Accurate conversion tracking and effective conversion attribution are two of the biggest challenges in paid search. After all, your optimization decisions are only as good as the data associated with your campaigns. Over the years, I’ve worked with many brands when their conversion tracking dropped off the website or was installed on all pages with over-reported conversions. One brand even had duplicate transactions and revenue reporting for an entire month, in Google Analytics, and imported that false revenue into Google Ads.
Recently, I worked with two brands that seemed to have a conversion tracking problem. After further investigation, both brands had different attribution discrepancies between Google Ads and Google Analytics, which is actually very common. In this blog, I’ll walk you through both of those scenarios and how Google Attribution (a standalone tool within Google Analytics) can help.
For one brand, the revenue reported in Google Ads was always significantly higher than what Analytics was reporting for Google cpc traffic. To dig into this discrepancy further, I imported Google Analytics transactions into Google Ads (with the setting “Include In Convs” set to No to avoid double reporting).
As you can see, according to Google Analytics, transactions are lower, and revenue is nearly $300k lower than what Google Ads tracking is reporting. (Note: in this example, the brand is using BOTH the Google Ads tag and the Google Analytics tag on their site.)
I contacted Google support to get a better understanding of the differences in reporting between Google Ads and Analytics. They said it was due to different attribution models.
This answer did not make complete sense to me because the attribution models, shown in the screenshot above, only affect how conversion credit is attributed across the campaigns within the Google Ads UI, meaning that the total revenue and/or total conversions attributed to Google Ads campaigns (google / cpc in GA) would be the same.
I continued to dig into this issue since I still didn’t feel like I had the full answer.
In another account, I imported Analytics transactions to observe the differences again. Despite all the settings, including the chosen attribution model for each conversion action, being exactly the same, there was still an $80,000 difference in revenue.
So, I asked one of our Google dedicated representatives to see if they could provide a better answer to explain the differences. They connected me with a Google Analytics lead who was able to provide a clear explanation.
The issue is that the two platforms use different reporting methods. Google Analytics uses “conversion time” reporting (e.g. the date the transaction occurred) while Google Ads uses “interaction time” reporting (e.g. the date(s) of the associated ad click(s)). Even though the transactions are recorded via the Google Analytics tag, as soon as they are imported into Google Ads, the system automatically converts the reporting of those transactions to ‘interaction time’.
The difference between “interaction time” and “conversion time” reporting produces what’s known as conversion lag. For example, if I clicked on an ad May 1st and later purchased on May 6th, that transaction (and associated revenue) would be reported in Analytics for May 6th and in Google Ads, it would be reported for May 1st. You can get a sense for your conversion lag by running a report using the “days to conversion” field.
As you mentioned with attribution models, the total amount of conversions aren’t lost, but they are potentially moved outside of the date range you use when comparing the two platforms. For example, the imported transactions were created at the end of May. However, in Google Ads, conversion values are reported as far back as March. That’s because transactions happening now are being associated with past dates when the last clicks occurred.
This was an important, missing puzzle piece.
Comparing GA Conversions Imported Into Google Ads vs. the Original Conversions in GA
For a better understanding of what’s going on, let’s compare the imported GA conversions within Google Ads to the same conversions in the GA UI. First, run a Conversion action report in Google Ads with all conversion values (by conv. time). In this example, revenue associated with GA transactions was $527,386.75.
Next, run an MCF report in GA, under Model Comparison Tool, and filter it to Last Non-Direct Click and the total value here between Google CPC and Google Display is $517,063.89.
The difference between these numbers is around $10,322. This could be explained by differences between time zones or cookies dropping.
What this information tells me is that Google Analytics imports conversion data into Google Ads using the last non-direct data from GA. That is confusing since we see “Linear” as the attribution model listed in Google Ads for those GA transactions (Scenario 1 data).
But what that means is that the conversions imported from GA will be attributed to the Google Ads campaigns linearly, not that the conversions were originally attributed in a linear way in GA, before being imported.
So, my next question was, “Which is correct: Google Ads tracking or Google Analytics tracking?” The Analytics lead added another note about the GA. vs. Google Ads conversion discrepancy, and then answered my question:
The discrepancy is due to the deduping that Analytics does. Google Ads tracking is a single channel, meaning it isn’t aware of involvement of any channels (Direct, Organic, other paid channels, etc.) outside of Google. Google Analytics on the other hand, de-dupes when there are multiple channels involved. Analytics runs on a Last Non-Direct Click model (last click, ignoring Direct unless that’s the only channel involved). See the screenshot below for context:
To your question about which is the right one, the answer is neither exactly. Google Ads, being a single channel, takes too much credit. Google Analytics, deduping on a last click model, gives too little credit. The truth is somewhere in the middle. The best way to report on this is through the new Attribution beta project, which you can find in the left-hand navigation menu of Google Analytics. This lets you pick different attribution models (like DDA) to use on the Analytics data, so the deduping isn’t based on last click. The Attribution beta is separate from Analytics, so it won’t impact or change anything in your Analytics reports. Instead, it’s a self-contained environment to better attribute data. You can even switch between conversion time and interaction time reporting within the interface.
Okay, that’s a lot. So let’s review the various factors that affect the discrepancy between Google Ads and Google Analytics conversion reporting:
- GA can show you multiple attribution models via the Model Comparison tool, but when exporting the conversion data to Google Ads, it sends the conversions or sales attributed to Google Ads using a last non-direct click attribution model.
- GA accounts for multiple channels, not just Google.
- So, in GA, when MULTIPLE channels are factored into a conversion path, a Google Ads campaign is less likely to be the last, non-direct click (compared to the limited conversion path view that Google Ads has) and thus Google Ads campaigns gets less credit, in total, in GA, before that data is exported to Google Ads.
- Google Ads uses interaction time reporting and GA uses conversion time reporting. The effect of this on reporting discrepancies varies based on the sales or conversion cycle. The longer the lag time between a click and the final conversion, the larger the discrepancy might be between conversion data imported from GA into Google Ads and the conversion data reported directly in Google Ads.
So now we understand the differences in how revenue is attributed and the differences between using Google Ads reporting vs Google Analytics reporting.
My next question was which one should you use for optimizing Google Ads campaigns?
The Google rep said he would always recommend using Google Ads because it provides more data points. However, this also means if your ROAS is 425% in Google Ads and only 350% in Google Analytics, the true ROAS probably lies somewhere in the middle and you should take that into account when setting a ROAS goal for your campaigns.
The Analytics lead did make a pretty good case to explore the Google Attribution beta to gain additional insights into conversion data. This beta can be found in your Analytics account on the side menu, towards the bottom.
Important Tip: You must have account-level edit access in order to Create an Attribution project. You can find the full instructions on how to set up your account here.
Once you set up the account, it will take some time before the data begins to populate. You may see something like this screen when you first set up the account.
Google Attribution Conversion Paths
One primary difference between Google Analytics conversion paths reports and Google Attribution conversion path reporting is that Google Attribution shows an exact percentage of credit given to a channel within a conversion path. See the example below. In the first listing, Paid Search is given 100% of the credit since it ignores the Direct channel.
The Google Analytics conversion path report doesn’t show percentages.
So, what did I learn throughout this experience? There is a difference in Google Ads between using Google’s website tracking and importing Analytics transactions. It may be better to use Google Ads tracking because it provides more data points for the system in automated bidding or general decision-making in the account. However, if you have strict ROAS targets, you may want to set the target a bit higher, especially if Google Ads is reporting a significant difference in revenue compared to Analytics. And if you do see a significant difference in revenue or conversions between the two, keep in mind that lag time created by interaction vs. conversion-time reporting.
Setting up the Google Attribution beta now may be a good idea so it can begin collecting data. This stand-alone platform may give you some additional insights. You may also want to allow yourself some time to explore the program before deciding to import Google Attribution data into Google Ads (when it moves out of beta). Also, keep in mind, this is still in beta and may not be the final version.