Performance Max Campaigns: Testing a Narrow Focus

Performance Max campaigns are hardly new, but for many advertisers, they’re still shrouded in mystery. Much of this comes from the automated nature of the campaigns. But it also comes from the results.

We’ve heard from many companies and PPC specialists that they struggle to get consistent results with their Performance Max (PMax) campaigns. Often this comes over time, where results start drifting lower, or with scale as budgets grow. While anecdotal, we often see PMax campaigns start off great, but over the course of months perform worse.

One thing we’ve consistently noticed is the best practices for building Performance Max campaigns focus on fairly broad targeting. The idea being that you give Google a lot of options for how to target your ads, and let the algorithm do the rest. We’re not sure this is always the best approach.

Better PMax Reporting Shapes Campaign Structure

As Google has evolved the reporting for Performance Max campaigns, some very useful insights have emerged for our clients. The most illuminating coming from the Search Term Insights report.

Using an online retailer as an example, we saw their own brand was the top converting term (a common occurrence if you do not exclude your brand and/or use the “bid for new customers” setting). But beyond that, we would’ve expected a lot more intent-focused searches around basic product categories. Instead, we saw much more specific searches focused on brands of products the site carries.

Structuring Performance Max for More Specific Targeting

Knowing this, we can take two approaches to better align our PMax asset groups with the way the campaigns target customers.

1. The Google way – allowing automation to take control. This requires using expansion settings for final URLs and assets. E.g. if we are Zappos and sell a lot of basketball shoes, we’d have a PMax campaign for Basketball Shoes and allow Google to craft the targeting, messaging and final URL choice for someone looking for Jordan Brand shoes.
   The Traditional Way.

The “ah ha” moment we had in thinking about this, is as search marketers we’ve solved this challenge years ago with search campaigns. So that’s exactly what we’ve done.

Instead of going broad and using automation to get more specific, we’re starting specific.

In the Zappos hypothetical scenario, we’re doing a PMax campaign for “basketball shoes” but then doing an asset group for our brands, e.g. Nike, Jordan, Adidas, etc.

This allows us to pick the best landing page for those customers and craft the right messaging in our ads. The latter is especially important when working with brands that have strong guidelines around messaging. We then narrow our listing groups to only show the products in that brand.

We’re taking the approach of trying not to go too much deeper than that, to still allow Google’s systems to have flexibility to find customers, but also to avoid competition across asset groups.

Audience Signals

Our approach so far has been to use a two-part audience signal.

1. A custom segment including the major keywords we’d typically target if this were a search ad group (e.g. “jordan low tops”)
2. A site visor audience for the pages that represent the items we’re targeting (at minimum, your final URL)

Assets

Because we’re targeting a specific brand+product category combination and limiting where on a site Google can send people, we can be much more specific with our ad copy, images, etc. to only speak to people interested in the thing they’re looking for. Our goal here was to increase clickthrough rates but also our optimization scores.

Performance Max Results

One thing we’ve pretty consistently seen with Performance Max campaigns, is when we exclude the client’s brand, we see better results in terms of return on ad spend (ROAS) via traditional search campaigns. While we need more time to know how well we can address the issues of scaling budgets and long-term stability, we wanted to use an A/B with our search campaigns as our initial measure of success.

Below are results for one such campaign. We’re including CPC because it’s an important factor in the results. The flexibility in where ads appear helps drive traffic at a lower cost, which can make up for lower conversion rates.

So far this has been representative of how this approach works for us. We’ll update as we learn more!