Is Google Analytics Under-Reporting Your Organic Traffic?

by | Dec 20, 2016 | Data & Analytics

Check your reports! Google Analytics is mislabelling up to 4% of organic sessions as ‘referrals’.

If recent news is anything to go by, 2016 will probably go down in the history books as the year of misreporting. From fake news to fake referral spam, those of us who work in digital and performance have had a lot to keep us on our toes.

And just in case that wasn’t enough, we’ve come across yet another example of misreporting – this time from the big G itself.

A few weeks ago, we noticed an interesting thing while conducting some analysis on a few clients’ Google Analytics accounts.

If you’ve seen the following two symptoms in your account(s), chances are you’re encountering the same thing we uncovered. That is, Google Analytics could be under-reporting your organic traffic.

The symptoms:

  • Significant increase in referral traffic YoY
  • This increase is largely confined to mobile sessions

3 Search Engines

In trying to answer the question of ‘why’ this phenomenon was (somewhat too conveniently) prevalent across the 3 accounts we looked at, we landed on the below.

  • These 3 search engines don’t seem to be automatically classified into the ‘organic’ medium by Google Analytics:
    • au.search.yahoo.com
    • duckduckgo.com
    • com.google.android.googlequicksearchbox

Of all of these, the last one (com.google.android.googlequicksearchbox) was the most puzzling and the most prevalent; puzzling because you’d assume Google understands what its own products are, and prevalent because it has the highest volume of all of the apparent search engines in the referral reports.

In one case, it had even garnered almost 81,000 visits over the calendar year to date, amounting to an astounding average of 340 visits per day (from nothing in the prior year!).

Almost 81,000 Visits

Could this just be referral spam?

We thought so too, but after a bit of further investigation, it turned out the source was in fact a Google search conducted via the Google App on an Android device.

Those of you who have downloaded the Google Android app on Google Play might’ve done so via the link below (notice that the ‘id’ is exactly as it appears in the referral reports in GA).

Google Play Link

We’ve gone a step further and have now successfully replicated the scenario via one of our own devices, and we can confirm that we did indeed register as a ‘referral’ from ‘com.google.android.googlequicksearchbox’ in GA, not as an organic visitor as would be expected.

We’ve also managed to test DuckDuckGo and, as with the above, the session is recorded as a ‘referral’ rather than organic traffic. Testing au.search.yahoo.com yielded the same result.

When did this start happening?

DuckDuckGo and Yahoo seem to have been happening for as long as we can tell. The Google App on Android seems to have happened from April 28 in all accounts we assessed:

Starting April 28th

What does this all mean?

In a nutshell, we’re probably under-reporting Organic traffic. How substantial our under-reporting is will be linked to how exposed the client is to the factors at play (mobile penetration, search engine usage patterns), but it’s probably worth checking nonetheless.

How much are we under-reporting by?

Across the 3 accounts we looked at (we’ll be conducting the same analysis across all our accounts in the coming weeks), the % of sessions being falsely attributed to referrals was as high as 4%, but the average is about 1.25%.

As insignificant as that may be from a percentage-based figure, with 340 sessions being misattributed daily across one client, it might be pretty substantial in its aggregate effect.

How can I look at the impact it’s had on me?

Easy – navigate in Analytics to the ‘Referrals’ report (Acquisition > All Traffic > Referrals) and apply the below advanced filter:

Include Source Matching RegExp: “Android|DuckDuck|yahoo” (without quotes)

Is there a way to reclassify this?

Glad you asked. You can set up a filter in Analytics with the below parameters to automatically reclassify these values into an ‘organic’ medium.

Something to be aware of is that the filter will only apply from this point forth and won’t change any data already in the account.

Details for the filter are as below (careful, these are case sensitive unless stated!):

1. Navigate to the Filter menu in the Admin section of your account.

Filter Menu

2. Input the following parameters:

Item

Input

Filter Name Doesn’t really matter, though I’d suggest the below to keep things consistent across accounts:
Reclassify incorrect organic referrals
Filter Type Custom > Advanced
Field A > Extract A Campaign Source:

^com\.google\.android\.googlequicksearchbox$|^au\.search\.yahoo\.com$|^duckduckgo\.com$

Field B > Extract B Campaign Medium: referral
Output To > Constructor Campaign Medium: organic
Other variables Field A Required: Yes

Field B Required: Yes

Override Output Field: Yes

Case Sensitive: –

 

In the end, you should end up with something like this:

End Result

Anyway, Google Analytics are forever updating their out-of-the-box classifications so we expect this to be fixed at some point. Strangely though, we’ve never seen something as blatant as this continue for as long as it has (almost 8 months now and counting!).

What You Missed Last Week In Digital – November 7, 2016

Last week in digital saw some sexy changes to your favourite social media platforms and an easier Google experience for mobile users. Read about it here.

Read More