On this Monday, “Facebook disappeared” and some websites faced an increase in traffic (those that do not depend so much on traffic from social networks), while others noted a decrease in traffic. You yourself may have looked at your google analytics for the evening of Monday, October 4th 2021 and seen that you have a lot of traffic from Facebook and Instagram.
If you think it’s probably because “Google is lying to you, because we couldn’t have visits from Facebook if it wasn’t working yesterday!!!” then it’s time for lessons in attribution. If you know that no one lies, you are an analytical savant and now is a good time to use this day for lessons in attribution, both for you and for your clients.
As many goals as possible, with as little money as possible…
We should primarily use Google Analytics as a marketing tool, not a sales tool. That means that, as marketers, we should primarily monitor the sources, of which there may be hundreds, people come to our websites. Not just how many users we had on our website, from which countries or which are our most visited pages.
Universal Analytics, as well as the new Google Analytics 4, show us which sources visitors came to our websites from. Did they come to our website via Facebook, Google ads, Instagram, our newsletter, email signatures, a banner that we placed on the portals, a QR code from the product packaging, some other way, or did they directly type our web address into their browser… At the same time, Universal Analytics presents its reports according to the so-called Last Non-Direct attribution model, which Google Analytics 4 calls Cross-Channel Last Click, and they mean the same thing.
When we define goals (conversions) on our website and set them up in Analytics, the marketing reports will tell us not only where users came to our website from, but also what quality those arrivals were. It will also tell us which marketing channels are worth investing in, because it may not be worth investing in Facebook if it brings us the most visits but not purchases or other conversions.
The aforementioned analyzes are the basis of performance marketing, in which we try to allocate marketing budgets to those channels that achieve goals, in order to achieve as many goals as possible with as little money as possible. If an agency tells you they are “the performance agency” and that they “count impressions, reach and clicks”, be aware that they may have only scratched the surface of performance and that they still have a lot of work to do before diving into optimizing conversions.
But what does that “last not direct click” actually mean?
If someone came to your website last Wednesday via banner (Display), on Thursday saw (and clicked on) a Facebook ad (Social) about your website, and on Friday finally googled a term related to your website, visited your website (Organic) and purchased a product, most reports in Google Analytics will show that Google (Organic) is responsible for the purchase.
Example: Display -> Social -> Organic = 100% Organic
If a visitor came to your website again on Sunday via Facebook (Social), and on Monday by entering your website address or by clicking on a newsletter link with no UTM-tags, that last visit is actually categorized as “Direct arrival”. However, as Google Analytics knows using cookies that you arrived via Facebook the day before, this last (Direct) visit will also have the attribution “Social”.
Example: Display -> Social -> Organic -> Social -> Direct = 100% Social
Due to the “Last Non-Direct” attribution model, some visits to your website recorded on the evening of October 4th will still be recorded as “Social” even though Facebook, Instagram, WhatsApp and Messenger were inactive. Due to the mentioned model, it also happens that in analytics we see visits from campaigns that have been stopped. This can also happen when one of the visitors bookmarks the address of your website, and that bookmark also contains information about the campaign (and that is the whole new blogpost).
You can check the accuracy of this statement and model by selecting “Direct Session” as the second dimension in reports (Secondary Dimension) in Google Analytics (Universal). You will see that the visits from Monday, October 4th may belong to the “Social” channel, but under the “Direct Session” dimension it still says “Yes”. That “Yes” means that the users arrived directly, but that it was not necessarily their first visit. If, for example, the column “Default Channel Grouping” contains the information “Social”, it means that their previous arrival was via social networks.
Familiarize yourself with various tools attribution models so that you can compare the results
Reports that use “Last Non-Direct Click” (the attribution model offered by Universal Analytics) or “Cross-Channel Last Click” (the model in GA4, which also ignores the last direct visits) will therefore always favor the last time visitors arrived on our website, and ignore other arrivals and their channels.
Because of this, marketers in performance marketing often wrongly focus on those channels only that “bring them conversions”, and ignore other ways of arrival. Other analytical tools, such as Facebook or Google Ads, also use their pixels to monitor the performance of their traffic, but they mostly attribute conversions to the last visits from their channel. For example, we can have a complex situation like this with 5 different visits to a webshop:
Display -> Social -> Google Search -> Corporate web (Referral) -> Direct = Purchase
What can be seen in the reports?
- Google Analytics will say that the corporate website (Referral) is responsible for the purchase.
- Google Ads will say that Google Search is responsible for the purchase.
- Facebook will say that the second visit was responsible for the purchase.
Therefore, we need to know the attribution models of individual tools well, so that we can compare their results and not to think that “there is always someone lying about something”.
More details on how Google Analytics can help you analyze the so-called “helping” or “assisting” channels that lead to conversions, you can find in the following article: “Attribution in Google Analytics: why we need it and how to understand it“. For those of you who are impatient, I spoke about that in a guest appearance with Petar Bogdan in his “Day for Podcast”, unfortunately in Croatian. So here’s the video lesson on attribution.
