Everyone that works in digital marketing, whether that be SEO, PPC, CRO, or email marketing, loves data.
We’re obsessed with it. We love scooping up data, breaking it down into segments and more manageable chunks, looking for anything we can tweak and exploit.
For a lot of us, Google Analytics is an invaluable tool. It gives us the data we need to make smart decisions.
Good digital marketers should always be using data to back up their hypotheses. However, there are times when we can blindly trust the data.
After all, numbers don’t lie, do they?
Well, yeah. They kind of do.
Especially when you consider how most marketers collect their data. We rely on tracking software and scripts to give us the juicy data that we crave.
Because of this, there’s always a real risk that the scripts and software fail, spitting out crazy numbers in the process.
With crazy numbers, we’re doomed to make crazy decisions, so it’s important to know when to call bullshit on the data you’re being fed.
Google Analytics can and will go wrong
Google Analytics isn’t a foolproof system.
However, in terms of price (free) and usability (pretty simple), it can’t easily be beaten.
In my opinion, the quickest, easiest and safest way to deploy the Google Analytics tracking code is via Google Tag Manager.
By deploying the code through Tag Manager, it can help reduce the chance of tag duplication, which is a classic way to absolutely destroy your data.
It also helps avoid deploying the code in the wrong section of the web page, which can also give some really funky numbers.
Now, I’ll admit, I would always install the Google Analytics tag by manually adding it to the <head> of a website, or by sending it over to a developer to implement.
I stopped doing this when one of my friends who works on big public sector contracts (with their notoriously unreliable web/IT departments) recommended deploying via Tag Manager, simply because it’s one less job you have to bug a developer take care of.
Once Tag Manager is installed, you can add further code snippets that are injected into the data layer without even having a login to the website.
That means if you have additional scripts you want to add to the site, you can do it without hassling a web developer who’s probably already up to their eyeballs with more pressing tasks. It’s one simple task and that’s it.
In fact, there are lots of things you can do with Tag Manager that aren’t immediately obvious, all of which are super flexible and can be implemented with minimal knowledge of code.
The time I pressed the magic ranking button and all the traffic came back
I was contacted by a site owner who was convinced their organic traffic had disappeared overnight.
They were convinced of this because that’s exactly what their analytics data told them. Check it out.
Before I got to see the data, I spoke to the client on the phone. They described the issue and what they were seeing to me.
My first reaction was that something like crawl directives could have accidentally been messed up. That’s one thing that can cause a website to immediately lose traffic.
My second thought was maybe it could be a penalty of some sort. Again, this is one situation that can cause a website to lose a lot of traffic in a very short timeframe.
Traffic doesn’t just vanish for no reason. Something was up.
I did some quick checks for obvious signs of manipulation or technical SEO errors.
Their backlink profile was squeaky clean - completely unoptimised. SEMRush suggested visibility gains throughout the period traffic ‘disappeared’. I was satisfied that the site hadn’t been misbehaving in any way, and there were no obvious signs as to where this traffic had gone.
So then I looked at Google Analytics and I discovered the culprit.
The tracking code was being deployed in the footer. In certain situations, it’s useful to deploy JS in the footer as it can speed up websites. However, this isn’t a great place to deploy the Google Analytics tracking code. You want the code firing as early as possible.
It needs to be high up on the page otherwise the firing of the tag could be delayed whilst other content and scripts on the page load.
That was error number one. Error number two was the speed of the website. It was seriously slow - like wading through treacle.
This created the perfect scenario for their data collection to totally fail.
The code was firing so low down on the page that a lot of people would be clicking through pages before the page had finished fully loading, meaning a lot of traffic potentially wasn’t being recorded.
They did some work on the speed of the site which resulted in dramatic improvement in page load times. As soon as that happened, traffic started recording properly.
To guarantee that all future traffic was recorded properly, I then set the client up with Google Tag Manager and deployed Google Analytics through that. This made sure that the Google Analytics code was deployed and firing from the right section of the web page.
The traffic now tracks correctly, with no more questionable numbers or readings.
Being a data-driven marketer is great. It offers feedback that lets you know you’re on the right track. However, sometimes, the numbers simply don’t add up.
When the numbers look wrong, they usually are. Data integrity should always be something that’s in the back of your mind.
I’ve seen people get overly excited over unrealistic single digit bounce rates. On extreme occasions, I’ve seen people celebrating the mythical 0% bounce rate (hint: not possible, even if you have the best website ever created).
I’ve also seen people, like the above, worry about situations that look disastrous at first glance. However, upon close inspection, everything was OK - it's just that data wasn't tracking correctly.
You should not be a slave to your analytics data. It isn’t foolproof, so always be prepared to call its accuracy into question if things look weird.
If traffic suddenly increases or decreases by a large amount and there are no obvious signals why this happened, you may want to check that your Google Analytics set up is tracking correctly. The same applies if your user metrics (i.e. bounce rate, time on page) suddenly change, or look to good to be true.