Why conversions in Google Analytics differs from Facebook Ads

One of the most common struggles in online marketing today.

One of the biggest misconceptions in marketing today is the usage of Google Analytics. My goal is therefore to give you as a reader a more truthful way of analyzing marketing performance – more specifically when advertising through Facebook Ads.

The issue

No matter how you slice and dice the data, you’ll never get the same number of conversions in Google Analytics and Facebook Ads. Normally, marketers tend to rely on Google Analytics in these cases – but why?

Reason #1: People-based vs. Cookie-based

The main advantage Facebook has is their people-based measurement, in comparison to Google’s cookie based measurement. Since the most of us have an active Facebook session on our devices, Facebook has no trouble attributing conversions cross-device. Additionally, since Facebook owns Instagram, Messenger and Whatsapp – they can also utilize user data and sessions from those apps to attribute correctly.

Google, on the other hand, are having trouble attributing cross-platform – since they rely on cookies. Also, if you’re relying heavily on referrer URLs (utm links), you’ll also miss out on the scenario when someone is clicking a Facebook ad, shutting down the website and then opening it again in order to complete the purchase. In this case, Google Analytics will not attribute the conversion to the specified utm parameters.

Some examples on what issues this causes

  • Eric clicks an Instagram Ad on his iPhone, then converts 2 days later on his computer. Facebook sees this journey and has no trouble attributing this conversion to the click on another device 2 days earlier. Google Analytics only sees a non-converting click from Instagram and then a converting visitor from “direct traffic”.
  • Emma sees a video ad on Facebook 9 times without clicking. Two weeks after she sees it the last time, she remember the brand’s name when a purchase intent is activated. Although, she’s too lazy to write the whole domain “acme.com” – so she simply writes “acme” in her browser. This results in a Google search, which then results in a click on an Adwords ad (for the brand’s name as keyword) – or an visit from organic search. If she converts, Facebook sees this conversion in their view-through attribution, while Google Analytics will present this conversion from a brand search.

In fact, both Facebook and Google have published studies showing that a great deal of conversions starts on one device and finishes on another – making a proper cross device tracking crucial.

Reason #2: Attribution definitions

Another significant difference is that Facebook attributes the conversion to the day of the click and/or impression (based on attribution window used) – while Google Analytics attributes it to the day of the conversion (since they mostly don’t know when the impression/click was made).  This of course increases the gap in difference between the two tools, which can explain day-by-day differences.

Reason #3: Different Protocols

Another issue arises when 3rd party tracking tools use referrer URLs to attribute conversions. Since 40% of people browsing Facebook uses HTTPS (in opposed to HTTP), an ad click can’t be attributed through a referring URL when they leave HTTPS for an HTTP environment.

Reason #4: View through

Looking at view-through performance is often scoffed at. Saying ad views doesn’t impact conversions is like saying that no offline marketing has ever worked (TV, print etc). You can’t click on a TV commercial, right? And since view-through matters and actually makes an impact, the fact that Google Analytics doesn’t track view-through performance makes it even less relevant.

Ad blockers also messes up the data, since they corrupts tracking javascripts and cookies. This will only impact conversions negatively, resulting in under-reporting.

Reason #5: Attribution models

These examples depends highly on the attribution model used in Google Analytics. Mostly, last click-attribution is used – which also is used in the examples above. The other attribution models aren’t doing anything more nuanced, since Google still doesn’t know people – they only know cookies. Therefore, trying to get a better understanding of your data with various attribution models won’t do you good – since the raw data is corrupt.

Learn the lesson: Vanity metrics

In this article, I’m specifically discussing the issue of discrepancies in number of conversions between Google Analytics and Facebook Analytics. Although, I also want to highlight why vanity metrics are really dangerous for your marketing.

Since Google Analytics is, by far, the most used data analysis tool today – it’s setting the standards of metrics and reporting. From the start, Google Analytics has been “educating” the world of marketers what metrics to look at – making us think that the metrics are true and actionable. But they’re not.

As we’ve learned in this article, there are a great amount of reasons to why Facebook Ads data differs from Google Analytics. This isn’t unique for Facebook Ads – it

Let’s take an example: Conversion rates. 

It’s one of the worst vanity metrics, but it’s widely used as some kind of qualitative measurement.

vanitymetricsTo make an easy example; Let’s say your goal is clicks. In this case, CPC would be the metric you’re analyzing. Vanity metrics in this case are CTR and CPM – since they’re not giving you more information than CPC is. Have a look at the screenshot to the right. This is an actual case from Facebook. In the case where CPC is the highest (48,70) – the CPM is the lowest. And the case where CPM is 11x higher – the CPC is the lowest (13,51).

If we were to give CPM any kind of meaning here, it would affect our decision in what Ad Set to run. But if we’re only looking at what really matters – we skip all the metrics that are before the truth.

So, if you’re an E-commerce and your goal is to get high ROI and a good volume – don’t look at CPA, CPC, CPM, CTR or conversion rates. Just look at ROI and the volume you’re getting. Everything else is completely irrelevant.

How to work with Facebook Ads

Everyone has their own ways of looking at data – and this is where the flexibility of the attribution windows comes in handy. Facebook Ads offers you to see the data in everything between 1 day and 28 days post-click – and the same thing with view-through.

In other words, if you’re a Facebook Ads sceptic – you could look at 1 day post-click, which simply gives the users 24 hours to convert after the click (otherwise, the conversion won’t be attributed at all).

Common sense is key, so try comparing the different attribution windows to see what insights you get and what setting you think suits your business the most.

Same goes with other marketing channels with their own pixel; you should look at the data delivered and interpret it the way you think makes the most sense for you. Don’t use Google Analytics as an intermediary.

Fully utilize the opportunity with Facebook knowing people (and not cookies). Do this by adapting to all tracking techniques that suits your business. In Facebook, you can use both pixel-based tracking, app SDK and offline conversions.

To sum it up

Google Analytics isn’t designed for professional decision making, based on data.

It sounds like a harsh statement, but I’m not the only one who realized the illusion they’re creating for marketers worldwide.

All the reasons – a short version

  • Google Analytics doesn’t know people, they mainly know cookies – and are therefore far from the truth.
  • UTM-links are often corrupt after a click is made.
  • Google Analytics is in some reporting sample based, which makes data unreliable.
  • Google Analytics is a bias tool that makes its users focus on vanity metrics.
  • Google Analytics provides various attribution models, creating the impression that there’s a sane way to look at the data.
  • Since 40% browse Facebook using HTTPS – the referring URL breaks and the clicks appears as “direct traffic”.

Notes & disclaimers

 

Disclosure: I work at Zalster. Although, I’ve had issues with Google Analytics the last 10 years – so this article isn’t biased :)