Why attribution is the first thing to understand in Google Analytics?
Business summary: Understanding attribution is fundamental knowledge in Google Analytics. And it's not only about traffic sources.
Because it affects almost every report! And this could be the end of the post. You've probably heard about attribution modelling as assigning the rightful volume of conversions to traffic sources. Undoubtedly the most discussed topic regarding attribution in Google Analytics.
Looking at attribution in general, means assigning credit to someone or something. In our case, something and not only to traffic sources. This post IS NOT about differences between attribution models. Try to google about 100 theoretical articles written about it. The point is to show the very basic data used daily, for which understanding the attribution is fundamental if you want to make the right business decisions.
The very favourite report especially when doing UX analysis. How many of you have at least once looked on the combination of "Landing page" and "Conversion rate"? And how many of you are driven by it? Spoiler alert: you might have a wrong data! And quite misleading. The reason is attribution. Try to imagine quite common buying process. You'd like to buy a winter jacket.
Let's assume you don't know much about it, so using one of almighty search engines is a good idea. The first keyword you search for is "winter jackets". Clicking on one of the links will get you to category page where you may find important information about how to choose a jacket. Then you browse couple of products and close the window. The first session is over.
After couple of days you're successfully retargeted by display ad and come back to the one of the products you've previously viewed. Now you do a bit more product digging and you become familiar with the website. Now you're pretty much decided what and where to buy, but you want to double-check at home with your husband/wife.
So, you have your final approval :), now just go directly to the website typing its website name into browser, landing on a homepage buying product and then just waiting for a delivery. The simple picture below visualizes the buying process.
Turning back to standard "Landing pages" report and considering scenario above only the last landing page gets credit for the conversion. The exact same bad last-click attribution as in case of traffic sources. Does it mean that category pages or product details don't work as landing pages? I wouldn't be so sure.
The whole point is that conversion rate isn't a good metric for landing page evaluation. So, you might ask, where to find the meaningful data? It'll require slightly more work and you won't get easily readable data. You have to go to multichannel report.
You might be surprised, but it's also very useful for landing page analysis. Just change primary dimension from "Channel" to "Landing page URL path". It's great if you know about this approach before the implementation. Implementing page types (homepage, category page, product detail, etc.) either as URL in separated GA view or via UTM_ID and custom dimension, will prepare the data for easier interpretation. Then you can expect something similar to this:
If you started to actively evaluate content you've probably also used Page Value. A great metric, but it's necessary to understand how it works. Let's just quickly remind how it's calculated. The logic of metric is to assign monetary value to pages according to their contribution on the conversion.
Simple scenario. The same user comes twice to the website:
The first session starts on homepage, then category page and ends with blogpost to which user was linked from the category page.
The second session starts exactly the same way. User lands on a homepage, then goes to category page, then is linked to a product detail and finally purchases.
Page value is calculated for every page as "Goal value" + "Transaction revenue" of the sessions during which page was viewed divided by unique page views. The goal value and revenue is counted only when a page view occurred before transaction or goal!
So for the case above our page values are:
|Page||Transaction revenue||Unique pageviews||Page value|
Does it seem familiar? The same problem as in previous case with landing pages and conversion rate. Pages substituted for landing pages. The pages viewed during the last session will get the most credit. Does it automatically mean that page product is the best? Or that blogpost doesn't work, at all? I wouldn't be so sure. Again, it will require a bit more advanced approach to understand the data correctly.
Slightly different case, but also it's kind of attribution! Sometimes it's worth to rememeber this one when setting up expectations before launching new website.
Simple scenario. Website redesign focused on mobile experience. Months of work and the closer to the release date the higher amount of sleepless nights. The D-day has come. Everything works perfectly, your tracking is fine too and you're going to check the business impact.
And here comes the moment of truth. Shareholders expectation was 30% growth in mobile devices revenue and no such a thing happened. You've hardly reached 5% and now it's time for a panic. Or maybe it's not. This is exactly the point when understanding attribution can save your ass.
The common case when you improve mobile experience is that desktop conversion rate also pops up. It can happen on the market where users are not that used to buy on mobile, so they only do research on mobile and mostly buy on desktop.
What to remember?
These were just 3 examples of where its worth understanding the attribution. And there are many more. Simo Ahava described how attribution works in enhanced e-commerce metrics. Rule of thumb is to think about attribution every time you're looking on the report where USER mostly makes more interactions with website until achieves website goal.
Happy attributing :)