What is Attribution Modeling?
For the purposes of this discussion, we’ll start here. Attribution modelling is simply a way of looking at who gets credit for a sale or lead. Depending on the model, someone might get credit for the whole transaction, or just part of it. There is no “correct” model, and selecting the most appropriate model will depend upon your business.
As a rule, if you advertise a lot of different places (Google, Bing, Facebook, Twitter, blogs… all the different digital channels), you’ll want to use a model that attributes a portion of the transaction to each ad that got clicked. This way, you’ll be able to better understand the return on your ad spend.
In the old days (so like 10 years ago), all we really had was either first or last click attribution. So the sale would either be assigned to the first or last channel that had a click in the customer path to purchase. This was ok, since we could at least understand the impact of part of our advertising budget.
What if there were clicks in between though? Well, they simply didn’t count. Which really isn’t fair, since they did help us stay top of mind with the consumer, even if they didn’t transact right that moment.
What Attribution Model do my Tools Use?
This is going to depend in some cases, but we’ll start with one absolute: Almost every report in your Google Analytics account attributes conversions to the last non-direct click. So in other words, if you’re looking at your standard Source/Medium report in the Acquisition portion of your account, anything in the Google CPC row is a visit where a Google ad was the last click on a measurable source. A user could come back directly to your website (typing in your full website address for the most part) and NOT overwrite Google CPC.
However, if a user finds you through organic search, then hits a remarketing ad or comes in through a coupon website, then that sale will get re-attributed to a new source/medium.
This is where the disconnect between Google Ads and Analytics comes in.
If you’ve set the attribution model to anything other than last click in your Google Ads account, these numbers will no longer match up. (Keep in mind, they will almost never match perfectly for a number of reasons).
In particular, if your agency or Google Ads representative talked you into the Data Driven attribution model, you’re going to be way off when you try to reconcile Ads and Analytics. The Data Driven model is a predictive model, and tries to show value across multiple ads or keywords based on consumer behavior. It doesn’t report a “real number” in the sense that we’re used to talking about them.
Reconciling Google Ads and Google Analytics
Let’s say for a given time period, Google Ads is reporting $1.7m in revenue, and Google Analytics is telling you your Ads only garnered $1.3 million. That’s a big discrepancy, so if people in your organization are looking at these things actively, it’s bound to become a topic of discussion.
So which do we believe? The answer is neither, without digging a little further.
Since we know that the Analytics report is last non-direct click, and in this case the Google Ads report is data-driven, we have to look at a different report in Google Analytics to find that additional $400,000 in revenue. This report is in the Conversions section of Analytics, as opposed to acquisition:
We need to do a few things to this report to find the hidden money (which we will soon see isn’t really hidden at all, but rather being credited to another one of your marketing channels). Here’s how you’ll want to set up the Top Conversion Paths report:
Your conversion type selected should only be the one that has actual revenue associated with it. Not Smart Goals, etc. You’ll note I have Type selected as All as opposed to Google Ads. That’s just so that I could get screen shots I can share publicly. You can save a step by selecting Google Ads. It’s also important that you select Path Length as All, to capture all the revenue.
Once set, something magical happens:
We’re now only seeing a discrepancy of $12,000, as opposed to $400,000. That’s a lot easier to swallow. So now that we know our tracking is in fact working properly, and we aren’t ACTUALLY missing transactions and revenue, it begs the question:
Why is traditional Google Analytics reporting off by $400,000?
From the report we just created, let’s walk through some examples of how this happens.
Attribution and Transactions: Who Gets the Credit?
Here’s 9 paths consumers took to order from an e-commerce website. You’ll note of course that they all include at least one click from a Google Ad:
In ALL of the above examples, a data driven attribution model in Google Ads is going to take credit for the sale, in one or more Ad Groups. Line 2 would split credit between a few different clicks in your Google Ads reporting, as would lines 4 and 9. And those three transactions would all be credited to Google CPC in your traditional Analytics reports.
Line 5 however, would show credit to 1 ad in your Google Ads account, but would actually put the transaction and the revenue in your Google Organic channel in Analytics, since organic is the last non-direct click.
Line 3 would still be Google Ads in ALL reports, since the last click is direct, so Analytics ignores it for attribution.
If you have a really strong affiliate program with lots of coupon and loyalty sites involved, you might see a bunch of transactions like number 6 above.
So in the end, the reason your Google Ads and Analytics accounts don’t show the same numbers is likely attribution. Start by checking with the steps outlined above before looking at your tracking code and other somewhat less likely scenarios.