GOOGLE ANALYTICS SKEWS 20% OF YOUR CONVERSION DATA

GOOGLE ANALYTICS SKEWS 20% OF YOUR CONVERSION DATA

By: Pavel Šíma, Roivenue CEO
[LinkedIn] [Twitter]

nnznHave you ever heard of Campaign timeout setting within Google Analytics? Almost no one has and it is skewing most of your data. Let’s have a look at the problem and how can we fix it.

 

What is Google Analytics Campaign timeout?

Google gives a very vague definition and explanation of what Campaign timeout (and Session timeout) are:

 

The default length of time for a session or campaign. An individual session or campaign for a given user ends after the amount of time specified here has passed (counting from the start of the session or campaign), so long as the session or campaign has not been stopped through another means.

 

Let me spare you the trouble and put this into words that most people will understand.

 

What it means is that whenever a user comes to your site directly, but they had come to your site previously from another non-direct source (like organic, cpc, or referral), if the first session happened within the time window specified by Campaign Timeout, then medium and source of the direct visit (or visits) will be overwritten.

 

You’ve read that right: they will all be overwritten. By default Campaign Timeout is set for a six months window. Yes, six months! All clear? No. OK. Let me give you a specific example.

 

If someone came to your website via a search ad in January and then again directly once each month in February, March, April and May, instead of one paid ad session and four direct sessions, Google Analytics will show five paid ad sessions in all reports except one.

 

Instead of one paid ad session and four direct sessions, Google Analytics will show five paid ad sessions.

mmzmoa

Understand Google Analytics Campaign Timeout
 

For more specific examples of how exactly Google Analytics overrides source and medium of the sessions, see this great blog post by Optimizestart.

 

Is Google being evil? 

The setting by itself is neither good nor bad. It’s just a different way of evaluating a performance of your campaigns. A look that favours non-direct touchpoints over direct ones.

 

The problem most businesses have with it is:

1. Exists in the first place, 
2. It is set as default,
3. It has been skewing results of their campaigns from day one.

 

The lack of transparency from Google is the reason why most marketers and analysts are outraged. It’s like dating someone for months only to find out that they don’t really own a Yacht in Monaco.

 

It’s like dating someone for months only to find out that don’t really own a Yacht in Monaco.

 

Can we blame them for having such feelings? Changing the source and medium in all revenue reports (outside of the specific Attribution reports) without giving an explanation is not being truthful. The only place you can find any information from Google is a joke.

 

For example it says that you should “Set the campaign timeout handling to the same amount of time the campaign is going to run or expected to be relevant. Clicks to a social media micro-campaign might not be relevant for more than a few days after launch.”

 

What should one think about such advice given that it is a property-wide setting that cannot be set-up for different campaigns separately? 

 

Ok, so how bad is it really?

From our experience, once our clients change the setting to normal (meaning to 4 hours which is the shortest time-interval possible), the participation of direct channel on conversions (or revenue) goes up by 20–30 percentage points.

 

Bellow you can see two Acquisition Report results from an e-commerce retailer for the same time period with all Google Analytics settings except Campaign Timeout remaining the same.

 

As you can see, the participation of Google / cpc goes from 20% (with 4 hours Campaign timeout window) to 27,8% (with 6 months timeout window) – a whooping 40% increase in the performance of the channel. Which consequently is going to also influence all metrics that you are evaluating with the help of Google Analytics data, including CPA and ROI.

 
smsmxmm
 
 
 
 
 
 
 
 
 
 
 
 

Same account, same time period. Source/medium share of traffic with Campaign Timeout set to 4 hours (left) and to the default 6 months (right).

 

The participation of direct on conversions (transactions) goes down from 41,6% to 17,2%. All of that credit is redistributed to other channels like facebook / social, google / organic or email with most of them “improving” their performance by 50–100%.

 

Take a closer look at the true numbers. 

nnopi

   Campaign Timeout 4 hours

 

mmnp

 Campaign Timeout 6 months 

 

Why is the Campaign timeout setting even there?

 Google of course will never comment on the topic. But think about it: the vast majority of Alphabet’s income still comes from its advertising business (85,8% as of Q3 2018).

 

At the same time, the majority of advertisers are SMBs who are far from being sophisticated with analytics tools and take the results presented to them at face value.

 

To these users, this Campaign timeout setting artificially inflates the ROI of search and other campaigns by double digit percentage points.

 

This Campaign timeout setting setting artificially inflates the ROI of search and other campaigns by double digit percentage points.

 

Is this a good thing?

Before you start telling that distributing direct visits to other sources is actually a good thing because the insights become more actionable as you have much more sway in influencing what’s going on in those channels, ponder this:

 

There is another very easy way to do this in Google Analytics. Within Attribution, you can easily switch between Last Interaction and Last Click non-direct Models (among others) which essentially does the same thing.

 

It does so at a place where you would expect it and the Google help text is actually very transparent and helpful in explaining its meaning.

 

w`pe

Taken together, if the only motive of Google was to give us a tool to model out where the credit from direct visits should go to, it wouldn’t need this behind-our-back Campaign timeout setting.

 

How does this tie into Multi-touch Attribution? 

The biggest sin of the approach Google has chosen is that it overwrites the source of the visit. Therefore, if you don’t upgrade to Google Analytics 360, which most businesses never will, you will not be able to access the raw data through Google BigQuery. You will never be able to see which visit sources were overwritten and never be able to see your true data.

 

The source of sessions are artificially overwritten and there’s no telling of which and how many were changed. Which in turn affects all your attribution efforts.

 

So, no matter which attribution model you choose to examine your customer journeys with within Google Analytics – being it last click, first click, linear or time decay – they are all going to work with data that has been artificially changed.

 

The biggest problem that follows is that you cannot seriously work with different attribution models until you solve this data problem. Very much the same way you shouldn’t be a judge in a beauty contest until you fix your vision first.

 

How to solve the Campaign timeout setting problem? ✅ 

The most obvious solution would be to immediately go and change that setting from six months to the shortest possible interval – which is four hours. But hang on a minute. If you do that, you will see a sudden drop in performance of all paid campaigns. It will throw all KPIs you have set up with agencies and/or channel managers on your team into disarray. And they will hate you for that.

 

Option n. 1: Run a paralel Google Analytics property ✔️ 

A smarter way to go around this is to set up a new property within Google Analytics with exactly the same settings as your original, with one notable exception: length of the Campaign timeout. Then run both of them in parallel.

 

Campaign Timeout setting can be found in Admin > Tracking Info > Session Settings > Campaign timeout.

 

aoao

 

 

Option n. 2: Get a paid web analytics product ✔️✔️

Option number two is to switch to a paid web analytics product like Google Analytics 360, AT InternetAdobe Analytics or Webtrakk that gives you access to the raw session level data.

 

Option n. 3: Get serious about attribution ✔️✔️✔️

Your third option is then to implement an attribution solution that uses its own measurement to track and classify multiple touchpoints independently on Google Analytics 😉

 

Why, Google, why?!

Don’t ever forget that Google Analytics is a free product. Google is not making money off it directly, so it has to make them somewhere else.

 

Back in the pioneering days of digital marketing, Google Analytics allowed most advertisers an easy way to measure results of their online activities. It played a historical role in giving data to millions of businesses which ultimately brought them ad dollars.

 

For one reason or another though, providing an accurate measurement somehow wasn’t enough for Google. They decided they needed to show the results in a somewhat better light. Like this, Google was “stealing” direct sessions that advertiser would be inclined to attribute to offline and brand campaigns for the online industry.

 

Everybody in digital marketing was a benefactor of this but Google by far the largest one.

 

Google brought the trust and ad dollars of millions of businesses into digital. But at what cost to its credibility?

 

The question then remains: Can you ever trust Google again? You be the judge.

 

The only thing we would recommend is that next time Google is pitching you an awesome accurate free product such as their most recent offer with Google Attribution, remember how they’ve kept everyone in the dark about Campaign timeout for a decade.

 

Trusting your ad spend decisions to black-box algorithms – which most attribution vendors are – is daring enough. Trusting it to a black box of a company that has a history of pulling your leg is a whole different game.

 

 

Further reading on the topic:

Understanding Sessions Campaign Timeout Settings in Google Analytics by Optimizesmart

Direct Traffic In Google Analytics And Last Non-Direct Click Attribution by Bounteous (formerly Luna Metrics)

Google Analytics Does What?! by Bluemoondigital

This reply on Quora by David Jurelius

 

 

Topics: Articles

Subscribe to Newsletter

Recent Posts