Dig in to get the truth from Google

In its quest for simplicity, Google Analytics is hiding the truth from you. This article will show you what you need to do to force it to give you the insights you need to be effective.

 

Wise marketers know that they need to optimize their website and funnels by tracking visitor and customer behavior.

So you need a tool to help you track metrics accurately. And for many, Google Analytics is the go-to tool.

It provides a wealth of information you can use to optimize and grow your business–all for free.

Only there’s a problem with this…

This data isn’t accurate. At least not on the surface.

Google knows that people want quick and easy answers, so it simplifies the data down. Sometimes even to the point of inaccuracy.

If you’re just logging into Google Analytics, taking the information at face value, and making business decisions with it, it could be costing you serious money.

Thankfully, there are steps you can take to force Google Analytics to give you more useful data.

I’m going to show you how to spot and unmask these tricks so you can start making better, more informed business decisions.

But first, you need to understand just how Google Analytics is misleading you.

Google Analytics isn’t being straight with you.

Google Analytics promises you detailed, accurate insights into how people are interacting with your website. And on the surface, it seems to deliver on that promise.

When you log into your dashboard, Google presents you with mountains and mountains of data. They track hundreds of metrics for every website visit. Each of those metrics can be filtered and analyzed by hundreds of dimensions.

list of metrics

Everything from bounce rates to geographic location to the exact order they click through pages on your site.

But with this bounty of data laid out before you, it can be easy to lose track of the signal buried in all the noise.

It’s easy to be fooled into thinking it’s really as easy as reading the tables Google pre-populates for you.

google analytics referral chart

But I see at least three misleading numbers in that screenshot alone… and by the end of this article, you will too.

Now, this isn’t just about bad data or even misguided conclusions. It’s about the business decisions that rely on them.

What if you run an A/B test to optimize your landing page and then drive $10,000 in AdWords traffic to it?

How much would it cost you if you chose the wrong winner?

What would be the cost of spending several months creating the wrong content while your competitor took over the organic rankings for the most valuable industry keywords?

But I’m not saying you should just toss out all regard for Google Analytics.

Despite its laziness, Google Analytics is an incredible tool. You certainly shouldn’t throw it away just because it isn’t straightforward.

Most of what we need to fix its flaws are seated right inside Google Analytics itself. You only need to adjust the proper settings (and your mindset) to get accurate, effective metrics.

Let’s uncover these tricks one-by-one.

Trick #1: Mysterious “Dark Traffic”

The other day I was talking with a friend about their site traffic.

They couldn’t believe how many people were accessing the site directly…

Google analytics direct traffic

But something smelled fishy.

 

Direct traffic is supposed to be recorded when the user types the URL directly into the navigation bar or clicks a saved bookmark in their browser.

No referrals. No backlinks. Just self-motivated visitors.

It sounds great! More direct traffic means that people are sitting down and deciding to head to your site without any outside push. A marketer’s dream.

The problem is this metric is flat out wrong.

Just look at my friend’s stats…

odd stat on Google analytics

Nearly half of his new users came from a direct source.

You mean to tell me that 45% of customers found your site for the first time by typing your name directly in the navigation bar?

Unless you’re running some very successful offline advertising or your brand is an established household name, you shouldn’t see metrics like that.

And it isn’t just my friend’s account. Nearly every Analytics account I see has (direct) / (none) as one of its top-two traffic sources. — despite the fact that direct traffic should only represent about 10-20% of your total traffic on average.

 

The problem is dark traffic.

Dark traffic is traffic that Google can’t seem to assign a source. So they just shrug and assume it was direct.

They do this despite overwhelming evidence that it isn’t.

In fact, in 2014 Groupon put it to the ultimate test by de-indexing themselves from Google.

They wanted to see what would happen to their direct traffic numbers. Guess what they found?

results of groupon deindexing test

(Image Source)

 

Direct traffic fell 60% during the window and returned sharply with organic results.

The truth is:

Google is much worse at tracking visitors than it admits. And it uses “direct traffic” traffic to cover it up.

Some other common places to find “direct traffic” are:

  • Links inside email applications
  • Messages on private social apps like Whatsapp, Facebook Messenger, and direct messages
  • Offline files like PDFs, powerpoints, and other documents
  • Image search results

“The most common mistake when people launch their first email campaigns is to forget to add UTM parameters for their links. They don’t realize it until they check the amount of traffic they are getting from their email campaigns.” According to Ajay Goel, Founder of GMass,

If left unresolved this dark traffic will dramatically skew your data.

For example:

If a large portion of traffic from Facebook is happening through links shared via messenger, you might mistakenly assume that social media isn’t performing well.

Fortunately, there are some things you can do.

First, make sure that you’re using UTM codes whenever possible. This allows you to set the source, medium, and other dimensions manually.

Tools like Growth Ramp can help you automatically create UTM codes to promote on social media and other platforms where you distribute content.

example of the UTM generator

(Image Source)

If people take this link and share it on different platforms, you’ll at least be able to attribute the visit back to its original source.

This is great, but it might not be enough.

As Groupon showed us, sometimes even organic traffic isn’t flagged. And we can’t guarantee that people will include the UTM code when sharing privately.

A second tool we can use to clean the data is our own brain (with a little outside help).

First, we want to filter out as much legitimate direct traffic as possible… it’s not all dark.

To do this, you would create a segment:

create a segment in Google Analytics
Google analytics new segment button
And include only (direct) traffic:

filter direct traffic google analytics

But filter out those pages that customers are likely to access directly such as your homepage or category pages:

filter customer direct traffic

Once you have saved this segment, you can examine the data distinct from the rest.

You can compare the trends in it with your marketing efforts and see if there is any correlation.

(Such as a big spike in dark traffic the day you posted a new article on Facebook).

spike in traffic shown in google analytics chart

Trick #2: Data is improperly fragmented

Dark traffic isn’t the only thing causing Google to misrepresent your data.

Some quirks in Google’s filtering process can also cause traffic to be misallocated.

Data fragmentation happens when there are minor differences in the URL which causes the same page to be counted differently.

There are three common situations where this happens:

  1. Case-sensitivity – even though the internet usually isn’t case-sensitive, Google Analytics is. That means /thankyou and /ThankYou would be reported separately.
  2.  Unnecessary query parameters – Some query parameters are helpful for accurately reporting traffic. For example, most UTM codes provide details about the audience (i.e., men), brand (i.e., Nike), product specifications (i.e., red), or other information that a page relates to. But some parameters provide useless information that will skew your analytics. You can identify this data when you see visits to a URL on your site that ends with a series of jumbled numbers or letters. Too many of these unnecessary query parameters will skew your analytics.
  3. Trailing slash – just like capitalization, even though most browsers couldn’t care less, /blog/ and /blog (no slash) may be treated differently by your Analytics.

This may seem trivial (“Big deal, the data’s still there”), but it’s not. Let me explain…

Good marketers are efficient. That means you should focus most of your energy on the top performing pages of your site:

session dropoff in google analytics

Do you see the problem yet?

Let’s say you wrote two popular blog posts “How to Fly” and “How to Swim.”

Your flying tutorial gets almost 5,000 hits.

…But your swimming tutorial only gets 3,500

Which are you going to spend more effort promoting and repurposing?

The swimming tutorial.

 

But what you didn’t realize was that in addition to the 3,500 hits on /how-to-swim, there were 1,500 visitors for /How-To-Swim (with caps) because someone was sharing a different link.

Another 1,200 visitors showed up at /how-to-swim/?=rel for some reason too.

If your top-10 cutoff is 4,000 hits, none of these will appear on the first page at all.

 

So while your dashboard will say /how-to-swim received 3,500 hits, it actually outperformed your flying tutorial with 6,200 hits!

This is just one simple example. Now think of all the things that could go wrong when traffic isn’t properly credited. Your data could be far more skewed than you realize.

 

Faulty data can devastate your marketing budget. Marketers should determine the site’s best content and then double down on their financial investment to promote that content.

But if you rely on faulty data, you’ll double down on the wrong content and potentially squander your budget.

The solution here is to build filters that normalize your data in Google analytics before it’s categorized.

Here’s how:

Go to the Admin Section…

google analytics admin section

Then, under VIEW, select Filters…

Google analytics view filters

Select +ADD FILTER…

Google analytics add filter

Then set the filter up like so…

Google analytics filter hostname

This will normalize your hostname—the main part of your URL. So MySite.com will turn into mysite.com.

You’ll want to set up a few of these too address the different problems I mentioned above. Here’s a comprehensive list of filters and the settings that go with them.

 

Trick #3: Google can’t keep time

So far these tricks have represented misleading numbers that were just hiding. But sometimes Google Analytics is just plain wrong and refuses to admit it.

What if I told you that one of the largest tech companies on the planet couldn’t keep track of time. Even with all of its supercomputers and servers.

That’s right.

Google has been passing you inaccurate time data since the very beginning.

Now, of course, I don’t mean that they have trouble counting the seconds pass by or that they’re stuck in some dimension where time passes at a different rate.

No. The error is in how they calculate it.

comparing time on page

(Image Source)

 

Google calculates “time on page” as the difference between the time they landed on a page of your site to the time they entered a different one.

So in this example, time on Page 1 would be five minutes. Likewise for Page 2.

But about Page 3? What if they exit before visiting elsewhere?

That time is reported as zero seconds.

It was hard for me to believe at first too.

What this means is that all your metrics like “Avg. Session Duration” can be wildly understated. Especially if the bounce rate is high:

avg session and avg time on page compared

This problem is most exaggerated on pages with high rates of exit or bounce and can really distort your comparative analysis between pages.

The good news is, this information isn’t mission critical to your optimization. It’s more of a correlation metric than a causal one…

correlation comic strip

(Image Source)

The bad news is that there is no easy fix for this one. Google doesn’t allow you to change how it records duration.

But that doesn’t mean they’re useless. Strong inferences can still be made by comparing trends in the duration. The key is to take them with a grain of salt.

Especially when comparing pages with very different bounce rates.

Trick #4: Last-touch attribution doesn’t show the full picture

So you’ve built an epic landing page using trust icons, meaningful customer reviews, and relevant testimonials to increase conversions and it’s converting great.

Congratulations! Let the sales pour in!

via GIPHY

Because you’re such a rockin’ marketer, you’ve already been running an effective Facebook Ad campaign for a few weeks now.

And now like it’s time for your two masterpieces to come together and dominate.

But after a few days of waiting, your checkout page looks more like this…

What gives?

The most common reason for this is that you didn’t account for all the work that went into to earning your reader’s trust and interest before the sales page.

You just went in Google Analytics, saw people hitting your sales page, and then converting because of your awesome copywriting skills.

But the truth is customers usually need to interact with your brand more than once before they are ready to buy.

avg customer online interactions in a chart by industry

(Image Source)

Even further, customers that have interacted with your brand 7+ times online are likely to spend almost double.

So it’s not just your conversion rates at stake, but the conversion values as well.

This is why Last-Touch Attribution is such a problem in Google Analytics. It oversimplifies the buying process and can lead to a poor understanding of the actual customer journey.

By default Google Analytics credits the whole customer buying experience to the last interaction they have with your brand:

last touch attribution(Source)

 

But in reality, it looks much more like this…

 

reality multi touch attribution

(Source)

If you don’t track all the steps leading up to a conversion, then you’ll be misrepresenting their value.

As a result, you might undervalue the power of social media because it isn’t driving any direct sales.

But what you’re missing is the number of customers that discovered your brand on Pinterest or Facebook before searching for your blog online.

Once on your blog, they downloaded your eBook and joined a newsletter.

They liked the content so much that they jumped on the phone with one of your salespeople or stopped by to chat with them at a tradeshow.

Your sales representative talked up your monthly webinar and convinced them to attend where you closed the deal with a persuasive pitch and time-sensitive bonuses.

Every step in the chain was critical to winning the customer.

While Google Analytics isn’t great for tracking offline events like sales conversations, it can be great for tracking contribution marketing.

 

You just need to look at your multi-step attribution model.

Navigate to the Conversion Reports section…

Then select the Model Comparison Tool…

conversion channels

 

By default, you’ll see the Last Interaction report we just talked about.

But they have a few other models to look at:

default conversion models in google analytics

There are use cases for each of these. But for now I’m only going to discuss the most common:

Linear, Time Decay, and Position Based Models

The linear model gives equal weight to each step in the process.

So if a lead interacts with any of the Channel Groupings or sources along their journey (even across multiple sessions), it will get equal credit for contributing to the ultimate conversion.

For example, a customer first found the site on through Facebook, then visited the site directly and was finally picked up by a retargeted search ad.

That’s Social Network, Direct, and Paid Search. Each would receive 33% credit.

model comparison tool

If you want to see more information about how these channels overlap and interact, you can under the MCF Overview tab:

Venn diagram of conversions

The Position Based model works very similarly. But instead of giving equal weight to every step in the funnel, it heavily weights the first and last interaction—giving them each 40%.

The remaining 20% is split evenly between the mid-points.

position based model

(Image Source)

 

The logic behind this is that the most important steps in the conversion process are to start the relationship and close the deal.

The important thing is to ask yourself if that’s true for your brand or product. For some, a Time Decay model might be more appropriate.

time decay attribution to different channels

(Image Source)

 

This model will give the most credit to the most recent touchpoint, and less credit to all those before it.

split credit across channels

(Image Source)

 

I can’t tell you which model is best for your business. But it will probably be one of these three multi-step models.

If you want a more sophisticated way to track the conversions your content is leading to you can use more sophisticated software like Salemate.io.

conversion tracking pie chart

Salesmate.io shows you which articles in your content funnel are generating leads. And by tracking the success of those leads, how valuable the content is to your marketing efforts.

Conclusion

Every marketer knows how important data is in making effective decisions. And of course, you want that data as efficiently as possible.

Google Analytics promises to deliver this.

But the numbers you see on the surface are very misleading.

Google Analytics is tricking you into thinking that people and their psychology are easy topics. It’s a lie we want to believe because it means our jobs are much easier.

Believing this lie is as costly as it is lazy.

Thankfully, you can catch these tricks and put a stop them.

At BuildFire, these methods are leveraged to properly track everything religiously, so you can take action on as accurate data as possible.

So make sure to take the time to understand what the numbers really mean. Apply filters to normalize and segment the data properly. Understand the value in each step of the customer journey.

If you do it right, Google Analytics offers you a wealth of free data that can be used to make smart decisions and crush the competition.

But it’s only possible if you think past the initial, misleading mask Google lays before you and dig deeper.

 

***

This is a guest post from the SpyFu Community.