Talia is a frequent keynote speaker at marketing conferences, teaching conversion optimization and growth on stages such as Google, Unbounce, MozCon, GMIC, CXL live, Search Love, Learn Inbound and many more.
She is the Co-founder & CMO at Banana Splash and was recently listed as one of the most influential voices in conversion optimization.
Latest posts by Talia Wolf (see all)
- How to optimize your site with user testing - June 3, 2018
- Are heatmaps actually effective? Interview with HotJar Founder - April 16, 2018
- Social Listening: How to Increase Conversions Using Social Media Insights - March 18, 2018
A successful conversion optimization program start with data.
There’s simply no going around it.
And Google Analytics can be frustrating at times.
We’ve all been there… Trying to fish out only the most important data, throwing loads of time, resources (and let’s face it, sweat and tears) into identifying where that leak in the funnel is, only to discover (after you’ve fixed that pesky leak)… NOTHING.
No change at all.
This happens for various reasons…
- You’re tracking may not be set up correctly so the data you’re collecting isn’t accurate
- Or the leak you found does exist, but it’s not as significant as other leaks in the funnel that can have a real impact on your bottom line if fixed.
There are many other reasons you may not be aware of.
Google Analytics is the first step to discovering where the leaks in our funnel are and prioritizing them.
To help make sure you’re getting the most out of Google Analytics and constantly improving, we put together a live google analytics training session with Annie Cushing.
Annie is one of the top experts in the world for Google Analytics and has a lot to of incredible insights to share with us.
Here’s what we’re going to learn today:
- What to track (and how to ensure the data you’re collecting is reliable)
- The top mistakes marketers make within Google Analytics
- The best reports to help you find meaningful insights
- How to prioritize A/B tests with the data you’ve found
- And how to find A/B testing ideas right within Google Analytics.
Getting Started with Google Analytics
Question #1: Many people find Google Analytics, overwhelming. How can they get started?
So first, just to address the overwhelming part, if you feel overwhelmed by Google Analytics, you’re in very good company. Because the Google team has iterated on it so quickly, especially in the past three years, especially since Justin Cutroni took the wheels. So they’re iterating on it a lot, they’re improving a lot. But we’ve almost hit a point of diminishing returns where there are so many changes and so many new things that it can really be easy to get tangled up in it all. So we’re going to address later in the webinar some resources if you want to learn more about how to really rock Google Analytics.
Google Analytics Channels Report
So if you’re new to using Google Analytics, a lot of people think that all the standard reports you see on the left of your dashboard is Google Analytics. However, these are really just templated where Google didn’t want you to start with a blank screen. So they gave us some templated reports. The real magic with Google Analytics happens in the custom reports, where you can go and get exactly what you want. If you haven’t worked with custom reports before, check out my video that takes you by the hand and walks you through how to create custom reports.
But you can still look at all the standard reports on the left as a general starting point.If you don’t look at anything else in Google Analytics, there is one report that I recommend, that in my opinion is hands down the report to start with: Channels report. The channels report is under acquisition → all traffic → channels.
When you go into this report, the default channel grouping. Up until 2013, we didn’t have this channels report. We had to look at mediums or maybe the source medium report, which is super noisy. I don’t like the Source/Medium report at all. Facebook alone causes so much drama, because Facebook can show up as up to 16 different referrals. But people don’t know that, so they see the first instance Facebook and think that’s their Facebook traffic.
So in 2013, Google grouped similar mediums together and gave us this report. One of the things that I absolutely adore about this report is depending on what you click through to, you are going to have different options.
The other thing that I like starting from the channels report, is that you get really valuable metrics like bounce rate and conversions. So if you’re running ecommerce on your site, you have ecommerce data. If you’re primarily using goals, then you can switch conversions here to all of your goals or a specific goal. Whereas none of the template social reports for whatever reason have conversion data. They also don’t have bounce rate. I think that aside from sessions, bounce rate and conversion data, are really some of the most essential metrics that you really need.
So to summarize, if you’re going to start with anything in Google Analytics, start with:
- The channels report – Hands down.
- Then, look at Site content. Go to behavior → site content, particularly all pages and landing pages. The difference is that if you go to ‘channels’ → ‘organic search’ and then switch to ‘landing page’, you will be looking at your top organic landing pages for that particular channel while here you can see all of your landing pages together (and you’ll have all of that conversion data just like we saw in the channels report, it’s all aggregated together).
Landing Pages vs. All Pages in Google Analytics
The difference between landing pages and all pages is that in landing pages, you’re going to see conversion data. In all pages, you’re not going to see conversion data, because the all pages report is showing you individual hits. So if someone comes to your site, they look at let’s say three pages. The landing page is one more descriptor of that overall session, whereas you can’t give credit to individual pages for the conversion but you can for a landing page, because it’s a descriptor, just as much as:
- What browser did you use?
- What country are you coming from?
- What source medium did you arrive on the site?
Simply look at the number of conversions you have and what pages were involved in those particular conversions and the pages that were included in the session that resulted in the conversion will have a higher page value.
For example, my home page is going to have a higher page value than my blog page. You can see here the home page has a page value of $1.73. My blog landing page has a page value of $0.31, whereas my resources page has an overall page value for this particular time period of $3.93. So that tells me it’s a good idea to send people to my resources page.
Is your Google Analytics Data Reliable?
Question #2: How can we make sure that the data we’re collecting is reliable?
There are a few simple ways to do that:
You always want to make sure you’re checking your Google Analytics against your backend databases. They’re never going to match exactly, and usually, Google Analytics is going to show less than your backend database. That’s because there are a lot more people choosing a Chrome extension or other extensions that don’t allow sites to track them.
Example: I just ran into this literally just last week with the client. Google Analytics was showing significantly more conversions and so we actually found that on some of their pages in some categories, of their site, they had migrated to Google Tag Manager, but they were still firing the Google Analytics beacon on the page as well as in Tag Manager.
So if you ever find that Google Analytics is showing more than in you backend databases, you most likely have something double firing. Either that or maybe you just have a lot of returns and you’re not using the measurement protocol to keep track of refunds.
There are a few tools that I really like to use to test:
- Google Analytics Debugger – The way it works is if you open up a page and then go into ‘inspect’ you’ll be able to click on console you’ll be able to see in here if the beacon is firing and you can see here the beacon was sent.
- Real time reports – This is especially valuable if you’re testing events, conversions or want to see if you have double firing issues. Those will show up in this report. You just first have to identify yourself as a user. Here’s a detailed video on how to do exactly that.
- Tag assistant is another Chrome extension from Google, very helpful if you’re using Google Tag Manager.
I go back and forth between debugger and the real-time reports. I really like the real-time reports, because I can see things as I’m doing them and that just gives me the most confidence.
Google Analytics Metrics Marketers Forget to Track
Question #3: What are the key metrics marketers forget to track?
#1: Conversion metrics:
Obviously, if you’re using ecommerce tracking, I use ecommerce where the new data is a pretty huge determiner of the success of my site, like blog post and stuff like that.
#2 Bounce Rate:
Bounce rate is really as Avinash Kaushik calls it, the golden metric.
- One caveat about bounce rate is you need to keep in mind the type of website you have. For example, blog posts by the nature of what they are, intend to have very high bounce rates, unless the movement between your blog and your money pages is really fluent.
- Particularly visitors coming from social channels, they bounce at a very high rate.
- For most sites, organic traffic is probably going to have the lowest bounce rates along with organic traffic and email, because people are in a different mindset than they are when they are on social platforms, for example – Facebook. When I’m on Facebook, I really don’t want to leave Facebook, unless it’s going to make me laugh. So if it’s something especially work-related, I’m going to just save it to park it and read it later.
Especially going back to your landing pages report, whether you’re isolating landing pages from a particular channel or looking at you overall landing pages, bounce rate is really the key, especially if you are paying for the traffic.
So I have much lower tolerance for a high bounce rate if I’m looking at traffic from paid search especially. Even with paid social, you still going to tend to have a higher bounce rate, but with paid search, you literally control everything and you are buying the keywords. You’re paying someone to send traffic to your site, so if I see keywords with a high bounce rate, landing pages with a high bounce rate, then that raises a red flag and I monitor bounce rate very closely.
Another really good thing to do is if you’re tracking revenue or you’re using goal conversions, look at the pages that generate high revenue but also have a high bounce rate → That’s automatic gold for testing.
You know that people want to throw money at you when they land on this page, but a lot of visitors are leaving. Test that page and see how you can squeeze more conversions out of it.
Top 3 Mistakes Marketers Make in Google Analytics
Question #4 What are 3 common mistakes marketers make with Google Analytics?
#1 Using mediums that Google doesn’t recognize.
This is the biggest mistake most marketers make.
Google has a list of mediums that it recognizes and endorses. There’s in fact even a list where Google actually tells us exactly what mediums it recognizes and what it puts into different channels. But if you go again back to the channels report (I don’t have a lot of traffic in mine, because I hawk over this, like nobody’s business) in this ‘other’ channel, nobody knows what this is, but this is a kiss of death in Google Analytics.
If you use a medium that Google doesn’t recognize, it’s going to put that traffic into this ‘other’ channel. So this is basically Google’s junk drawer. I can tell you, even the top publishers like Jezebel.com, Buzzfeed and gawker make this terrible mistake and get this wrong. I can look at the tags that they share, look at the links that they share on social media that are tagged and my eye goes right for UTM_Medium, and if it’s a medium that Google doesn’t recognize, I know it showing up in that channel and whoever is managing this is (usually social media that gets hosed the most with email coming in solid second), all of that traffic is just going to get routed to this ‘other’ bucket that no one looks at. Because you wouldn’t think like, “oh, my email traffic might be showing up under other”, or “my social traffic might be showing up under other”.
So one thing that I highly recommend that you do is go into Google Analytics → channels report → ‘other’. Now you’re always going to have some, because people will tag links that they put in their newsletter or their email or whatever and they’ll tag links to your site (even if their not supposed to, that’s like painting your neighbor’s fence). So you’ll have other people’s tags show up in my other.
If you go to “other” → then click on other and change it to medium. This will tell you what mediums Google doesn’t recognize.
When I’m doing an audit, I will go through these line by line and I’ll see ‘oh, retargeting, Google doesn’t recognize that as a channel’, it absolutely should, it should have a default channel for every targeting, it should have a default channel for a paid social (I’ll get to that in a minute) but it doesn’t.
Also one thing that is very important to know, is that sometimes mediums like email and display can be found in the ‘other’ channel because Google doesn’t recognize the medium if it’s capitalized. It is a silly mistake on Google’s part in my opinion, because a lot of sites have traffic showing up in ‘other’ simply because they capitalized email or display.
So that’s the number one thing across the board you need to do, go through this list of mediums and pick those out because it really can cause significant mayhem. For example, I did in analytics about 4 months ago, and their ‘other’ bucket was their second most trafficked bucket. So that’s just awful to see.
#2 Working from Google’s default channels.
As I mentioned earlier, there’s some really important channels that are missing, for example Google doesn’t have a channel for paid social.
Google recommends using the medium ‘CPC’ for paid social. I don’t agree with that at all, because if you tag your social ad’s medium as ‘CPC’, it’s going to show up in the paid search channel, however an ad on Instagram is not paid search, an Adwords ad is paid search, because that actually shows up in a search engine. Check out this post on how I recommend to track social ads (which is to use a custom medium → and then create a custom channel, which I show you how to do to track paid social.
Some other ones that are missing as I mentioned; retargeting.
You can track retargeting as display, I just don’t recommend it because retargeted traffic behaves very differently from display traffic. Display traffic is almost throw away traffic, It’s good for branding but it doesn’t generate a lot of conversion. Retargeting, especially done well, can be one of your top revenue generators so if you put it into your display channel, you just have to make sure you site map that data, so you drill down into that channel again.
My personal preference would be to create a custom channel for your retargeting. But then, also creating custom channels for whatever is important to you:
- If you’re a publisher, you might have syndication agreements with other larger publishers.
- You might want to create a partner channel for those syndicates.
- If you’re doing a lot of guest blogging on different sites, you could create a channel of just the sites that you have guest posts on so that you can track that traffic differently from joesixpack.com for example.
- If you’re doing a lot with PR, you may want to have a PR channel. With that, only include things like press releases and reviews and things like that.
- For me, I have a lot of Google Docs that I created for marketers, so I have a Google Docs channel, because I use a custom medium that I just tagged it UTM_Medium=GDOC. Then I created a custom channel for Google Docs.
- I also have a channel for templates, because I create different templates that marketers can buy and use. A site auto template and analytics auto template, a marketing strategy guide all of that traffic would show up at my site as direct, because anything coming from a work doc or PDF doesn’t pass referral data, so that traffic all shows up as direct. Whereas by tagging it, now it’s going to show up in my templates channel. So that’s really important.
#3 Not removing query parameters from your URL’s inside of Google Analytics.
So GA gives you the option to filter out query parameters. Query parameters, you see them a lot on ecommerce sites, you can sort price from high to low or from low to high. So it might say price equals low or size equals large, color equals orange. Those are all query parameters. When you get a bunch of those query parameters in your URL’s, they can generate a significant amount of duplication in your content reports.
By content reports I’m talking specifically about:
- “All pages” report
- Or your ‘landing pages’ report
- Also the ‘exit pages’
But especially these 2 pages (All pages and landing pages), because what most marketers will do is if they’re looking for a particular page, they’ll scroll down, they’ll find their page and they’ll just click on it and say okay, “here’s the traffic that I’ve had to this particular page” and not realized that if you have query parameters reporting, you could have tens of different versions of that page. I have 1 client; they had more than 1000 versions of their top money pages.
One way to find those, is if you just drop an equal sign in your all pages report, you can start to identify query parameters and then you just go into your view settings here and there’s this option here ‘exclude URL query parameters’. Go through the whole process of picking which query parameters you want to exclude and which one’s you want to keep.
How to Find Leaks in your Funnel with Google Analytics
Question #5 What are the best reports to use to find leaks in your funnel?
Talia: What are your favorite reports for finding the leaks in your funnel and identifying what needs to be optimized in Google Analytics?
#1: If you’ve moved to Universal and you’re tracking e-commerce, I highly recommend enabling enhanced e-commerce. There is something called a ‘shopping behavior analysis report’, it will show you where the drop off is occurring.
If you are using goals, then there is a little known report under ‘goals’ called the ‘reverse goal path’. This is your goal completion location. It will show you the previous steps and the most common previous steps.
If you want to use funnels for your goals, go into ‘admin’ → ‘goals → choose the goal you want → “Goal details” and turn on “funnel” view.
If you have that funnel set up, then you’ll actually see the funnels in the funnel visualization. One a huge caveat here is it’s pretty buggy, you’ll see people supposedly leave in the first step, but then come back in step 4… You know that that can’t really happen so most likely what’s happening is your Google Analytics tracking code might be too low on the page or something like that and they’re just clicking through to the next step before Google has a chance to register that piece of the funnel.
Prioritizing AB Tests in Google Analytics
Question #6: Once I found the leak in the funnel, how do I prioritize what’s important to fix and optimize?
Talia: Many times (and I’ve seen this a lot, it’s really frustrating) you can do all the research, find the leak in the funnel and then spend a ton of resources on fixing that leak which generates absolutely no change. This happens because you’ve dedicated your entire time to fixing something that won’t really have any effect on your bottom line. What would you recommend doing in order to prioritize?
I’m so glad you asked this question, because I’m not sure about the CRO space as much as the search space, but in general among marketers, automation is all the rage right now – identifying issues and automating the changes as much as possible.
But along the lines of what you just talked about, when you’re talking about testing, that can be very detrimental, because you could click all of these resources into rolling out these global changes and like you said have little impact.
So what I like to do is pull a universe of pages: pages that I’m testing or advance or whatever it is that you want to test. I mostly prioritize by the pages generating ultimately the most conversions.
But I like having smaller test groups that you can test vis-à-vis and see if that moves the needle. Create a custom report and throw that into a dashboard, so that you can monitor it.
Practically, we can talk about A/B testing, multivariate testing on an academic level, but the reality is when I’m working with a client, if they run any kind of test, it is a no-brainer. I’m still working with a dashboard tool, like Tableau or Klipfolio or something like that, then I set it up in their tool.
But if not, then I just take whatever (we’re almost always dealing with the customary report or a filter report), go into ‘Behavior’ → ‘Landing Pages’ to identify the pages that we’ve actually changed → then change the filter to: ‘landing page’ matches RegExp and just put it in the individual pages and separate them with the ‘pipe’ character (which just means ‘or’) and then apply those changes.
Bounce Rate Optimization
What is considered a high, average or rock star bounce rate for a website?
Social Traffic Bounce Rate:
A good bounce rate when we’re talking about social, can run as high as in the 80’s, especially if they’re looking at blog pages. So I’ll generally segment out my blog pages from my money pages. I even have done that with something called ‘content grouping’, so that you can look at your traffic differently, depending on the pages. Here I have my home page, I have blog, sales and then my member content affiliates:
On my site, in the 80’s is very normal when going to a blog page because they come, they read the blog post and leave. When they land on my resources page, that bounce rate even from social is going to come down significantly.
So blog pages especially from social, it’s almost like it seems like it’s thrown away, but it’s really not. Because each time they come to your site, they’re building hopefully more confidence in your brand, they’ve made a mental note, they’ve started to identify ‘oh, Annielytics talks about analytics’. So it’s very valuable from a branding standpoint.
PPC traffic Bounce Rate:
With any kind of paid search traffic (legitimate page search traffic not Facebook), I don’t like to see anything above 50% bounce rate with fairly rare exception.
So I will just hawk over particularly the keyword report, because I’m looking to see ‘okay, if I have a particular keyword that’s sporting a 70-80%, sometimes even 90% bounce rate, that’s just money I’m throwing away’.
With paid, like I said, you control everything so the problems could be:
- That might not be the best keyword
- Your matching is too broad.
- Maybe you’re using broad matching, which is a great way to tear through your budget in no time.
- Or maybe you need to optimize the page better.
All these, definitely warn greater analysis.
I’ve seen destination URL’s (which is not a landing page for a PPC) have really high bounce rates. To see that go to ‘Acquisition’ → ‘AdWords’ → ‘Final URL’s’, export that list and run it through Screaming Frog and let the client know which pages are redirecting. If they’re redirecting, the Adwords parameter can drop off so you’re reporting is going to be messed up.’
But also you would be absolutely amazed by how many paid landing pages are 404 and that you definitely don’t want to happen. Everything in between display tends to have a very high bounce rate, it’s all branding for the most part.
To summarize, in social, the more customized you get (like people coming from my custom channels → they’re coming from ‘my templates’ channel) the more they tend to have lower bounce rates because they’re in the midst of trying to do their first site audit or their first Analytics audit. They may come look at that particular page to learn more and come back later. I definitely get a lot more repeat traffic from my custom channels. I would say, overall, the one’s with the lowest bounce rates should be:
- Paid search.
- Direct – tends to have a much lower bounce rate.
- Email – tends to have a lower bounce rate.
- Organic search – tends to have a lower bounce rate. I would say anything under 50-60% is still within the acceptable range for organic search, because you have very little control over that.
Definitely anything you’re paying for, I would caution against paying for campaigns that have high bounce rates.
Spam and Bot Traffic Optimization
How do we remove spam traffic from GA?
One thing to keep in mind is email tends to be a really big culprit in sending direct traffic and it’s the most important medium to make sure you add your campaign parameters to. If you don’t add these little tags to the end of the URL it won’t work (you have to make sure you have at the bare minimum, the medium and the source).
But why tag a medium’s words and not give it a campaign name? It’s like having a kid and not giving him or her a name. If you don’t tag every link from your email that points back to your site, you are going to have a very high percentage of that show up as direct.
The reason is that no referral data gets passed from desktop apps. So if you’re using Mac Mail or Outlook, referral data doesn’t get passed, so it shows up as direct. Mobile apps show up as direct; no referral data gets passed. And now even email providers, such as Gmail and Yahoo! that default to a secure server, also don’t pass referral data.
So only a tiny percentage of email traffic will show up, but it won’t show up in your email channel, unless you actually tag it. You can only use the medium: ‘email’. You can’t use the medium newsletter or something like that.
If you want to use a custom ‘medium; you’ll have to go into your ‘Channels’ (your custom channels) and customize them. For example, go into, your ‘Email’ channel and in addition to ‘System defined channel matches email’, which is set as default, also accept ‘anything’.
Side note, the reason I did this is to fortify against uppercase. I suggest adding ‘Medium matches RegEx email’, even if you don’t use any RegEx so that it’s not case sensitive. Now you could say: ‘or newsletter’ and from this point on you would be able to use the medium of newsletter and it will show up in your email channel.
So that aside, spam is a huge issue. It’s really come to the forefront recently, since a certain spammer who got mad at Google indicted everyone’s reports with language spam (if you go to audience → overview you’ll see your languages) so in most people’s reports, you’ll see all kinds of crazy things like Google.com and messages about how Trump is the best. All kinds of crazy Russian jumble.
Two things to know:
- When Google Universal came along, they gave us this option for Bot filtering under ‘Admin’ → Select your particular view → ‘view settings’. But for whatever reason, they didn’t select it by default so no one knows it’s there and they don’t select it. I don’t know why they did that, because I can’t imagine anyone saying ‘what the hell, what happened to all of my bot and spider traffic’? So this is a no-brainer. Definitely go into every view that you use and select this.
- Now to protected from language crazy spam, identify the hostnames in your Google Analytics account.
- Step 1: The best way to see the host names that are reporting is to go to ‘Audience’ → ‘Technology’ → ‘Network’, change the primary dimension to ‘hostname’ and this will show you all of the subdomains that have your Google Analytics tracking code on it (this is also a good way to identify sites that are scraping your site). Then you’ll see I only have www.annielytics.com and this one here because I have a filter in place. Before I put this filter in place, I had many more host names showing up here because of sites scraping. Also lots and lots of spam sites.
- Step 2: Create an ‘include filter’: Go to ‘filter’ → give your filter a name → Then go to ‘custom’ → Choose ‘include’ → Then under the filter field, just choose ‘hostname’ and tell Google ‘this field here is sensitive to RegEx.’ This lets Google know anything from annielytics.com is okay. If you had other domains or subdomains that you also wanted included in your reports, then you just use this pipe character here and put those as well.
So if you’re using Salesforce or something, you might have a Forex domain or something like that. So you would just drop those in here and click ‘save’ and from this point forward -it won’t affect your historical traffic- that will get rid of your spam. If you find that even with doing this you still have issues with language spam or whatever else spammers come up with to spam (because that’s their job description) then you would just add a filter.
Depending on what it is, you might want to use the exclude filter. If it’s language spam and your host name include filter isn’t’ doing the job, then you could just identify ‘anything that contains Trump’. Whatever they choose to spam or exploit, this is the place to use ‘exclude’.
The ‘include’, when it comes to the hostname- I think it’s a much better option, because they’re always going to be coming up with new spam domain. So you don’t want to have to try to keep on top of that. It’s just easier to say ‘only include in my reports traffic that comes to Annielytics.com’.
Talia: This has been great and really in-depth. Thanks Annie!
Want to learn more?
- Follow Annie on twitter – @anniecushing
- Or visit her site: annielytics.com
Watch the full Google Analytics Training webinar:
Powered by Facebook Comments