In the last blog, Rylea Sloat did a great job of summarizing AdWords attribution and how it can change the performance of the campaigns. If you haven’t read it yet, start there and then come back to read about how attribution modeling impacts the performance of branded terms.
First, let’s start with a quiz about attribution modeling. The above diagram shows three different click paths. For each click path which click is credited with the conversion? Of course the answer is “It depends.”
Why? Because it depends on the conversion model that you’ve selected. Using AdWords Last Click model, the last click in the click path gets 100% of the conversion:
The problem with this model is that it gives 100% of the conversion to the branded terms. The problem with this is that it makes that performance of the branded terms seem very good by creating the illusion that they are only terms that convert. I’ve heard advertisers say, “I should only advertise on my own branded terms,” or “Advertising on non-branded terms is too expensive.” However, many times there are other clicks that led to the branded term click.
Prior to AdWords changes in Attribution Modeling, the purchase funnel was a good theory, but couldn’t really be optimized for without critical analysis of click paths which is too difficult for most businesses.
Consider the following diagram showing the buyer’s journey along the purchase funnel. They begin researching with generic terms and become aware of products that exist. They then continue to do research about product features and make comparisons with competing products. Then they consider the three “P’s” - product, price, promotion - of their top choices before finally coming to a purchasing decision.
On the AdWords side, this progresses from generic “top of funnel” terms (awareness), and then becomes category or market terms (research), to specific product terms (brand or product terms).
Without Awareness at the top of the funnel, the buyer wouldn’t progress. AdWords attribution has made it possible to distribute credit throughout the purchase funnel. But how much credit should we give to each stage? As always, it depends!
As a general rule, the two things we’re interested in are average number of clicks in the path and the weight of branded terms.
Path Length: You can find this report under Tools >> Attribution. If your average click path is very close to “1,” then select whichever model you want, it doesn’t matter. If it is close to or exceeding “2” then you’ll want to choose another model.
Branded Terms: If you estimate that over 50% of your conversions come from branded terms then you’ll want to “devalue” these terms as much as possible. Don’t worry, they will still get plenty of conversions and still continue to perform well. But what we’re more interested in: finding the top and middle funnel clicks that are driving toward the branded terms.
In order to devalue these terms, we would select the Linear model. With a path length greater than “2” it would put the greatest emphasis on early clicks and the least amount of emphasis on the mostly-branded last click.
For businesses who don’t have as many conversions from branded terms, then you would use the Position Based model. This would emphasize the click that gets buyers into the funnel and the clicks that convert in the end.
This diagram shows the difference using the two models. Notice how the branded terms are devalued using the Linear model when there are more than 2-clicks in the path? This becomes more obvious on even longer click paths.
In summary, branded terms are great, but their conversion performance alone doesn’t tell the full story. By distributing conversions to the top and middle of the path, you’ll have better insight on how other keywords/ad groups are contributing to the success of the campaign.