of traffic that's coming to your website each month. And if we're tracking kind of the main things, whether it's, if it's an E commerce site, it'd be purchases. If you've got more of a B2B thing where you want someone to fill out a form and you're going to get back to them and have more of a discussion, more of a sales pipeline, then we're going to make sure we have forms tracking. Essentially we're going to know how much traffic is coming to the site, where is it coming from? Is it coming from LinkedIn ads, Google Ads, organic search, whatever you happen to be doing, email. And then as you're getting these forms booked, we'll also understand not just, okay, this traffic came from these places, but which of those places actually ends up leading to the most forms. So which channels have the highest conversion rates? Alongside that, you probably have a bunch of different content on your website trying to speak to various different aspects of your business. And so we can also look at what pages did they land up and within the various pages that we're sending traffic, which of those are doing the best job at driving the conversions, whether it's again, form submissions or e commerce purchases and what have you. And so that's kind of a good starting point. Usually once we get to that point, we start looking at the data, we get a baseline for here's how it is today. That's where we can say we've collected data for a month or so, assuming it's a new project. And, and now we're looking at this and we're saying, does these numbers line up with like, basically what you're seeing within your business? And there may or may not be, yes it does, or no it doesn't. And so if it doesn't, there's more of a question of what's missing from the tracking. Are we seeing numbers that are too high? Does it seem like it's is inflated, or is it in fact that the numbers are too low? And so we think we're missing something so we can dig in and start to sort out what's going on, make sure that we got the tracking dialed in with the business. And once, once we are at that point and we've tweaked it to a point where we feel confident, then we really do have a little bit of time to collect more of a baseline and get familiar with things. And once we look at various data points, whether it's the sessions, it could also be, uh, what kind of devices are people using? All sorts of different things you can collect. At that point. There's probably now like, again, you've always got the business goals, so how is this now aligning to the business goals now that we know that this is tracking accurately and where within this data do we feel like we have more levers to pull in terms of if this number was bigger in middle of the funnel as an example, maybe that would lead to more leads or whatever. If everything converts the same way, if we just had more here, that would drive this kind of bottom line, if that makes sense.

Speaker A: Yeah, that makes a lot of sense. What types of businesses would you say are best suited to go with partner, uh, like yourself for analytics and this infrastructure work?

Speaker B: Yeah, I would say it makes sense for sure. Like you want to be operating your business, you don't have any questions about how the business works. Right. You've at least gotten to that point where you're up and running. You've probably been going long enough to have a website that's working on at some level at driving whatever activity is you're trying to drive, whether it's leads or purchases, and you're kind of that point where you maybe feel like you're starting to grow and you're not exactly sure, kind of there's so many things you can dabble in. Right. And you're starting to feel like your time's More precious and you're not totally sure which of these things is driving the most results. That's when you really want to get that data dialed in so you can make more effective decisions about how to spend your time in a way that's m more meaningful and impactful for your business and tends to be. Businesses that are 10 to 20 million-plus are going to probably be, uh, well enough established that they're looking for that extra. What's that thing that's going to get us to turn the next Corner sort of 10x again. Right. Less than that. I think companies are still more in a growth stage where they might still be figuring out some of the things. And not that I couldn't help a company in an earlier stage, but it uh, tends to be that like in those earlier stages there's more logistics, more things with the business you're trying to sort out. And as you get larger and you start to really kind of focus more on the growth aspect of it and asking questions about what's worked historically. And maybe you feel almost like you're plateauing and you're trying to figure out how do we keep that trend going that other direction.

Speaker A: Yeah. And when a company gets to that level, how do you approach the conversation of should we go with an external partner or an internal hire?

Speaker B: I think, uh, it's something that you probably intrinsically have a gut feeling that either we have this kind of resource internally that seems to be willing to dig into this stuff, or we feel you want to bring in an expert. And I think they're. There's plenty of times where you might actually have that guy on the team that's like, hey, this. I'm, uh, able to do a few Google searches. This sounds simple enough. I think I could probably implement this. And actually that's probably a scenario that's a potential leading indicator that's maybe you're uh, about to feel like you need to bring in a third party. Because a lot of times the story that I hear is something to the effect of, yeah, we've had a few people in here that kind of seem to dabble with this. We don't really know if they knew what they were doing, but they were the only people in house that had any idea what was going on. So we just kind of let them run with it. And now some other people have looked at it and we're like feeling like it seems like it's a little bit of a mess and we just need someone else to look at it to tell us, does this look like it's working correctly or is it just completely messed up? And some examples there. Typically you want to put tracking links on any URLs back to your site in any ad platforms, right? So you have UTM source equals Google and then medium equals CPC and campaign campaign equals the name of the campaign. It's an example. These are all just parameters that get appended to the end of the URL that you may or may not have noticed in the past anytime you click on various random links. And so that's one way to track traffic coming to the site. People learn about that when they're first new to Google Analytics and they'll be like, hey, I want to figure out if somebody's clicking this button on the homepage or this other button and they'll get this bright idea that they should just add the UTM parameters to that button on the page. But if you do that, and now somebody actually came from like a meta ads campaign, they come to your site, they click this button and it's, it's saying the source of traffic was the homepage button instead of the actual source. So you can, there's nuances like that where you can fragment the data and it can get messy and confusing pretty quick. And then you layer in things like the stuff that's changing with cookies and all sorts of browser changes and security updates, gdpr, ccpa, privacy policy type stuff. And pretty quickly it's consuming more time than you probably want to spend on the tracking. And you're more excited about the things that are maybe moving mountains, so to speak, on the business side that you'd rather deal with. And that's a great example of peace of mind. Right? I'm not going to crack open a my car and try to fix anything. I'm going to go hire the mechanic because I don't have the time to figure out all that stuff because I'm doing this.

Speaker A: Yeah, you mentioned such an amazing point about somebody thinking that adding a UTM for some internal value or internal button is going to help with tracking and it ends up inadvertently harming attribution. And I've definitely seen that happen in my career. What other pitfalls around attribution and just generally marketing data have you seen that businesses should be on the lookout for?

Speaker B: Yeah, so that's one. There's depending on the technology that you're using for your website, if you're using a tool like WordPress and you're using various plugins, they may or may not load content in a way that is helpful or hurtful to how things are tracked. So one common thing that happens is that form tool might do something where it handles the data through ajax, which that's what the developers would say, it's an AJAX form. And what happens there is that can be coded up in a way where the, the tools that we use to track the form submission can't actually detect that form is submitting. So somebody's gone in there and they think they've configured the form submission to track. But because this is doing something where it hijacks what the browser thinks is going to happen, then it causes some other things not to be able to see the forms being submitted downstream basically. And so then that can cause it to look like the form is broken. From a tracking perspective in a situation like that, I would work with a developer to make sure we have a data layer push at the exact moment that the form is submitting so that the sort of JavaScript that's hijacking the normal browser behavior can just say, oh hey, by the way, tracking tools, this just happened. So that's an example. Another thing is iframes. There's a lot of scenarios where an iframe might have the tracking information and you've got the main website outside of that. And certainly like with Shopify checkout extensibility is this thing now where there's pulling the pixels and the JavaScript out of the thank you pages on Shopify. And so that becomes another scenario where you get walled off environment, so to speak, that's like considered safe. As in no bad actors could get in there and inject anything, what's the word? Quarantine from the actual website in a way. And so like those things sound great for security, but there's scenarios where if the stuff is not configured the right way and the cookie information isn't passed around the right way, we could have the events actually tracking. But because it's missing a parameter or some kind of detail, it could just end up going to Google Analytics. And Google's cool. We got a purchase event here, no idea where that came from. So we're just going to say it's direct traffic. Whereas if we were able to make sure we're passing the right information when we send that, Google can see that it was associated with that organic search visit that just came a moment ago and tie those together.

Speaker A: Before we continue, a quick note from our sponsor, which is the Searchmaster app, uh, that I've developed. So in less than 30 seconds I'm going to show you how to run a AI competitive search analysis for your business. So you go to Searchmaster app and scroll down. You don't even have to log in. It'll show you your AI search rankings. Let's say your business is Rolex and you want to see are the AIs ranking you as the best watch company or is there somebody ahead of you? We can see that, ah, all of three of the AIs are ranking Rolex in the number one position with Patek Philippe at number two. So when people are asking about which watch they should get, you can be confident that the AIs are recommending you first. You can see a full list of competitors available at Searchmaster app. Ah, sign up today. And now back to the podcast. So I've heard pushback about analyzing that sort of stuff in Google and Analytics because of sampling for large data pool or data data sets. What's your general opinion on the sampling?

Speaker B: Yeah, so sampling is real if it depends on the reports. So if you're using some of the standard reports, there's less sampling happening. They have certain reports where they're running the queries and crunching the numbers, then saving that so they don't have to necessarily rerun all the stuff in the background. Every time you pull up that report, it's printed in the way it's aggregated. But then you decide to segment it. Then all of a sudden that's when it's, oh, we have to really run this query. Then depending on how complicated the query is, there may or may not be more or less sampling. So there's various different interface alerts within Google Analytics that tell you if you are dealing with sampling or if thresholding has been applied. And the concept of thresholding is a little different, but like similar as in say you have a, uh, website that has lots of different URLs. Like one example would be if you had like users logging in. So uh, they see accounts orders, wishlist, whatever. Let's just say there's 10 different pages you could see as a user. But because you have a hundred thousand different users and there's a, maybe a unique identifier in the URL, you'd have more than 50,000 combinations of URLs that Google Analytics might see. And so you'd hit like a high cardinality dimension at that point, which could cause what they call thresholding as far as we're not going to comb through all the possible pages. So with the ga4 to bigQuery integration though, everything that goes to ga4 gets sent to bigQuery. And you can write your own queries and there is no sampling. So if you're building like this data warehouse, you do get around the sampling thing. So that's a neat thing. If you were like, I don't like Google Analytics, I don't want to use the interface. You could still technically use Google Analytics to configure tracking, use the ga4 to BigQuery integration to put it all in BigQuery and then you've got your analytics collection in place and it's all already just going to this data warehouse where then you can just work with SQL and Looker Data Studio and not even have to touch Google Analytics if you don't want to. So there's kind of a lot of possibilities there depending on sort of the appetite you have for sampling or not sampling. And generally speaking, you've got to have kind of a lot of traffic to really feel like you're hitting problems with sampling. Some of the scenarios where sampling is more problematic are the scenarios where we're like, well, interest in the people that definitely saw this page and they might have seen this other page and then eventually they did. You have all this criteria going and then you're like, but only desktop or only mobile on top of all of that, right? That's where you're going to get. You're going to run into some of those queries. And then furthermore, let's say you actually like the sampled data for some reason because maybe you feel like the statistics are good there with the way the sampling calculation works. BigQuery has some of the same kind of aggregate functions where you can fire queries that run m. More. The advantage here is, let's say you're got a lot of data in BigQuery and you're trying to save on compute power. At that point you might run a query that like samples that data set rather than running the entire thing. Some of those queries that are under the hood in GA4 are available to you within BigQuery.

Speaker A: Got it. That makes a lot of sense. What transformations do you apply in GA4 that then make it into the database that you query manually?

Speaker B: A big part of what I like to do when I'm tracking a website and a big part of what GA4 is all about is being able to track a lot more than just what we think of as the web analytics basics that we got from like the previous version of Google Analytics. Right. And so previously if you installed Google Analytics, it would automatically just track basically page views and then anything else beyond a page view. Like a form submission or E commerce stuff would be considered like custom tracking. It's not part of the base code. So. So with GA4 they were like instead of page views we're just going to say everything is an event now, but when you do install it, it still gives you the basic page views. But because of this like event based mentality, you can track the limits are limitless in a way because the old version to compare had like event category, event action and event label. And if I was going to track something there started to think of those as like a hierarchy, right? So the category we might say main navigation and then the action or actually no, let's say that the category is navigation, the action would be main. So the main bar that you're interacting with and then the label could be like whatever part of that nav that you clicked on. For an example, right? So you'd come into the reports and you click on navigation and then you could see if you also had a footer nav and a utility nav. You could see like how's the navigation activity distributed across these sort of user Interface elements. With GA4 I've found that the naming conventions shift a little. What I tend to do is give it an event name of something like Navmain. In GA4 I can still search for events that start with nav. To get that navigation thing we could do navmain, navfooter, nav utility, whatever kind of section. However you want to think of the navigation navbutton just to capture all the button clicks on the site, right. And then as we're tracking those, what does the user see? They see some kind of click text or what is the button text. What does the main nav say? Login, learn more what have you when it captured that, but then also where they go in. So the click URL is another thing that's worth tracking, right. So that's a uh, sort of a glimpse into how I look at some of the custom events. Just as it would relate to tracking the main navigation. When you are interfacing with GA4 it would be hard to hold together a report. You kind of have to go in the explore section and then play around with that in order to get like a sense of how the NAV is working. So bringing that over to BigQuery, we can then make a query where we can associate the client ID from all the sessions to various NAV activity. We could then compare that to other sessions that have purchases in them and you can start making groups of queries where there's essentially sessions with various activities that you're interested in and then you can compare that against all of that other data. Another thing early on with GA4, I think they're getting better with some of the raw data now. But the concept of a session didn't really exist in GA4 when it first came out. It was all uh, user level data and users contributed to events. Right. So that was it. And then there was a third scope that was related to like products. Right. If you had E Commerce site. Right. Product level schema. And so that session part was missing and they've been building a lot of that back in. But early on some of the queries I would do were related to combining the cookie ID with the session value because the session value is a timestamp. So believe it or not, two people could actually do something at the same time, which would be wild. But so by combining this user cookie value with the timestamp, you can essentially create like a new session id. And so there's some kind of aggregation that, that I've done to recalculate sessions in a way that's more similarly aligned to earlier versions of Google Analytics. And some reasons for that would be like you have the historical data and you wanted to continue to analyze it that way. So you might try to have the UA data also stored in BigQuery. And when that stops and GA4 starts, you can have your queries. Like we were looking at this one and now we're looking at this one. But the graph and the chart that you put together allows you to see all that data seamlessly because it's joined up with all the same column headings. And so, you know, that's another example tying together your historical analytics data with your new analytics data.

Speaker A: Yeah, yeah, that makes a lot of sense. Thank you for that. That's. That is really applicable, I think to a lot of listeners. What is, have you used amplitude before?

Speaker B: That's sounding like something I've dabbled with

Speaker A: Google Analytics, but it doesn't sample so it's a little bit more self serve in that sense.

Speaker B: Nice. Yeah, I can't say I've done anything with this recently. There was as far as tools outside of Google Analytics. I've worked with some of the Adobe products back in the day, I think when they were calling it like Omniture, if that's right. I get them mixed up. Yeah, yeah, it was Adobe Omniture at one point. I think now it's, I think they're calling it Adobe analytics today. I'VE worked with Mixpanel and Heap. But yeah, the thing with Google Analytics to get back on the sampling thing for a second unless you're like seeing huge volumes of traffic, like you're probably not experiencing that. And then if you do get to the point where it is a problem, like I've seen like probably in all my years, one account where they had literally enough data points happening in a given day that they were just like over the limits. And so they had to sign up for the GA360 and that used to be priced uh, in the older days with ua you'd be like oh, the fancy expensive version of Google Analytics that nobody can afford because it's like 150k a year. It now starts at 50k a year for I think it's like something like 30 million, maybe 25 million. I can't remember the base number. There's a base number of events and then the per million increases after that are like 17 or something along those lines. So once you get on the plate, the paid platform M, you can scale up and have no sampling. The GA360 accounts have no sampling. But again if you are, I think it's pretty rare like in, in my experience when I've compared the numbers that we're seeing in GA to like the actual numbers like, like ad block is something that we're all concerned about. Like 30% of users have ad blocks so we're missing all of that data. And it's hasn't usually been that bad definitely. We tend to say if we see numbers that are slightly smaller, we're not that concerned about it because we know that adblock is a thing but like it's not so much that it's like a third smaller than the reality usually. I think that some of the folks that tend to use adblock tend to be some of the more tech savvy users in the first place. So people are using AdBlock than what you might think. Yeah, yeah. So with the. I uh, yeah, I'd be shocked if like you compared like something like amplitude or Mixpanel or anything else to Google Analytics and that like the ones that claim to have no sampling would have like varied enough data that it would be like, like shockingly different. I could imagine a few different industries where that might be the case. But yeah, I think for the majority of businesses like Google Analytics is a great solution. It's the most installed web analytics platform out there and with its integration with BigQuery today, it might be doing a pretty good job at selling BigQuery for some people, the amount of data you can put into BigQuery I think it's like something like the first terabyte. I can't might be free, I don't know. This pricing probably changes all the time but I feel like $5 per additional terabyte or something like that after this a terabyte is like a lot of data so you can really go that route and if you're trying to be more bootstrapped about it it's still a great way to go.

Speaker A: Yeah. Have you used HubSpot for attribution and what's your general take um on using it as an attribution platform?

Speaker B: Yeah, I've used it probably alongside Google Analytics and I compare notes there. Within HubSpot they do have like their own kind of take on like the channel mapping things. So if you're got the HubSpot tracking code installed it's going to tell you what it thinks organic search and all these other things are and the numbers are usually pretty similar. I think the scenario where I'm using HubSpot in that example is like usually when there's a sales team involved and like the marketing team is needing more kind of upfront automation and like forms that can help funnel things in. I know that HubSpot's got a lot of like add ons now that allow for adding or like managing stuff like they're trying to make it more centralized and uh, yeah I think you can get good results with Hubs. HubSpot I think if I'm personally like looking at a marketing automation tool. I did work at a company called Drip back in the day and so I still have a sweet spot for Drip. I think it's one of the best marketing automation platforms on the market and then outside of that Klaviyo is one that like seems like it's everywhere today as well. So yeah experience uh integrating with all

Speaker A: of those solutions and do those do attribution as well or are they focused mainly on automation?

Speaker B: I think those are. There are reports within all the tools that may or may not give you the attribution that you're looking for. I think that the email tools specifically are more like ad networks as in um, they're going to give the credit to the most email that most recent email that was sent or campaign that went out or automation that sequence, what have you. So I think it's skewed a little bit that way. I think again yeah, for a full kind of channel channel picture I tend to rely on Google Analytics, they have another thing about their attribution, right? So it previously was touted as being like a last click attribution, right? So if you had come to the site through email and then through TikTok and then finally did a paid search ad, and then that was the time that you decided to buy because. Because the last click was on a paid search ad, like Google would give credit to that more than the other things. But as time has gone by, like people may or may not know, there used to be a handful of different attribution models that you could compare within Universal analytics. And with GA4, they got rid of that in favor of the Data Driven Attribution. And I think what's going on under the hood with the Data Driven Attribution is that they're still looking at all these other models. They're just not surfacing it within the user interface. And then they're using machine learning to predict based off of all these models saying that this credit goes here or there. The data driven takes all of that into consideration in a machine learning take. That's my guess at what it's doing anyways.

Speaker A: Yeah, yeah, that makes a lot of, um, sense. Wow, there's so much here. Do you have any final thoughts to share with the audience?

Speaker B: Yeah. If you've been listening to this and it, uh, sounds interesting and you feel like you've been struggling with this sort of thing, or you're just potentially wanting to learn more about what it would take to get a better tracking solution set up, you could reach out to me on LinkedIn. My name is Joshua Lauer, and if you just find my profile, it's a picture of myself at a wedding probably about 15 years ago, looking like a happy young man. Send me a message with the words deep dive audit and I'll get back to you with some time with a calendar link and we can have a chat about your business and how we can get the tracking all sorted out.

Speaker A: Amazing. Thank you so much, Joshua. You're doing really important work and I'm sure we'll talk again soon.