Enhanced Campaigns have been pretty controversial, as I’m sure you’re aware. While some of us here at PPC Hero have experienced some success due to the massive influx of mobile traffic, others are still wrestling with the total loss of device segmentation and traffic control that we previously enjoyed.
Chief among our concerns are business models that focus solely on mobile traffic. We’ve got a few clients with that business model in-house, and Enhanced Campaigns have not been kind to them. While you can maneuver around the clumsy bid modifier system to limit desktop traffic as much as possible (credit goes to Jake for actually pulling it off – he’ll tell you about it soon), it’s still a major oversight in the AdWords system.
One of the potential workarounds to aid with your Device segmentation woes is the Ad Device Preference. Theoretically, this system allows you to define certain ads as Mobile-preferred, displaying different ads to Desktop/Tablet and Mobile users — and allowing you to take advantage of Mobile-specific landing pages.
So, does it actually work?
To find out, we’ll be comparing two different accounts. The first one is a traditional PPC account, focused mainly on desktops with some mobile thrown in. It’s all lead generation, focusing on tightly-controlled geographic targets. It’s been opted in to Enhanced Campaigns for the last couple of months, so performance has had time to stabilize. Mobile bid modifiers range from a meager +3% bump, down to a -80% adjustment.
Here’s how the device performance broke down for the last 30 days compared to the Device preference on the ad:
This is surprisingly segmented. Over the last 30 days, traffic was funneled correctly between Mobile-preferred and All device-preferred ads. The only Mobile performance statistics accrued in the “All” category are by design, coming from Image Ad-based Ad Groups. In this account, every instance of an Ad Group containing both “All” and “Mobile-preferred” ads, traffic was allocated correctly.
Our second account is Mobile-only (or as much as it can be in a post-Enhanced Campaigns world). As such, it makes heavy use of positive mobile bid modifiers and depressed general bids to limit totally irrelevant desktop traffic – which cannot be removed, no matter what. Here’s how the Device statistics performed relative to device preference over the last 30 days:
This is more troubling. Not only do we see evidence of Mobile-preferred ads serving to Desktops and Tablets, “All” ads are still serving to Mobile devices. For businesses with a need for distinct Mobile-specific landing pages, this lack of segmentation is a problem.
In my initial investigation, I wanted to make sure this wasn’t a phenomenon that occurred due to the Enhanced Campaigns forced transition on July 22nd. To that end, here are the same account statistics over just the last five days:
The trend persists even to this day in a very fresh account snapshot. In addition, each Ad Group contains both an “All” and a “Mobile” preferred ad – so this problem isn’t due to oversight, nor is it an account structure issue.
So what accounts for the difference in traffic segmentation in these two accounts? I have a theory.
While these accounts have different aims, and leverage their mobile presence to different ends, they have one main difference that could account for this discrepancy: Ad Rotation.
Account #1 has properly segmented mobile traffic and makes use of Rotate Indefinitely, Rotate for 90 Days, or Optimize for Clicks.
Account #2 has poor device segmentation and primarily makes use of “Optimize for conversions” as its Ad Rotation setting.
Examining the data above for Account #2, two things stick out: Impression volume for Mobile-preferred ads is significantly higher than “All” device-preferred ads, but Click and Conversion Volume usually favors “All” device-preferred ads. This is in line with my theory that “Optimize for conversions” ignores device preference. To that end, it will serve whichever ad (and corresponding landing page) it thinks will get you more conversions – regardless of your stated preferences.
This is something we’re going to test going forward, so stay tuned. We’ll have some followup data in the coming months to see if we’re right. In the meantime, if you’ve experienced the same phenomenon in your own accounts, let us know in the comments. Thanks for reading!