Over the years, I’ve come to the realization that pay-per-click (PPC) is perhaps my favorite form of digital advertising, based entirely on the power of conversion tracking.
PPC advertising is the one method that offers, in my opinion, an accurate way of quantifying success. Other systems are in place that attempt to quantify the impact a billboard or a TV spot has on sales or leads, but they just aren’t able to do it with the level of accuracy provided by PPC conversion tracking.
Often in the world of PPC you will try to get exposure in as many parts of the shopping funnel as possible. You can advertise at the top of the purchase funnel to generate brand awareness on broader search terms, and at the other end of the purchase funnel with remarketing to site visitors and advertising on your own branded search terms.
And while a customer might have converted after clicking on one ad generated by one keyword search, it doesn’t happen very often any more. It’s more likely that they clicked on multiple ads across many days on several devices. So how do you give credit where credit is due?
Google Ads’ PPC attribution models allow an advertiser to decide how credit should be divvied up for each conversion.
There are six standard models for PPC attribution. Let’s start with the most basic:
1. Last Click
Gives all credit for the conversion to the last-clicked ad and corresponding keyword.
2. First Click
Gives all credit for the conversion to the first-clicked ad and corresponding keyword.
These two models are very straightforward, and probably the most commonly used. Last-click attribution gives credit to the last touchpoint in a customer’s journey, while first click gives all the credit to the customer’s first one. Last-click attribution is the default model for Google Ads. Google Analytics, on the other hand, uses first-click attribution. That may be why sometimes your Google Ads and Google Analytics data don’t always match up.
Last-click attribution is the most problematic and should rarely, if ever, be used. Let’s say a customer does have multiple touchpoints with your ads during the conversion process, the last touchpoint before a conversion will probably be a branded search, or a remarketing ad. The customer would have already interacted with your ads and visited your site, so they’ll be actively searching for you. Therefore, last click might overinflate the impact of remarketing or branded search terms when in reality other touchpoints also played a role.
First-click attribution works if your business goal is centered around growth. Typically the first touchpoint a customer will have with your advertising efforts will be more towards the top of the purchase funnel. Because of this 100% of the conversion credit will typically go to the shorter-tail, broader search terms. Which could lead to some inefficiencies over time. This is the second most problematic attribution model and should rarely be used.
Lets go over the other four, which more accurately help you see your attribution path.
3. Linear
Distributes the credit for the conversion equally across all ad interactions on the path.
The linear attribution model definitely caters to the idea that all of your ads play a role in a conversion. This model is a bit more intricate then the first-click or last-click models, but it’s the simplest of the multi-touch attribution models. It gives equal credit to every single touchpoint a customer has with your ads.
For example, let’s say that during a customer’s path to a conversion they click on four of your ads. Each click would get 25% of the credit for the conversion, or 0.25 conversions. This explains why you’ll sometimes see a keyword or product group with a partial conversion.
One potential downside of this attribution model is that you might not be able to correctly pinpoint the keywords or search terms that are doing most of the heavy lifting and truly driving the most conversions. But it won’t overlook any ads that helped at one point or another in the conversion process.
4. Time Decay
Gives more credit to ad interactions that happened closer in time to the conversion. Credit is distributed using a 7-day half-life. In other words, an ad interaction eight days before a conversion gets half as much credit as an ad interaction one day before a conversion.
The time-decay attribution model adds yet another layer of intricacy into the attribution puzzle. Like the linear model it’s also a multi-touch model, but it gives more credit to touchpoints that occur closer to the conversion date, with diminishing values given to the earlier touchpoints.
Essentially it’s a combination of the linear model and the last-click model. And because of its similarity to the last-click attribution model, you see issues with overvaluing the effectiveness of branded keywords and remarketing. However it also gives some credit to the top-of-funnel keywords as well.
5. Position-based
Gives 40% of credit to both the first and last ad interactions and corresponding keywords, with the remaining 20% spread out across the other interactions on the path.
The position-based attribution model gives credit to the top-of-funnel search terms that might introduce a customer to your website and the last search before the actual conversion. It considers the first and last touchpoints to be the key touchpoints a customer has in the conversion process. Obviously this leads to a potential undervaluing of the touchpoints in the middle, but as you’ve seen so far, no attribution model will be able to perfectly attribute the proper credit to every touch point.
However, this is the attribution model that makes the most sense if you’re not able to use the data-driven attribution model coming up next.
6. Data-driven
Distributes credit for the conversion based on your past data for this conversion action. It’s different from the other models, in that it uses your account’s data to calculate the actual contribution of each interaction across the conversion path.
The data-driven attribution model allows machine-learning technology to dole out the credit for each conversion. It takes the data from your unique ad account and determines which keywords, ads, ad groups, and campaigns should get credit for driving conversions.
This model is only available to Google Ads accounts with enough data: over 15,000 clicks and 600 conversions within a 30-day period, according to Google’s general guidelines. The more volume your account gets, the more accurate the machine learning will be in attributing credit.
If you have the ability to use the data-driven attribution model, use it. There is a little bit of a lag in reporting, but it’s worth the wait for better attribution.
If your data level drops below 10,000 clicks or 400 conversions within a 30-day period, you’ll be alerted that you have 30 days to get back above to the threshold. If you don’t, your attribution model will default to linear.
Testing Attribution Models
If you’re not sure which model is right for you, you can compare them in the Google Ads interface.Click on the “Tools and Settings” button in the top right of the screen to display the drop-down menu. Select “Attribution” under the “Measurement” heading, then select “Model comparison” in the left-hand navigation. You’ll be able to cross-examine how each attribution model compares to another. You can also look at the top paths that a customer has taken to a conversion.
It’s an excellent way to see which attribution model makes the most sense for your account, based on the goals you have set for your campaigns.Whatever attribution model you use will also affect any smart bidding strategies you may be running, like Target Cost per Acquisition, Target Return on Ad Spend, or Maximize Conversions. If these campaigns aren’t performing well, consider your business goals and pick a corresponding attribution model before pausing entirely. And remember, Google says you should allow at least 14 days of learning time after making a change before making further changes to your campaigns or just giving up.
