Advertisers pumped up to 90% of their Google ad spend into its AI tool, and some are now questioning whether it’s actually performing.
- Google is encouraging advertisers to use a tool called Performance Max, which uses artificial intelligence to place ads.
- Some advertisers intend to reduce their spending on PMax because they cannot control where their ads appear.
- These advertisers claim that measuring different ad formats and targeting ads is difficult.
After Apple’s privacy update made it difficult to target consumers online, Google has bet on artificial intelligence to improve the performance of ad campaigns and restore ad revenues.
One of its most important products is Performance Max, a two-year-old tool that uses AI to determine where ads should run across Google sites such as YouTube, search, and display. Google has recently begun to promote this product to smaller and mid-sized marketers.
According to multiple ad agencies, retailers are particularly interested in Performance Max, which they use to find customers and drive sales.
Tinuiti, for example, reported that during the third quarter, more than 90% of retailers’ spend on Google’s popular shopping listing ads was purchased through Performance Max. According to Tinuiti, Performance Max increased retailers’ spending on Google’s shopping ad products by 16% year on year.
Other performance-driven ad agencies have increased their investment in this tool as well. According to marketing firm Belardi Wong, 68% of Google ads for retailers are purchased using Performance Max, up from 50% last year. According to Arm Candy, Performance Max handles 45% of retailers’ Google spend. According to a fourth agency, Performance Max receives 60% of retailers’ Google shopping ad spend.
Some advertisers are wary of Google’s AI
However, as Performance Max spreads throughout the ad ecosystem, advertisers are growing wary of its claim to algorithmically improve ad performance with no transparency into how it does it or where the ads actually run.
Arm Candy will reduce its Performance Max spend this year, according to Zach Thompson, its director of ad operations, because it cannot control where Google’s AI decides to place ads, making it difficult to track and measure. Performance Max, for example, frequently purchases a mix of YouTube, search, and display advertising for a campaign, but advertisers use different formats for different goals.
For example, YouTube is commonly used to increase brand awareness, whereas a search ad is used to increase sales or conversions.
Performance Max’s buying can also be inconsistent, with the company focusing on YouTube inventory at times and display ads at others, according to Thomas.
According to Arm Candy, he will redirect Performance Max spend to traditional campaigns managed by humans. According to him, Arm Candy has also developed its own internal tool that analyzes where ads ran to get a more granular view of Performance Max.
“We’ve found that turning off Performance Max leads to better performance,” said Thomas.
Separately, Dentsu-owned digital agency Merkle discovered a drop in performance after using Google Performance Max, according to trade publication MediaPost.
According to Brendon Kraham, Google’s VP of global ads for search and commerce, Performance Max intentionally buys a mix of different ad formats, and performance dips are based on how many ads are available for it to buy.
An executive at a second performance ad agency, who asked to remain anonymous in order to protect their relationship with Google, also stated that their company intends to pull spend from Performance Max and reallocate it to managed shopping campaigns.
The source specifically stated that the return on ad spend of Performance Max campaigns began to slow down this summer. ROAS is an important metric that determines how many sales can be attributed to each dollar spent.
According to this source, after the decline, Performance Max began to improve its results by purchasing brand search campaigns. Brand search campaigns occur when businesses purchase search ads against their own brand names, such as Hershey’s purchasing “Hershey’s” keywords. These campaigns are usually successful.
“Brand search terms were being pulled into Performance Max, which is what was making Performance Max look so good,” said a representative from the agency.
The agency executive was taken aback when he discovered that Performance Max was purchasing brand search terms. Marketers typically prefer that these campaigns be managed by humans rather than AI to ensure that budgets are not depleted too quickly.
According to Google’s Kraham, Performance Max can supplement other search budgets. He also stated that buying search ads in conjunction with Performance Max is frequently a best practice for marketers and can increase conversions by 18% on average.
“PMax may buy additional impressions and search that you were not buying, but that will deliver the exact ROI that you cared about,” he went on to say.
Advertisers that like Google’s AI say it helps them find new customers
Google has attempted to allay advertisers’ concerns about what Performance Max purchases. Google introduced the ability to exclude specific keywords from Performance Max campaigns in July. Other recent enhancements include the ability for marketers to limit Performance Max to only finding new customers.
And, despite the fact that they can’t fully control the campaigns, some agencies, such as Belardi Wong, continue to spend with Performance Max because it has helped them find new customers by locating people higher up in the so-called “sales funnel.” These customers are just starting to think about making a purchase.
“That kind of funnel expansion — as challenging as it is for marketers in losing some visibility and control — is a reason why we’re shifting more spend,” said Calla Murphy, Belardi Wong’s VP of digital strategy and integrated marketing.