So What? Marketing Analytics and Insights Live
airs every Thursday at 1 pm EST.
You can watch on Facebook Live or YouTube Live. Be sure to subscribe and follow so you never miss an episode!
In this week’s episode of So What? we focus on out-of-the-box attribution models. We walk through the different types of models available in GA3, which ones are most appropriate for your type of business, and what to do with the analysis. Catch the replay here:
In this episode you’ll learn:
- The pros and cons of each model
- Which model to use for your business
- The implications of GA4
Upcoming Episodes:
- Video SEO – TBD
- How do you benchmark a website’s performance? – TBD
Have a question or topic you’d like to see us cover? Reach out here: https://www.trustinsights.ai/insights/so-what-the-marketing-analytics-and-insights-show/
AI-Generated Transcript:
Katie Robbert 0:19
Well, Hey, everyone, Happy Thursday. Welcome to so what the marketing analytics and insights live show Today on the show we are talking about out of the box attribution models. So a lot of times the topics that we cover, you know, sometimes the feedback feedback we get is that, you know, it’s really interesting information. But it’s unattainable. Because, you know, it requires special software or skill sets, you know, around like development and coding and advanced analytics. And so we also want to make sure we’re covering topics that, you know, are a little bit more accessible to the majority of marketers. And so today, we’re covering out of the box attribution model, which is low to no cost, because it’s literally out of the box. If you have Google Analytics, using the free version, you have access to all of these free analytics, attribution models. So Chris, where do we want to start?
Christopher Penn 1:13
Well, I think it’s actually important to say that these models are shared among different pieces of software. So Google Analytics has mainly these off the shelf models, they share a common heritage with Google ads. So you’ll see them in when you’re running Google ads, you are actually asked him to set up a campaign to choose an attribution model for your campaign. And even Facebook, you know, when you when you’re building out your Facebook analytics for your Facebook campaigns, it also asks you to specify an attribution model. So these, these common models are pretty standard. And they’re, they’re not terrible, for the most part. They, some of them are good enough to get by until you can you spend some more time digging into the data. So probably, what would make sense start with are what the heck are all these different models? Because there’s a bunch of them. So Katie, do you want to talk through a couple of the different more obvious ones, I’m gonna, I’ll bring up here, our attribution project and Google Analytics. And you can see right on the screen here, some of these different ones, we’ll start off with, like, what a couple of these are and why you would use them?
Katie Robbert 2:22
Yeah, absolutely. And so if you don’t, you don’t have to be in the attribution project. So Chris, I think you’re in Google Analytics. Four, is that correct? Three, okay. Oh, you’re so okay. So you’re still in three. In Google Analytics, three, you have access to not only an attribution project, which we do recommend you set up. But also just the general attribution modeling, that’s out of the box. So Chris, you’re looking at this in the attribution project, if you don’t have an attribution project set up, you can still access these attribution models, through the main navigation on the left hand side of your main view, it’s the model comparison tool. And so there’s a handful of different models right out of the box. So you’ve probably heard first touch last touch. So in Google Analytics is called first interaction last interaction. And one of the things to note I like to give this little pro tip this little, you know, cheat is that if you look at the icons for each of the attribution models, then you can actually see where the credit is going to be given. And essentially, an attribution model is just that it tells you what gets credit for the action being taken. So a first touch or last touch is really good for a very transactional website that doesn’t have a lot of pages on it. So there’s not a lot of, you know, other calls to action other than buy this thing, or contact us now. And so if it’s very singularly focused, and you’re don’t have a very large mix of marketing channels, if you’re a smaller business, and you really only use maybe email marketing, and social, the first and last touch interaction might be really good. And what comes to mind. JOHN is actually the marketing over coffee website, it’s fairly straightforward. It’s fairly simple. And there’s really only one, maybe two things you can do from that website. And I know that you use email marketing, and you use a little bit of social and so it probably makes sense for where you are, if you’re looking for a free version of an attribution model to use either first or last touch model. A lot of companies make the mistake of using a first or last touch a lot of times as the last touch, especially if it’s a larger company, and they have a large mix of digital channels. What ends up happening is it doesn’t matter what happened prior to that last touch, that last touch gets all the credit. And then if you’re trying to resource you’re unbalanced. So you can use First and last touch is not the greatest if you’re a larger company using a lot of Digital’s, even if you’re a smaller company using a lot of different digital channels, it’s not really going to tell you what’s working, it’s just going to tell you the very last thing, or the very first thing that somebody saw, so it doesn’t give you the full picture. Another one of the standard models is what’s called a linear model. So regardless of you know, where in the process or how many times, or the different mix, every single thing that led up to that specific action gets the exact same credit. So you don’t know if Facebook is working harder than email. Or if your display ads are working harder than your organic social, you just don’t know. So everything gets the exact same amount of credit, you know, we don’t have a really good use case for this one, unless you are really just working against a lot of, you know, internal infighting and bureaucracy and red tape, this might actually be something to just sort of like, at least start to get people moving in the same direction of, hey, look, you guys are all important, you know, just have that conversation with people. time decay, is probably one of the better models to use if you’re using an out of the box attribution. And if you are using a good healthy mix of digital channels,
because what it does is essentially what it sounds like it, it decays the credit over time for whatever it is, that has led up to that specific action. And these actions that we’re talking about, those are goals. So the goals or the goals that you set up, you know, contact us download a paper, subscribe to a newsletter, you know, certain amount of scroll depth on a page, you know, the goals really vary depending on the type of business. And so you also want to make sure you’re matching up the model to the kind of business that you have. The other one position based, is it’s kind of like a choose your own adventure in the sense that you can determine how much credit you want something give. So if you you know, double down really heavily on email marketing, and then only occasionally do social, you can then determine, say, I want to give email marketing, two times the credit of social all the time. And so you can sort of set those parameters so that you know, there’s a little bit of customization there, which is good. But if you’re just looking for a straight, you know, our best recommendation for out of the box attribution, we would recommend time decay.
Christopher Penn 7:32
One of the things to keep in mind with a lot of these things is you want to have a sense of what’s happening in your data. And like how long the path to conversion is. Because the longer the path to conversion, the more complex a model you need to what Katie was saying, if you have somebody just comes in, clicks on something buys it and they’re done. Yeah, you first touch, the last touch is perfectly fine. So if you’re going to go into Google Analytics into an attribution project, you need to set this up and you need at least 30 days worth of data set, and is based on your existing Google Analytics goals. So if you haven’t gotten done all that stuff, that’s a prerequisite. But then you go to your conversion path length here and you look to see how long is the path to conversion? How many touch points to conversion does it take so this is my personal website, I would actually be okay with the first touch model, because 90% of my revenue and 85% my conversions are one and done. They come in, they do the thing and they leave. Right. So in this case, this is this would be a business where Yeah, it probably would make sense to use the last touch model because you know, it’s pretty quick. On the other hand, if this was like 70%, or 60%, or 40%, and you had a more of a curve of conversions, you know, going out pretty far then yeah, you definitely want a more complex model to deal with the fact that there are all these different touch points that are happening. So one of the tips that we suggest is you look at your conversion path length that say, how long does it take to convert somebody? Now, the other thing you have to ask yourself is, if your conversion path is really long, especially for something that isn’t a complex sale, then you got to ask yourself why and go. I got conversion pathway seems to be really, really, really long for we’re just trying to get people to sign up for an email, like, what is it about our site or our offering or whatever, that sucks so much. Nobody wants, it takes like 20 touches, to convince them to buy something.
Katie Robbert 9:28
And now Chris, correct me if I’m wrong, but I, I believe with these out of the box models, you’re restricted down to 90 days in terms of what they’re going to look at. So if your sales cycle is longer than 90 days, or you know, Chris, to your point, if it takes someone longer than 90 days to make a decision to subscribe to your email, you’re not going to get really good information because Google isn’t looking at the data beyond that timeframe.
Christopher Penn 9:55
That’s exactly right. There’s a 90 day window in Google Analytics for all the stuff you have A process that is longer than 90 days, you need to be encoding that data in your CRM. So each touch point has happens. And then you build either build a custom model, from your CRM data, instead, we have a couple of clients that do that, where some of their touch points are a couple years long, and so they have this longitudinal data they use to build those models instead. And that’s, that’s the workaround.
Katie Robbert 10:24
So, john, I’m gonna keep putting you on the spot, because you know, it’s fun. You know, you talk with a lot of prospects and customers around attribution modeling, and those questions that they have, what are the kinds of questions that people are asking? That leads you to say, I think you need an attribution model?
John Wall 10:45
Yeah, well, everybody wants to just get to the holy grail of like, being told that, Okay, look, you just need to put all your money on these three marketing programs here, and you’ll triple your business, you know, next quarter, and that’s great. And unfortunately, there’s like a million Gremlins hiding beneath the surface, beneath the surface of this thing. Because there’s so many things that can happen, a big thing that people need to address up front is, you know, do we care about awareness, because for some companies, really, it’s just about awareness, it’s just about getting them to know you’re there. And then you trust your sales process that Yeah, so if they’ve showed up and come in the door, sooner or later they are going to buy. So it is really that first touch, and then at the other end of the spectrum is like, ya know, there’s a lot of people milling around, but unless, like, they hit the pricing page, and, you know, maybe they take this survey, you know, they’re never going to be a customer. So those are, you know, tend to shift towards later stage. And then the other thing that gets messy with, you know, complicated sales, is how you, you know, when you attribute those percentages, making sure that those match back to revenue. And sometimes that’s even account based, you know, you may have five people doing things that all attribute back to one deal. And the stereotypical mess, you know, when we see a senior management Come come to us with this, it’s always the CEO is like, well, marketing said, you know, they attribute 2.8 million in deals this year, and we, you know, our net revenue was 1.3 million. So obviously, something is very wrong here. And, and, you know, then they’re right, because stuff keeps getting attributed back. And you don’t want to end up at the other end of the spectrum, where you’re running these models, and like every marketing program is contributing, you know, $4.38, and you don’t know what to cut or where to double down, because nothing seems to, you know, be the silver bullet. So, yeah, there’s so much going on between having your data, right? And then is it you know, awareness versus close deals? And how do you roll in all the data, and then you hit the other one with timeframes to I mean, it’s insane. You’ll see a deal where, you know, you dig and dig and dig, and you find out that Oh, yeah, Bob worked with Susan, eight years ago at this place. Like, that’s actually how they knew about this. And you know, that’s not going to cover it in the model. So you need to work around all that kind of stuff.
Katie Robbert 12:56
So Chris, I think it would be helpful. So we’re looking at the model comparison tool. And again, if you have Google Analytics set up, you have access to this exact same view. So can we walk through what people are seeing on the screen? Like, what are they looking for? What do the different, you know, toggles mean, what do they do?
Christopher Penn 13:17
So the model comparison tool is really to understand your data and understand how, how to attribute credit for different kinds of models. And what you’re looking for are? A is the model that best describes your business? And be based on that? How does it differ from, say, a traditional last touch of first touch model. So let’s go ahead and take a look here, I’m gonna first set the look back window to crank up to 90 days, because I do know that I’m a b2b websites, my personal is that but it is b2b site for the most part. And so the longer conversion window makes more sense, because, you know, some people may see a tweet out two months ago, and I’ve gotten cookied in. So that’s the first part. The longer the window, the more you’ll have things like assisted conversions and multiple channel touches, because there’s more opportunities for people to interact with you. The shorter that window is, the less stuff you’ll have there. Again, if we were to look back at that path length stuff, either in this tool or you can actually do it within time lag right here in the in multichannel files menu, you can see Yeah, some for, for if you restrict your model down to, you know, three or four days, now, it’s going to look a lot more like the last touch model because there’s just not enough data to give those assists. So next thing we want to do is choose what kind of conversions that we actually want to to modeling. In this case, I have e commerce on my site, people can buy my book and stuff there. They can also buy our data science one on one workshop which can find a TrustInsights.ai dot AI slash days sign SWOT one and there’s also non financial transactions. So newsletter subscription Amazon Gale. people visiting my speaking page, and so on and so forth. For this, let’s just go and choose, let’s click none. And let’s go with let’s go with newsletter subscriptions, I think it’s a good one, I’m going to crank up the time here here to actually look the last 90 days. Otherwise, it’s kind of silly to to have a 90 day look back window. And what we should see is the last touch model that says, you know, these are the channels that have helped conversion. So direct traffic, which is unknown, is 34% of the conversions in the last search model. 24% was organic search 22%, referral, traffic 11% email, and social networks 6%. Now, my next step would be okay, what if I want to compare it to, as you were saying earlier, at a time decay model where you distribute credit over time? What’s the difference? And what you’ll see here as a percentage change of conversions, some things are more important in a time decay model than they aren’t unless such models is there’s a difference in how users are interacting, for example, referral traffic is 5%. More important in a time decay model than less sexual means that in a time decay model, there may be some things happening earlier on in their interactions that do yield more conversions. Now, what do we do with this information? Well, in a time decay model, especially, we want to look at, what are the percentages of conversions by channel, right? And then ask ourselves? Are we allocating resources appropriately, to match those conversions? So let’s say I am, I try to build my newsletter list here. And I’ve invested, you know, 80% of my marketing budget in in Facebook, right? I do go all in on Facebook, and try and get my grow my list from Facebook. If I were to do that, and say, 80% of my budgets going there, and 6% of my results came from there, we’ve got a pretty big mismatch. That’s, that’s clearly done something wrong. And so I would know, oh, well, in this case, maybe I need to re balance what it is that I’m doing. Or if I am going on, you know, if I’m looking at investing in my SEO team, and my SEO team, I’ve got resources, you’ll 5% of the marketing budget, but it’s generating 23% of the conversions, like I need to resource more into SEO, because organic search is working for me. So that’s what you do with a lot of these models is to have a look at look at the comparisons, look at the different performance of the models, and then said, am I resourced appropriately? And are there any surprises? Is there something here like, Hmm, I didn’t expect to see that like, in this case for newsletter conversions, I didn’t expect referral traffic to be that big contributor.
Katie Robbert 17:58
And so then, yeah, you would have to dig into, you know, which sites are referring? And does it make sense? And so, you know, Chris, you just walked through? Probably the most common scenario as to why using any kind of attribution model is useful to your marketing, because the question of where do I put my money? What Reese’s resources do I need? And what’s working? Like? That’s the most common question we get is, you know, what, how do I budget for this? Where should I put the money? I was just given $100,000 for the next quarter? What do I do with it? So number one, is making sure that you have your tracking set up correctly, for all of you know, whatever campaigns you’re running, or whatever the actions on your website. And then number two is using some kind of an attribution model. Unless you’re a b2c and just a straight ecommerce site, probably, you know, a time decay, looking at that 90 day look back window is going to be your best bet. And so if you do see that surprise, you know, that Chris mentioned, so let’s say referral traffic is just killing it for you, you can look at your attribution model in the form of source medium. And so Chris construct to see here, okay, what is referring you to my website that is then you know, causing people to take the action I want them to.
Christopher Penn 19:22
Yep, exactly. So number four Elegant Themes. The Elegant Themes affiliate program is actually sending traffic to my site, which doesn’t make sense because I should be sending traffic to them so that people buy this stuff. So after that’s a mystery for me to go solve medium comm referring converting traffic to my site, which is amusing and ironic, because I got banned for medium for spamming. Oops. T dot SEO is a miss code. In this case, so this is Twitter traffic, but because of the way Twitter does it source medium always come into Google Analytics as referral traffic. And you can’t fix that. It’s that’s just the way it is. You can change in Google Analytics through which channel grouping it’s in. And we do that for our clients. But you cannot, you cannot fix the XLNet. So you can fix it in Google Analytics four, but it’s very dangerous to do that. But yeah, so Twitter, traffic is 3%. They’re a website, I can’t remember her first name. But this is a marketing blogger, but 2% of my traffic. So my next steps here, I’ll look at YouTube. That’s nice. We have LinkedIn and YouTube and Twitter, all of which are social. Exactly all which is social media. So now my next step is to go okay. I’ve got converting traffic sources in my, in my out of the box attribution model. Maybe I should be focusing a little bit more on some of these things. So one of the questions that I asked myself at the beginning of 2021, was, should I discontinue my my daily YouTube show? Right, and I stopped, I actually stopped it in December in December. And because it didn’t seem to be converting, and it was the right choice, and still is the right choice. Because even with three, almost three years of daily shows up on YouTube, it’s still referring only 1% of my conversions, right? Am I going to, if I have 45 minutes in the day, to allocate to myself to write content, create content? Should I be putting in the channel that converts at 1%? Or should I be writing blog posts in the child compared to 22%? Which is organic search? The answer, obviously, is I should be converted, I should be making good blog content stuff, because and feed the organic search piece. So that’s, that’s what I did. But from here, I can look at this and go, what things are surprises? What things make sense? What things should I maybe stop doing? Because investment isn’t just budget, it’s also time, like we only have one or four hours in the day. So where should I be putting my time based on this? I should be putting my time in SEO, email, and then maybe outreach, just maybe some social media?
John Wall 22:06
You know, I was gonna ask I hadn’t thought about it this way before, but looking at the way the time decay runs. If all campaigns were considered equal, can you consider this your funnel? I mean, because it’s the stuff that’s scoring, the lowest obviously is earliest in the cycle? Or is that too far stretch our most clients not, it’s more the campaign’s suck.
Christopher Penn 22:25
It’s the campaign’s suck, this is not a timing thing. They’re all what you do see you do you see a difference in the time you have compared to the last touch, if the conversion number is lower, it means that probably is a little earlier in the process than it is but you can’t get timing out of this, you’d want to use like a machine learning attribution model to get the exact timing.
Katie Robbert 22:48
Make sense? So Chris, there’s a couple of other toggles that you haven’t walked through. So one is right above the last interaction time decay right now it says conversion and CPA? Is that the correct? The two correct columns to be looking at? Considering you are? I mean, is the CPA more appropriate? If you’re looking at Google ads? Or is it appropriate for this view?
Christopher Penn 23:15
It depends. It depends on what you guys said. And for me, this makes no sense because I don’t run ads. So both CPA and Rojas are stupid to have here because I don’t do ads. If I did ads, I would probably want to look at be investigating both those this case, I’ll switch over to, to conversions and value, because there are dollar amounts associated with a lot of these things. But if I was running Google ads, or anything where it had an ad spend data being uploaded into Google Analytics, I could then look at this and say, okay, is the cost per acquisition where I want it to be? And am I getting a return on adspend? That’s, that’s high enough to meet my goals. So if I, if we flip over to conversion, and return on adspend. In this case, still all zeros because I have anybody, this is where you would start pulling out your industry basic rules, like, for example, there’s sort of a general rule of thumb that your return on adspend should never be below 400%. Because return on adspend is different than ROI, you’re not ever computing the cost of profit and return on adspend. You just, you know, just the return on adspend directly. And so in order to make sure you do have positive ROI, most folks say you know, your target Rojas should be never less than 400%. So this would be a good way to check on that at the bottom of the funnel.
Katie Robbert 24:32
But so in this case, because you’re not running any ads, really you should be looking at conversion, value and value. Got it. That’s right. So if you switch back over to conversion value, then next to that, to the right of that you have another drop down, which right now is percent change and conversion. You can also look at it as a percent change in conversion value. Are you. And to be clear, these values that Chris has seen these dollar amounts are based on the value that he is assigned to this specific goal. And we’re looking at his newsletter signup goal. And Chris has assigned, you know, whether it’s $1 or $10. That’s, you know, depending on what makes sense for his business, that’s where those value amounts are coming from.
Christopher Penn 25:22
Yeah, we value news or subscribers basically being worth $1.
Katie Robbert 25:28
So, Chris, john, you know, if someone is really you know, they don’t have the resources, they can’t afford something advanced, you know, something custom, and they’re really, you know, trying to make the most of an out of the box attribution model, what are some of the other things that they could be doing to set themselves up for success?
Christopher Penn 25:52
The biggest thing, is this all this? Is it dependent on goals? Right. So if your goals and goal values are not set up to be anything intelligent, none of this works. So there’s a goal in here I have, you know, speaking page visits, right, that’s a pretty squishy goal, I set that up for myself, because I just wanted 100 people visiting my speaking page to hire me to speak at events. But that’s kind of a dumb goal, right? It should be like, you know, speaking form fill requests, like somebody actually filled out the form for us, because that could literally just be bots, that wouldn’t, you know, that might not even be humans. So, if you’re going to spend time on attribution modeling, and you should be 80% of your time should be on the goal and planning upfront and the infrastructure to make sure it’s all working. Because if none of that’s working, this is all pointless exercise.
John Wall 26:41
Yeah, that’s funny. See, I was gonna come at that from a different side of saying, don’t go crazy with your 100k budget, looking at the model, until you’ve also gone through and looked at all the goals and made sure that you know, this tracks, because the last thing you want to be doing is throwing around all this money. And then you look behind the scenes, and you’re like, Oh, actually, we only had four goal completions this year. So you’re that total dartboard. situation there. So. But yeah, obviously, the whole thing is built on that stuff being mapped out correctly.
Katie Robbert 27:09
I would add to that, you know, it’s dependent on your goals being set up correctly, but it’s also dependent on your Google Analytics system being set up correctly. And the advice that we always give is that Google Analytics infrastructure out of the box is set up incorrectly. And that’s because of the channel groupings. And so out of the box, it’s wrong. And so you would need to go into your, your settings, and, you know, resolve all of that, you know, so as Chris showed the example, you know, Twitter referral or LinkedIn referral, that’s incorrect. And so you would want to include that information and make sure that social, it’s set up correctly. The other one that we know right off the bat that’s always set up incorrectly is email. For some reason. Gmail is not automatically categorized as email out of the box, even though it’s in the same family of brands. It’s all owned by Google, you know, why they do that? We don’t know. So the other thing that we want to chat about a little bit. So we’ve talked a lot about Google Analytics, three, which is where everyone should be right now, you should have your Google Analytics three systems set up correctly, you should be confident in the data. And then in parallel, Google Analytics, four, it was released a few months ago, it’s still working through some, you know, new features, it doesn’t have everything yet, we do recommend you set it up in parallel. But don’t shut off your Google Analytics three, if you want help with that, obviously, reach out to us. So Chris, attribution modeling looks a little bit different in Google Analytics for
Christopher Penn 28:48
what’s going on there? Well, for one thing, that’s the whole thing about default channel groupings has been solved by Google because they’re gone. Now you’re stuck with a broken source medium, and there’s no way of fixing it.
Katie Robbert 28:59
Well, that was a terrible idea.
Christopher Penn 29:02
Unless you go into Google Analytics for admin, and you actually make instream modifications to your data stream, I generally don’t recommend people do this unless you know what you’re doing. But you can actually manipulate the raw data itself inside of Google Analytics for and and make substantial changes to it. Again, not the kind of thing that you probably want to do, you know, just just on a whim. The second thing is for attribution modeling, and which model of shoes Google has solved that too, it’s gone. There is no attribution modeling in Google Analytics for at all. And the reason for that is that attribution modeling, as we’ve done it, and then showing here, is it part of Google Analytics it is it’s a reporting mechanism. It is not an analysis mechanism. And so Google in its design philosophy, Round Google Analytics for us said that this is a business intelligence tool. This is an analysis tool. This is not a recording tool, and they want you to do the reporting somewhere else. Typically would be Google Data Studio. Here’s the catch. There is no attribution modeling and Google Data Studio for Google Analytics for either, which is, which is kind of awkward. The best that you can do right now is cobbled together two different types of mediums. There’s user medium, and session medium. And you can do with source medium combo. And what that will give you is essentially the first touch for a user. And then the last touch for conversion within that session. And so you sort of like get a first and last touch model in ga for I’m hoping that the omission of attribution was an oversight. In ga for I’m hoping that it eventually does come into the analysis how because they do have some templates in here that that are not bad attribution modeling, not one of them. Because there’s no way to do it in Data Studio, either. The only place that you can do attribution modeling, as we’ve been talking about it is in the raw BigQuery data itself, which means you need to a provision ga for to be recording and staged in a big way. And be you then have to get the data out of BigQuery, which is a SQL database, and then process the data yourself to do an attribution model. I was playing around with that earlier this week, getting set up to start pulling that data in and start building attribution models with ga four. It will be better than what’s available ga three because you get 100% of your data. And it’s really nice. But it’s also inaccessible for the average user, the average user is not going to sit down and write code against the BigQuery API and extract stuff out. So I’m hoping that there, there will be something that Google puts together, that allows us to do attribution modeling, because right now, when you go in here, and you look at your conversion events, see a junior subscription, I can dig around and see like, what was the event source and things like that. And, you know, add a few columns here and there. But it’s nothing even close yet to to a real attribution model. Like there’s, there’s just sort of the event campaign immediate, this is not a super helpful stuff.
Katie Robbert 32:27
And you know, so everything you’re describing is exactly why even if you’re setting up Google Analytics for we highly, highly, highly cannot emphasize enough, do not turn off Google Analytics, three, there’s no penalty to having both of them running against your website, it doesn’t slow things down, especially, you know, if you have it set up correctly through Tag Manager, you know, and so, continue to use Google Analytics three, until it’s completely discontinued by Google, which we know in terms of these rollouts is going to be a couple of years, because as Chris was mentioning, there’s a substantial amount of features missing. And so Google has been slowly rolling things out and adding on to it and upgrading. But by no means is it ready to replace Google Analytics three?
John Wall 33:17
Do you think they’ll roll something out in Data Studio perhaps? Or is this going to be one of those Google Reader things where it’s like, yeah, it was good, but it doesn’t fit the mission. So it’s not going to be around anymore.
Christopher Penn 33:28
We don’t know, I would imagine, they will have to release something. Because there are plenty of competitors to to Google Analytics, I would like to take a shot at the throne. And with the changes that are coming in advertising with ad tech, and not being able to track certain things, Google Analytics, l and having it on individual websites will be one of Google’s only ways to be able to see what’s happening once user leaves a Google ad and goes to a website, right. So without that, if their lack of you know, attribution modeling, causes people to switch analytics applications, it diminishes their ad system, potential revenue as well. So there’s, there’s a clearly financial interest for Google to maintain its visibility into millions of websites out there.
Katie Robbert 34:21
So as we start to wrap up, we’ve covered out of the box attribution models from Google Analytics, three, the pros and cons and the use cases of each of those. You know, we’ve talked about which ones are appropriate depending on the type of business and really our recommendation, you know, is generally using the time decay model. It is the out of the box model that most closely mimics true user behavior, but really the only you can’t get good attribution modeling out of the box. You know, we would really recommend if you can find the resources or the budget to do something more like an advanced attribution model that you would get out of machine learning, using all of the same data, but putting it together in a different way, that’s going to be a more accurate representation of what’s going on with your digital marketing and where you should be budgeting and resourcing with channels get the credit. We’ve talked through the implications of Google Analytics for the bottom line is, don’t completely switch over, because you’ll lose a lot of the functionality that you know, and use in Google Analytics three, you know, what are some other, you know, parting thoughts for people who are still, you know, constrained down to using something out of the box, because, you know, it’s cost, you know, it’s, you know, it can, it doesn’t kill their budget, the words are not coming out.
Christopher Penn 35:44
One other thing that I think is really important for people to do, is having a backup analytics system in place that is running locally on your site that’s capturing local data. This is a package. It’s an open source free package called matomo. We have it running on our TrustInsights.ai website. And obviously, I have it running on my personal website, this, again, gives you raw data, right? So yes, you may not be able to do the attribution modeling in here because it’s, it’s it’s even more primitive in here than it is in ga for but you have the raw data in your own database on your own site. So even if we get to a point where say Google Analytics is no longer able to track people, because tracking pixels just get wiped out, which is not inconceivable at some ad blockers already do that. You can get server level data from your own server, and be able to process that and turn that into usable data. So again, systems like this are open source, there’s there’s actually not that much of a technical hurdle to getting it set up. It is free of cost other than disk space on your server. And it’s just always a good idea to have a backup, because at the end of the day, you don’t control Google. You don’t control Facebook, but you do have control over your website and the data it generates.
Katie Robbert 37:02
So I think that that is an excellent place for us to start to wrap up. You know, if you have additional questions that we didn’t cover, feel free to drop them in the chat. Otherwise, you can always contact us if you have more questions. And we’re happy to help too. We’re happy to help you get things set up. So, guys, I think that’s it.
Christopher Penn 37:23
All right, we’ll see everyone next week. Thanks for tuning in. Thanks for watching today. Be sure to subscribe to our show wherever you’re watching it. For more resources. And to learn more. Check out the Trust Insights podcast at Trust insights.ai slash ti podcast and a weekly email newsletter at Trust insights.ai slash newsletter. got questions about what you saw on today’s episode. Join our free analytics for markers slack group at Trust insights.ai slash analytics for marketers. See you next time.
Transcribed by https://otter.ai
|
Need help with your marketing AI and analytics? |
You might also enjoy: |
|
Get unique data, analysis, and perspectives on analytics, insights, machine learning, marketing, and AI in the weekly Trust Insights newsletter, INBOX INSIGHTS. Subscribe now for free; new issues every Wednesday! |
Want to learn more about data, analytics, and insights? Subscribe to In-Ear Insights, the Trust Insights podcast, with new episodes every Wednesday. |
Newsletter
Podcast
