Measured Opinions #10: Better metrics for web measurement
This week Dan and Dara dig deep into the ABCs of Google Analytics metrics. Defining what they actually mean (not what you may think they mean) and discuss their relevance and/or place in to modern technological age.
Dan’s Hyper Island course they’re attending is https://bit.ly/3lTDYPr.
Dara’s new find Agency Hackers is https://bit.ly/3hY0VQr.
In other news, Dan and Dara watch some TV!
Leave a rating and review in the places one leaves ratings and reviews, or suggest a new topic by emailing Dan and Dara at hello@measurelab.co.uk.
Transcript
[00:00:17] Dara: Hello and thanks for joining us in The Measure Pod, a podcast for analytics enthusiasts where we try and make sense of some topics in the analytics industry and have a little bit of fun along the way. I’m Dara, MD at Measurelab, joined as always by Measurelab’s longest serving Analytics Consultant Dan.
[00:00:35] Dan: Hey folks.
[00:00:35] Dara: Hey Dan. How are you and what have you been up to in this last week?
[00:00:39] Dan: In this last week? Well I am fine first of all, thanks for asking. I have been actually attending a course that you recommended, from a company called Hyper Island on building and creating effective digital teams it’s kind of cool, learning lots of stuff and yeah, hopefully take a lot away and put into practice. How about you Dara, what have you been up to?
[00:00:59] Dara: Uh, well in a slightly or somewhat similar vein. I’ve been looking at a website called Agency Hackers which was recommended by Steve, our Commercial Director. So it’s basically an online community for agencies, and it’s got lots of online events. It’s got something called a war room, which is peer-to-peer coaching for agency leaders. So it’s got a bunch of stuff on there really. And I’m just starting to look through it.
[00:01:25] Dan: Awesome.
[00:01:25] Dara: Okay, so what is our hot topic for this week Dan?
[00:01:29] Dan: Well I couldn’t think of a better name, but it’s a very descriptive name, and it is “better metrics for web measurement”. So yeah, no prizes for names there, but basically just going through a bunch of the pretty standard metrics that Google Analytics throws at you and digging into them a bit more around what they are, how useful they are. Can we make them better even?
[00:01:50] Dara: Yeah, I’m glad you stated the title. Think it makes us both cringe a little bit, but the actual content is very relevant. This is something we talk to clients about a lot. So everybody’s very familiar with the bog standard metrics that come out of GA, but there’s a lot of quirks, let’s call them that. There’s a lot of misconceptions. And also there are some tips and tricks on how to get a little bit more out of them, or maybe some alternatives to some of the common metrics that are a little bit more specific and insightful.
[00:02:17] Dan: So what we’re going to be really focusing on is any basic report in Universal Analytics. They categorize them, or I refer to them as the ABCs, the audience, acquisition and conversion metrics. These other kinds of common metrics that you may be familiar with or see all the time in any dashboard like users, sessions, bounce rates, conversion rates, those kinds of things.
[00:02:35] Dara: Okay, so where should we start?
[00:02:37] Dan: I reckon looking at the data quite often, the metric you see the most of is users or sessions. And I think that’s where we should start maybe pick into those and see, well, maybe start with defining what they are and are they useful?
[00:02:52] Dara: This is already bringing me back to a previous episode. The debate episode. Long live the session.
[00:03:04] Dan: Well, that’s a really good point. I mean, go listen to that episode if you want to get the argument. It was a civil discussion, but what we came down to is the idea that sessions are not super relevant in today’s age. No one logs onto the internet and has a session on the internet, it’s a bit more quick fire. Open my phone, close it. Open my phone later, close it again. So we have these kind of shorter snappier bursts on websites than we used to, or at least when the session was created. We also talked about the concept of a user being less and less reliable. Users are an analogy or a proxy to people, or at least they were, and they’re less and less relevant to that nowadays. And the users in Google Analytics is a first party cookie, so with cookies come issues around things like browser technology and resetting cookies and not accepting analytics cookies. Well, that’s not saying anything around if you share devices. So two people can be using the same device and be tracked as one user. If I have two browsers on my computer, then I’m two users. So users are a bad gauge for people and it’s just becoming harder and harder. So the long and short of what we got to is that users exist. They becoming less and less relevant to call them people and sessions exist, even though they’re a bit archaic. And I think what we did in that last episode is we, we met in the middle and we created this thing called session days. The idea of a session day be in how many days has someone visited the website over the period of time. So even if there’s 10 sessions in one day, we count that as one session day. If they come back the next day, then there’ll be a second session day.
[00:04:35] Dara: Even with sessions, which is probably the most commonly used metric from GA. Not everybody, knows that, for example, a session in GA will be triggered if a user is on the site at midnight. So if you’re on a website at 11:59 PM and then you’re still on a, at 12:01, it’ll actually count as two sessions in GA. The other trigger for new session is if you leave the site and come back through another marketing campaign. If the UTM information changes, then that will trigger a new session as well. So there’s a few nuances with the session metric and GA there are worth being aware of, because I think my argument is always that users don’t tend to be people and with all the issues with your cookie deletion and the unreliability of cookies, my argument is always that that users could be very far away from the actual number of users. So I tend to say, well at least with sessions you know where you stand. As long as you do know where you stand. So there are some points to be aware of in terms of how sessions are calculated within GA, or any other analytics platform for that matter.
[00:05:46] Dan: Yeah, exactly. And the timezone thing always catches me out, and I can imagine lots of other people, in that that’s the time zone you set in GA. So it actually is irrelevant whether the user browses their local time over midnight, it’s actually a flattened time zone. The same way GA flattens currency. Everything in GA is reported as one time zone, one currency. So it makes kind of sense when you think about it, like when I’m in a report and I say give me the data for the last seven days, what’re we actually asking of it. We’re asking it for the last seven days based on a specific time zone. We can’t just say the last relative seven days to every user, because we’ll probably have to wait an extra day before the reports are ready. And although that sounds actually like, it could be super valuable. Most people don’t want to have to wait a day or two before they can report on the data. They want to be looking at the data, firstly, in the morning or Monday morning for the last seven days or whatever, whatever that looks like. So knowing that GA flattens all this stuff, we know that sessions trigger going over midnight or coming back by or a different UTM, or just leaving the website coming back via payment gateway even. Looking at something like session day still hits the same hurdles, right? What does a day mean? But at least it avoids the UTM issues and things like that.
[00:06:53] Dara: Yeah, and of course the other time based trigger for a session is 30 minutes of inactivity. So for most websites, that’s not going to be a problem. But if you’re a content or media site, if you’re somebody like Netflix, for example, and somebody is sitting watching long form content and then they click onto a new page, that’s going to trigger a new session as well. So for your typical website, 30 minutes of inactivity is going to cover it. But for other websites, that’s not going to be enough and you might want to look at customizing that setting, otherwise you’re going to have some session inflation.
[00:07:21] Dan: And you can customize this of course, as you just mentioned, but people don’t. The 30 minute timer is the default out of the box, vanilla Google Analytics setup which is, as you said, fine for most cases. But the fact that you can change it doesn’t mean people will. And what we’re at a point now is that, you know, people have had their analytics tracking for 10 plus years in this very specific way. And then we’re saying, oh by the way you can change this it makes perfect sense, you should do it. Or they think they should, but they don’t because all of a sudden, they’re going to be comparing apples versus oranges for the last 10 years worth of data. And we’re basically saying to them that all of that is redundant because you can’t use that data as a comparison without doing some heavy lifting, you know, in BigQuery to recalculate all of that stuff. Although again, these things are things that Google provides a configurable option for, most people end up choosing not to configure it once they learn that it can be configured because of the historical data not being affected by the new change.
[00:08:18] Dara: So what’s your take on new versus returning?
[00:08:21] Dan: Oh, don’t get me started. Yeah. I mean, with all this stuff, we’ve just mentioned around cookie issues with browser technologies or with consent management. The new versus returning is, I don’t use it, put at that way. I actively tell people not to use it or put any stock in anymore. The reason I say that, again a user is a cookie. So if we use the literal definitions, the users in GA account of cookies that have been seen. So new users are new cookies that have been created. But if I’m on Safari that deletes my cookies after either 24 hours or seven days, let’s say it’s a grocery store and online grocery store. And I come back and do my fortnightly shop. Every single time I come back, if I’m using safari, I’ll be treated as a new user. Even if I’ve not actively clear cookies or I’ve not opted out of tracking or anything like that, I will be a new user. So yeah, really the new versus returning isn’t accurate and I don’t think he ever really was cause it was still always based on cookies. But now, as we stand in September 2021, I don’t think it’s good enough to put any stock in. So I wouldn’t base any decisions off the back of it. In which case I take it out of any dashboards or surfacing that to anyone in the business. I just take it out cause it puts an interpretation on the data that that could be completely misleading and it’s best to just remove it. If it’s not trustworthy, then I don’t bother.
[00:09:42] Dara: I totally agree. I almost forgot, I’m glad I didn’t. Something that I’ve seen too many times, lost count of this, is if they’re doing say weekly reporting and then roll up to a kind of monthly summary and the add users together. Which is the big no-no. I’ve had so many, it’s not, not, not always easiest thing to explain. And I’ve had borderline arguments with people in the past. where I’ve tried to.
[00:10:08] Dan: I’ve been the other end of that.
[00:10:10] Dara: It’s not easy necessarily for people to get their head around it. But with the user, it’s unique to the timeframe. So if you take, let’s say you have 10 unique users in a week, and then you look at the second week and you have 10 unique users. You don’t actually know how many of those users are the same across the two time periods. So the way to get the deduplicated count is to take the full two-week period whereas if you take it weekly and add them up, you’re going to duplicate and you’re going to over count the number of users. But it’s amazing how common it is for people to actually think you can do that and end up with a user count that’s even more wrong than the problem than it probably is to begin with.
[00:10:48] Dan: And it’s exactly the same principle when you looking in Google Analytics and it will say users and new users, but it doesn’t give you the returning user count. And it exists, you can go get it, but it’s just not in the default reports. So what some people might be thinking they might be able to do is take away or subtract the new users to get the returning users. But again, it’s exactly the same issue where you can’t do that because it’s unique. Just because I’m a new user, it doesn’t mean I’m forever a new user. So I can come visit the website twice in this period of time. The first time I’m a new user, the second time on a returning user. But overall, I’m still one user. So it’s one of those things that makes users tricky to work with, especially for people like us, that building dashboards and reports it catches people out. It’s just mindfulness, I suppose, that users are unique and you can’t sum a column, especially if you’re exporting through the API, Data Studio or pulling the data out into a spreadsheet and just pivoting. This is where it doesn’t recognize or appreciate the uniqueness. And we have to be mindful of that.
[00:11:47] Dara: Let’s move on to the B in the ABC. So the behavior section of metrics. I’m going to let you go with the first one, your favorite, your favorite metric.
[00:11:58] Dan: My least favorite metric in the world, bounce rate. Is that what you’re referring to? Yeah, so everyone’s got an opinion on bounce rate. And if you don’t have an opinion of bounce rate, you don’t know what it is. And I think that’s, that’s the way it is that when I, well both of us Dara, we’ve run how many GA training sessions in our lives, like many, yeah many many. And most of the time when you talk about bounce rate you get kind of quizzical eyes of like, what do you mean bounce rate isn’t necessarily bad? Or we get the question, the classic question of what’s a good bounce rate? Or why is my bounce rate so high? This is things that our clients genuinely ask. And just to start at the very top, a bounce rate is the percentage of sessions that contain one page view. And there is a nuance there I’m sure we’ll get into in a moment, I can see you smirking already. But ultimately that’s it. The bounce rate is the percentage of all sessions that they have that. And I think this is where we start thinking about the stuff that you mentioned Dara just now around the nuance around sessionisation. So if someone browses over that midnight timeframe, it resets the session. And as an example, let’s say I had 10 page views at just before midnight, midnight happened and I saw one page. Which means that now Google Analytics tracks that as two separate sessions, one non bounced and then one bounce session just after midnight. So my user will have a 50% bounce rate. You said about the 30 minute timeout, the Netflix example you said. If I click to the next episode after an hour TV show, that’ll be another session, but a bounced session. Because every time I go to a page, it will be a new session and a single page session because of the timer will elapse. So bounce rate can be really misleading thinking of it as a negative number. And this is where Google annoys me slightly with these kinds of numbers is they like to put up and down percent change at arrows and they color code them. So bounce rate going up is a red arrow and bounce rate going down is a green arrow. But actually, I don’t know if it makes much difference at a kind of aggregate site level.
[00:13:54] Dara: I, I agree. I think it can be. I think your, your last point’s a really valid one about it not being useful. And this is the same for for all metrics really. Especially percentage metrics, looking at them at an aggregate level is not really going to tell too much because that aggregate number is going to contain lots of difference in the case of bounce rate, lots of different page types. And for some of those page types, bounce rate might be a good thing. So if it’s a blog, chances are you want somebody to come on to that particular article, they find it by searching for something. They read the article top to bottom, they leave the site. They might come back another time again, they might interact further with your brand, they might end up buying something from you and that’s a great thing. But the blog itself is probably just going to bring them in quite high up the funnel. So you might expect it to bounce. If it’s a product detail page, you probably don’t want it to be a bounce. So not every page is the same. So it’s definitely worth going down a level deeper and looking at the different types of content on the site to try and get a better idea. So it’s comparing like with like, rather than looking at an overall bounce rate. And one of the features in GA that a lot of people don’t know about is the weighted sort feature. So it’s not available for every metric, but for percentage metrics you can actually sort the data by weighted sort. It’s really useful things like bounce rate or e-commerce conversion rate because if you sort from high to low, you get all of these low volume pages. So obscure pages maybe that you have on the site that have maybe one or two page views and a hundred percent bounce rate. if you flip it and sort it from low to high, you get a lot of maybe similar pages that have one or two page views and have 0% bounce rate. It’s not really telling you what is the, quote unquote, best or worst pages. If you use weighted sort, it takes into account the volume and the metric itself, so you’ll get pages that have a high volume, but also have high bounce rate. So you can actually hone in on pages that have maybe a bounce rate that’s too high. And again, you would want to do it by looking at like for like pages rather than just everything as a whole.
[00:16:02] Dan: Yeah, and the same could be applied for campaigns as well right. If you’ve got a ad campaign, let’s say you’re running a PPC campaign and you’re driving to a specific landing page. That is a big page that just has one button that says, click, click, click, click me. And they don’t click that, then you can assume a bounce rate is bad. But actually, you know, for the most of the campaigns you might be running, you might be taking them, you know, the better the marketing is, the more relevant the ads are, and the more likely you are to be able to get them to the right place of your website quickly. And I think actually this goes beyond just bounce rate, but actually what I’ve come across a number of times is companies that have got some kind of optimization campaign running, they’re optimizing user experience or they’ve launched a new website or they’re improving the user journeys or the efficiencies. But then they have this other side of them or the other part of the business that goes, our numbers are down our pages per session are down, our session duration is down, our bounce rate is up. But what that basically means is they’re doing a good job. You know, you’re trying to get people to where they need to go quicker. So less clicks, less page views and in this time. And maybe you even get them where they need to go straight away in which case they might bounce. And all of these things are positive things based on an agency or a team in your business is literally trying to do. You’ve got one team that’s trying to make those numbers go down, and another team that flags that saying there’s an issue when they go down.
[00:17:18] Dara: And there’s other issues as well with metrics like session duration.
[00:17:23] Dan: Oh, yes. Yeah. So going back to my favorite, bounce rate. The concept of a bounce really affects a lot of the numbers within Google Analytics, especially to do with time. First of all, Google Analytics doesn’t track duration, it can’t. So the way it works is every single time a page loads, it has a timestamp attached to that. And then the duration metrics in Google Analytics are just differentials between two different timestamps. So if I land on the website, uh, land on the homepage and that’s maybe at 10 o’clock on the dot and then I click the product page and I go there at 10:05. That is a five minute duration on that homepage. And that’s the way that Google Analytics assumes time works right. But what if I opened a page, went and made a cup of tea came back and then click to the next page. That entire duration, that delta of those two timestamps has been attributed to that previous page. Even though I wasn’t actually on that page or looking at the page or anything like that at all. Same as if you open the browser in a new tab and never even look at the page, it still tracks the page view it still starts that timer. But the thing with bounce sessions is that when you’re trying to work out time based on two numbers, what if there’s only one? So a bounce session is one page view, which means there’s only one timestamp, which means there’s no second timestamp to work out the duration on that session. And so what that means is that every single bounce session has a zero session duration. So when Google does averages, it doesn’t take that into account. It actually takes into account all of those zero session durations, where they’re bounced as well as all the non bounced durations as well. So when you’re looking at a 60% bounce rate, that means 60% of the numbers in your average calculation are zero. That’s skewing that number right down. If you think of them as outliers, rather than as actuals, that number session duration in most cases will be a lot higher at least perception.
[00:19:16] Dara: meaningless.
[00:19:17] Dan: Yeah, it’s completely meaningless. I mean, we, even, even if you were to take out that and you can do that through segments, you can create a quick segment in Google Analytics that just looks at non bounced sessions and you can go get that number super quick, super easy, and give you some, some semblance of meaning behind it. But you’ve even got the issue of the last page of any session that even if it doesn’t bounce. So whenever you get to the last page of a session, there is no next page view to work out the time difference. Because otherwise it wouldn’t be the last page of the session, that would be the last page of a session.
[00:19:47] Dara: Going back a step to bounce rate again, one thing that you can do to improve the usefulness of the bounce rate metric is to have um, if, if it’s relevant to have interaction events. So even though you mentioned Dan and strictly the definition is a single page view session, but interactive events will actually count against the bounce rate. So if you have a page with a lot of content on it and you are, so your example earlier was, is you’ve got this big call to action button on it. If you’re tracking, okay, presumably that button would link to another page, but if it was something like viewing a video or if it was downloading something, an action on the page that didn’t involve you navigating to a second page. And if and if that was the action that you wanted from the user, then if you event track that action and set it to be an interactive event, then that won’t count as a bounce. And that would give you in that case, a more meaningful bounce rate.
[00:20:43] Dan: Yeah, absolutely. I’m always in two minds with the interaction level events. And I know you’ve heard this before a hundred times Dara, but I’m in the camp of keeping the metrics consistent, whether we like it or not. And it goes back to that thing of why people don’t change the session time out of 30 minutes, is because all of a sudden your historical data and all the numbers that use that are going to be different, going forwards. And it’s absolutely something you can configure and something you can change, but the idea of something being different and everyone’s perception of a number being called the same thing and changing the definition is really hard to grasp and educate everyone internally that is now different and stop comparing it. Let alone, if someone just logs into GA six months later and not realizing. I always just steer away from that whole can of worms and just leave bounce rate for what it is, which is single page sessions. And just understand where it’s relevant. Like you were saying around you looking per page and just moving on.
[00:21:34] Dara: Yeah, fair point. I mean, my hypothetical argument, It was a brand new website. Never tracked before. No, you’re you’re right. Anything that’s going to change a fundamental metric where people have an understanding of it, that does have to be done with caution. But it is an option, and I think if everybody is using the data understands that, then that can be an option to consider. There’s something actually, I only just realized recently, and I had misunderstood for quite some time is if you’re using the unique page views metric in GA, if you filter that metric in any way, it’s not actually duplicating the number. The unique pages should be, it’s not quite like for like with sessions. But if you’re looking at page data and you want a rough measure of how many sessions viewed those pages, you can use unique page view metric. But if you use the filter within GA, so let’s say you only want to look at unique page views for pages that contain slash blog. The number it gives you in the score card isn’t actually deduplicated.
[00:22:34] Dan: It’s what you were saying about the uniqueness of users, right? It’s exactly the same, just a different scope. So we’d taken the users at a session level of unique, and now when we’re at a page level, the sessions become unique. What they are doing is they’re summing it, which they shouldn’t be doing. This is the thing that introduced me to the whole world of data scopes in GA, I was asked way back when to get how many sessions contained the homepage. And I thought, you know, as the question is written down, I will put page and sessions in a report together and then filter for homepage. But that’s not how GA works. Actually, if you include a session-scope metric like sessions and a hit-scope dimension like page, what It’ll actually do is look at the first hit of the session. So what that actually gave me by combining page and sessions together is sessions landing on that page. Which could be fine, but not why I wanted to get. So in my head, I expected it to do what I thought it was doing, but it wasn’t. And I had to use unique pages to do that. I didn’t think it would touch on GA reporting scopes just in this conversation. But, um,
[00:23:36] Dara: It finds a way to the surface.
[00:23:38] Dan: It always does.
[00:23:40] Dara: I’ve one more or maybe a two-parter, a very under utilized metric in my view is the page value metric. Looking at the page value metric can be quite nice yardstick to measure the performance of certain page types. You can use that page value metric in combination with something like exits. So you can actually work out a potential, it’s not highly scientific, but it’s a good way to work out how much potential lossed revenue you’re leaking from that page on the site. So if you multiply exits for that page by the page value, it’s approximately or you could argue it’s revenue that you’re losing from that page. So those pages with the highest number when you multiply those could be the ones that need the most attention and might need some optimization efforts.
[00:24:25] Dan: Yeah, that’s a nice metric to use in CRO or at least, you know, we’ve seen using CRO analysis. Not the most glamorous of names, leakage, but it’s still valuable nonetheless. Um, that the only thing with page value is to ignore the standard pages through a purchase. By the way, taking a step back, page value is going back to my favorite topic of attribution. It is a linear attribution of the revenue across all pages of the session. So you see 10 pages in the session and you got a 10 pounds of revenue, then it be one pound per page, and that’ll be your page value. And it does that across every session and aggregates up. But thing like the basket page, the checkout and the thank you page, they are going to be in every single purchase. So you can almost ignore the page value from those specific pages, because it’s going to be quite high because you would imagine that every single session that makes a purchase has to go through those pages. Where it becomes super relevant is things like content pages and non ecom focus pages. Especially landing pages or category pages where you don’t necessarily need to hit those pages to make a purchase.
[00:25:28] Dara: I’m amazed, we’ve talked for this long on the behavior section.
[00:25:31] Dan: I know. How about you move us on and talk about your favorite, conversion rate.
[00:25:35] Dara: I do love the e-commerce conversion rate, probably the metric that gets the most questions. Number one of which is always is my conversion rate good. And it’s, don’t want to say everything’s like bounce rate, it’s it’s not like bounce rate at all, but it’s similar in the fact that it’s session-based which is problematic. It’s not taking into account the fact that users will rarely convert in their first session. There are probably some rare cases, some very lucky businesses where they tend to get people converting in the first session. Usually that doesn’t happen. So if you’re looking at conversion rates and this goes back to our last week’s topic around attribution. If you’re using a last non-direct click attribution model, which is the default in GA, you’re going to devalue channels that tend to be driving people into the funnel rather than getting them over the finishing line. So session-based e-commerce conversion rate is problematic and it’s confusing as well because a lot of, um, at least I’ve experienced this. A lot of businesses will be looking to, use a user conversion rate. And I’ve found that often people even confuse the two and think it is a user conversion rate.
[00:26:50] Dan: Yeah, well, one of the two numbers is session-based right. And that’s the thing that gets me, is that it’s not even a truly session-based metric. You’re taking sessions and transactions, and transactions isn’t session-based. If I purchased three times in one session, that will be a 300% conversion rate because there’ll be three transactions and one session. The issue I come up against with things like the out of the box e-commerce conversion rate in GA is that it’s not fully in one camp or another. So I’ve seen it many times where you have a higher conversion rate than a hundred percent or a very high conversion rate that’s questionable, and it’s because of things like this. If you have a website where you are likely to make multiple purchases, let’s say you are buying tickets to a gig, but you can only go through one at a time. So you want to buy five tickets and you have to go through the checkout process five times. It might be a super quick checkout process, but you’re still transacting five times and you get a very high conversion rate. So where I come down on this is, you’re absolutely right, sessions are questionable and whether or not we use sessions is the key, but either use sessions or don’t. So let’s look at converting sessions rather than total transactions. So how many of my sessions were successful?
[00:28:00] Dara: Well I agree with that, what do you do with a 300% conversion rate, retire I suppose?
[00:28:05] Dan: Yeah. Yeah. Your job’s done. I think there’s, for me again, it goes back to the idea of it not set into either camp. So I want to know, did the click I paid for potentially, was that successful. And I don’t think that the e-commerce conversion rate in GA right now does that. I think we need to create one or use a goal. Goals are actually slightly different to e-commerce conversions in GA because they are session-based. Let’s say my goal is signing up to a newsletter. If I signed up to the newsletter three times, for some reason, in one session, that would be one goal conversion. And so goals are truly a session based metric where we can see. X percent of my sessions were good or were successful. And you can apply that to a user as well. How many of my users did a thing I wanted them to do as a percentage. So, yeah, not really anything to do there. It’s just about taking what it is. Understanding that can be weird nuances with there being greater than a hundred percent conversion rates in GA or at least e-commerce conversion rates, and then even creating your own metrics as a better or replacement or a comparison that are fully within one scope or another, there is a session-based conversion rate and there is a user based conversion rate. Neither unfortunately are what GA has by default.
[00:29:16] Dara: That’s pretty damning.
[00:29:19] Dan: No, I do. I don’t want to sound super negative with this. It’s just the thing I always say when I’m talking to anyone about GA, especially if they’re relatively new to it or not had any formal training, is that don’t assume that it’s doing what it says on the tin. And that’s the worst thing you can do. It’s not saying that these numbers are unuseful or unusable. It’s just that if your interpretation is different than you might be applying it in the wrong way. And so just having an understanding that these metrics aren’t maybe what you have interpreted them to be, is actually all you need to do. You can then decide on what’s useful, what levels are good, what levels are bad, how your business applies and adopts those. That’s only something you will know how to do. All we’re saying is the facts that Google Analytics tries to make it super simple by renaming numbers to sound more relevant. But actually sometimes that can be misleading.
[00:30:09] Dara: Yeah, no, I joked about the damning assessment, but you’re right. And the main thing is, and this is one of the, I guess this is one of the ideas behind this particular episode. It’s not just about there always being necessarily a better alternative, but it’s even understanding the metrics that you are using now because we see them misapplied or misunderstood a lot of the time. As you said, it’s knowing what it is that it’s actually telling you, and it’s not always what it seems.
[00:30:34] Dan: And looking at the transaction and revenue counts. I think you mentioned it in the GA audit episode Dara, around how they can be inflated somewhat. If you have a purchase confirmation page that you can access multiple times across multiple days across multiple sessions. You actually be tracking that purchase in Google Analytics over and over and over again. But ultimately just be mindful that what we’re actually looking at here, this isn’t your balance sheet. I wouldn’t be sending out your orders based on the data GA collects, there’s all sorts of nuances with web tracking. But it should be close enough, within around five to 10%, I would say. But assuming that you’re relatively happy with those numbers, then we good to work with.
[00:31:08] Dara: Yeah. And therefore useful for trends and kind of high-level analysis, but not for financial reporting.
[00:31:15] Dan: No, no, no, no, definitely.
[00:31:16] Dara: Okay, I think we’ve covered a lot. So to summarize, I would say there’s nothing wrong with using the bog standard metrics that are provided from GA, but it is worth taking the time to understand some of these nuances, of being aware of some of the misconceptions that you might have that we’ve probably had and we’ve seen people have. So it’s understanding those metrics, what they mean, what some of the limitations are or how those metrics are calculated. And then also looking at some of the additional metrics that you might not be aware of or not just metrics, but features. So things like weighted sort, things like using segments to exclude bouncing sessions if you want to look at that things like a verage time on site. Looking at calculated metrics like user conversion rate, for example. So there are some extra metrics that you can create or find or manipulate in some way. And then there’s a whole world of things that you can do if you have that data in BigQuery. But first and foremost, take the time to have that understanding of the limitations with the existing metrics in GA.
[00:32:21] Dan: Yeah, it sounds good.
[00:32:23] Dara: All right Dan, w hat have you been up to outside of work? What fun things have you been doing?
[00:32:28] Dan: Well, what fun things? So me and my wife, we’ve been watching Sex Education series three, and it’s awesome. We’ve managed to binge it in one weekend, as is the way with any new Netflix TV show when they release it every year or every 18 months, whenever they do it. The same thing we do with Stranger Things, for example. But if you haven’t seen it Sex Education is well worth a watch.
[00:32:46] Dara: I’m waiting for Ozark. That’s the next one for me.
[00:32:48] Dan: I stopped Ozark, I should get back into it.
[00:32:51] Dara: It’s so good.
[00:32:51] Dan: So Dara, when you’re not weighted sorting reports, what are you doing to wind down?
[00:32:56] Dara: What else would I do? Well, what better way to wind down. Eh no, mine is also TV related. Grand Designs has just recently come back on, new series. I know you’d like it to, but you had your choice and you went for a Netflix series. Yeah, Grand Designs I’ve watched, I don’t know, not not every series cause there’s been 400 or something. But I’ve watched a lot of them. And especially more recently, I’ve been an avid watcher of it and it’s great. It’s just so much fun. There’s drama, you think It’s not going to work. They overspend, they think it’s going to take them six months to build the house and it takes six years. It’s really good fun, and something I’d love to do one day, I’d probably end up grabbing that. But in the meantime, I’m just going to keep watching other people struggle to build their own houses. Okay, as always, you can find out more about us over at measurelab.co.uk. Or you can get in touch via email at hello@measurelab.co.uk, or just look for us on LinkedIn and feel free to ask us any questions or suggest a topic for us to discuss. Otherwise, join us next time for more analytics chit-chat, I’ve been Dara joined by Dan. So it’s bye from me.
[00:34:12] Dan: And bye from me.
[00:34:14] Dara: See you next time.
Daniel Perry-Reed
Subscribe to our newsletter:
Further reading
Customer Data Platforms (CDPs)
Grand Designs: “The Warehouse to CDP conversion”
