I put this image together thinking I could come up with ten things I hated about Google Analytics (GA), but I couldn’t get all the way up to 10. I love GA and recommend it or Google Analytics 360 for pretty much any size of organization unless you really absolutely must have some specific Adobe Analytics functionality. I also really wanted to keep the 10 things graphic because it’s so fancy.

I did manage to put together five things that I hate about Google Analytics. These frustrate me as a digital strategist and as a technical specialist, but are sometimes just misleading to webmasters trying to get their head around understanding their website.

Bounce Rate & Time on Site Tracking Sucks

I’ve mentioned this before, but Bounce Rate and Time on Site metrics are my least favourite metrics in Google Analytics. Neither Bounce Rate nor Time on Site metrics are even remotely accurate by default. You’ll get a reported 30s average time on site when it’s actually closer to 15 minutes. To make matters worse, Google uses these metrics in email reports and on the default dashboard implying that they’re somehow useful:

It doesn’t help that many analysts don’t recognize this particular flaw in GA.

Back in 2016 I fixed the bounce rate tracking on my blog by using my heartbeat solution, but in July I changed my whole Analytics implementation for nefarious reasons and forgot to re-enable the heartbeat. Here’s the result:

Once the heartbeat was disabled you can see that Google suddenly thinks that close to 90% of users bounce (leave the website within a couple of seconds) and, on average, users spend 20s on the website. This is dramatically different from the actual behaviour of users on the site which you can see prior to July. Generally users will sit on my hyper-technical 6,000 word blogs posts and try to implement solutions as they were reading through.

Bounce Rate and Time on Site have been broken since the beginning of time and Google has done nothing to fix them. Worse than being misleading, it’s frustrating that Google then gives them to webmasters as gospel.

Read more on bounce rate issues in GA here.

Terrible Spam Controls

Google Analytics is very simple to “hack.” I’m not going to get into the details because I’d rather not encourage the behaviour. Suffice it to say that absolutely anyone can screw up your GA data with the default implementation. And yes, Google has known about it since the platform launched. The result can be spammy, inaccurate reports that ruin the whole analytics experience.

One widely abused security flaw is that Google Analytics doesn’t care which website is sending data to your GA property. If I copied your analytics code and put it on my website, GA would happily pollute your GA profile with my data, no questions asked. This really shouldn’t be the default behaviour.

A quick and easy solution from GA’s perspective would be to prompt the webmaster for feedback any time a new domain started sending data to your analytics profile. “Hey, do you recognize spammy-website.net?” If the webmaster says no, then don’t allow that data to pollute their GA profile.

That in itself would dramatically decrease the degree of spam on most websites. Of course every specialist worth their salt will create a custom filter to block invalid hostname spam, but I have no idea why Google hasn’t resolved this before.

Forced Last Touch Attribution

Let’s say someone comes to my website for the first time through an ad on Facebook. She loves the service that I offer, but she wants some time to think about getting in touch. A couple of days later she searches for Delta Growth on Google and submits a contact form.

She will be a lead associated to the Organic Search channel in Google Analytics. Which is kind of odd, because she would never have heard of us without our Facebook Ads. Google Analytics associates conversions with the last channel used by a user before they converted. This is called Last Touch Attribution.

In our situation, Last Touch Attribution is extremely misleading. Our user would never have heard of us without our Facebook campaign, so why aren’t we giving Facebook the credit? In this case we’d have much preferred to track her acquisition channel as Facebook, because that’s where I should increase my spend if I want more leads like this user.

The odd thing is, Google Analytics does have an understanding of First Touch Attribution (tracking conversions against the first channel they used to access the website):

… but it’s only available in one report under Conversions -> Attribution -> Model Comparison Tool. You’re stuck with Last Touch Attribution in any other report, generate or pull out of an API. This makes it effectively useless, despite the fact that GA clearly has this data internally.

It bothers me that the Google Analytics product doesn’t provide access to these other attribution models in useful reports. It forces analysts to build complex attribution models to make good business decisions. Giving us access to the data that’s clearly already collected would be fantastic.

NB: There absolutely is a way to do this yourself using custom dimensions and fancy third party libraries like Source Buster (highly recommended for the tech savvy marketer), which is one of the ways we help our clients build complex attribution models. You can also get this data through many marketing automation platforms like Hubspot, Pardot or Marketo.

Blind Alerts

Google Analytics has what appears at first to be a fantastic feature: Alerts. Don’t want to comb through reports every day, week, or even month to find out when something awful or brilliant happens? Just identify metrics that you absolutely must watch and create alerts around them. You’ll get an email any time something goes wrong.

At least, that’s the theory.

The problem is that GA alerts have no understanding of what the expected value should be. Here’s a quick, human example:

  1. In December last year, we had 10,000 Organic Search visitors.
  2. For the past three months we’ve had on average 20% more Organic Search visitors than the same time period the previous year.
  3. What should our Organic Search traffic be this December?

If you guessed 12,000 (which is 20% more than what we had last December, just blindly following the trend), you’d probably be in the right ballpark. It’s not reasonable to compare many metrics as straight Month over Month or Week over Week due to seasonality. For example, do you expect your traffic and revenue to be the same over the Christmas Holidays vs. the first week of summer?

So really what we want is to be alerted if we see an increase or decrease relative to our forecast (+20% growth compared to the previous year). I’d be worried if, instead of growing 20% versus the last December, we grew 5%. I’d be excited and want to know if we grew 40% year over year versus the expected 20%. Unfortunately that’s not an option in the default alerts:

This makes alerts useless to me outside of very specific scenarios. And it’s a relatively simple option to add, one I’ve duplicated into a spreadsheet that Delta Growth uses to monitor hyper-specific issues on a daily basis for our clients:

But what gets me is that Google Analytics 360 doesn’t even offer this feature–forecasted alerts–which Adobe Analytics has had for years.

Google Analytics Goal Funnels Suck

This is a funnel fallout report in Hotjar:

This is what a fallout report should look like. A funnel is broken down into Steps, where each step is defined as a page, a group of pages or an event. For example:

  1. Any Service Page (group of pages)
  2. Click on Apply Now (event)
  3. Complete First Page of Application (page)
  4. Complete Application (page)

A user can “Drop Off” by not proceeding to the next step in the funnel.

From the screenshot above you can imagine how this funnel might be useful. You identify the most common funnels used to convert into a sale or lead on your website and build out your funnels. Once you’ve collected enough data it would be easy to see which step has the largest “dropoff” or “fallout.” Knowing what step causes your users to leave lets you fix the user experience on that step. You’d end up with a lower dropoff and a better conversion rate.

Google Analytics has something similar in its Funnel Visualization Reports:

This actually works fine for some websites, but there are some big issues:

  1. You can’t use events as part of your funnel. Most of the clients I work with need to use of events to track conversion steps. It’s very common, so why can’t we use events in goal funnels in GA?
  2. You can’t define more than one funnel for the same goal. An eCommerce site may want to analyze the funnel flow for users who come to the site through a Product Page (Nike Air Max LD-Zero), versus users who come to the site through a Category Page (Men’s Running Shoes), versus users starting from the home page. Technically you can create multiple goals for this… but there’s no good reason why funnels in GA are defined in the goal.
  3. Once you have defined the funnel you’d expect to see the funnel conversion rate in most reports. The actual reported conversion rate, even with the funnel fully defined, is very odd. It’s calculated as, “Total number of conversions divided by all sessions on the website.” That means that if you get more blog readers then your eCommerce conversion rate will drop… That’s not very useful.

So what can we do about these issues?

Luckily all of these issues can be resolved through technical implementations. Unfortunately, the majority require an analytics specialist and a developer. The worst is that many can’t be resolved in the Google Analytics platform at all and you’ll be left finding your own solution.

  1. Bounce Rate & Time on Site: I’ve built a fix for this that I’ve been calling Heartbeat. Check it out! It’s free, just a bit of code you can add to your site.
  2. Spam: Add a hostname filter to your GA profile.
  3. Attribution: This one is tough. If you aren’t going to use a marketing automation tool or CRM that allows you to build different attribution models, you’re going to have to put together a completely custom solution. Delta Growth built ours using Source Busted, cookies and custom dimensions within GA itself.
  4. Alerts: I’m not aware of any off the shelf solution to do this. Delta Growth built our own solution that we use just for our own clients.
  5. Fallout Reports / Goal Funnels: Hotjar has this functionality but it’s probably overkill for most webmasters. We’re currently building our own solution for this as well, though we’ve been using manual report creation for Conversion Rate Optimization work. Google Data Studio, Power BI or Tableau could likely let you build this type of report though!

Good luck and let me know what you hate about Google Analytics!