Leverage Attribution Modeling for PPC
Ryan Moothart May 28 2014
Are you using attribution modeling or assisted conversion data in your PPC reports? If you’re not, you’re missing out on some golden opportunities to leverage data to your advantage.
We all know the LNDC model (last non-direct click model—the default used by Google Analytics) is not comprehensive enough for us to see the big picture. Continuing to rely on this model for year-over-year analysis may be necessary for the sake of consistent reporting; for a long time, that was the only model Google Analytics offered. But by also highlighting the important trends witnessed by comparing attribution models, you can develop a more comprehensive PPC strategy and tell a whole new story with your PPC reports.
Here’s a typical PPC segment report using Google Analytics’ default LNDC attribution model:
This data tells a very predictable story:
• Branded keywords dominate, but expansion opportunities are limited.
• Non-branded keywords can spend a lot of money, but don’t get much in return.
• Product Listing Ads are robust, high-spending, and provide decent return.
• Earth is a sphere, the sky is blue, and global warming is actually happening.
These points of interest are things just about every single PPC strategist in the world would have been able to guess before even looking into this account. By leaving our attribution modeling at this, we miss out on the bigger picture. So how does this story change if we apply a different model and compare it to the default?
Here’s the same data comparing the LNDC model with the First-Click model, which gives all conversion credit to the introductory click:
The conversion value increases by over 30% for Product Listing Ads and by over 40% for non-branded keywords. Right away, this tells us that Product Listing Ads and non-branded keywords are much better at introducing quality users into the site’s marketing funnel than they are at closing the deal and acquiring the conversion directly. Now you, as a PPC strategist, know that there are opportunities within these two segments to strengthen your overall conversion strategy—opportunities you may not have seen by just looking at the default model.
What if we looked at another model? Here’s the same data comparing the LNDC model with the Linear model, which gives equal credit to each visit in the conversion process:
The conversion value for branded keywords decreases by over 20% while the other two segments stay the same. This tells us that branded keywords don’t play as much of a supporting role as our other segments; they close out the conversion process more often than introduce or assist.
From here, you can use this insight to optimize your ad copy and/or landing page strategy. If you know branded keywords close the deal, ensure all of your ad and landing page messaging for these campaigns are designed to close the deal: offer incentives or discounts.
Using and comparing attribution models is paramount to developing a robust, comprehensive PPC strategy. Use them to your advantage, include them in your reporting, and tell the full story from multiple points of view.
How do you use attribution modeling to your advantage? Do you already use model comparisons in your PPC reports? Share your answers and ideas below.
Ryan is a PPC Architect and has been with Portent since 2010. He has over seven years of hands-on PPC experience including large-scale e-commerce, international B2B lead gen, and everything in between. Graduating from Willamette University with a BA in Rhetoric and Media Studies, he became a published author in 2016 with the release of his book, Towards Cascadia, which is a non-fiction exploration of Pacific Northwest identity, bioregionalism, and nationhood. He and his husband, Paul, enjoy traveling and are avid followers of Sounders FC, Seattle’s Major League Soccer club. Read More