An Australian consumer services business, operating nationally for over 10 years, had been growing steadily, with leads and sales coming from a mix of Google Ads, organic search and Facebook. Google Ads had been managed in-house for several years alongside many other marketing initiatives, and while it wasn’t perfect, it had been working reasonably well. Then a Google sales rep suggested switching on Performance Max and AI Max, and enabling settings which allowed Google’s algorithm to choose the keywords, write the ad headlines, and choose which landing pages visitors were taken to. The campaigns were changed to follow Google’s advice, and some of the automatic “recommendations” in the Google Ads interface were also applied. Within 3 months, spend had risen, appointments had fallen, and far fewer of those appointments were turning into sales. With large fixed costs, the business went from profitable to unprofitable almost overnight, and if nothing changed, it would soon face difficult questions about its future. Knowing that something had gone badly wrong, the business researched Google Ads specialists carefully, then got in touch for a review of their campaigns.

Summary

Previously burnt by Performance Max and AI Max, but now back on track, with 59% more appointments turning into sales, at an estimated 42% lower cost per sale.

What went wrong

In the 3 months under Performance Max and AI Max:

• Google Ads spend rose by 21%
• Appointments from Google Ads fell by 10%
• Cost per appointment rose by 35%
• Follow-through from appointment to sale fell by 34%

Taken together, this meant an estimated 41% fewer sales from Google Ads, at roughly double the cost per sale.

Many hours were spent unravelling the account’s change history, to understand why. The changes in spend and conversion rate didn’t line up with any of the changes made by people. Instead, most of the volatility and decline appeared to have come from changes which Performance Max and AI Max had made on their own, to the searches being matched, the ad messages being shown, and the landing pages visitors were taken to.

The search query data showed hundreds of clicks from searches related to free advice and helplines. Performance Max and AI Max had also written ad messages which promoted free advice, and implied a free 24/7 hotline. This appears to have been taken from the business’s website without its intended context, since the business does offer 24/7 support, but only for its own customers. Many of these searches were also being matched to the brand campaign, which had AI Max enabled. Since brand campaigns typically have a lower cost per conversion than other campaigns, Google’s algorithm had bundled in large amounts of low-quality traffic alongside the brand keywords.

In short, Google’s algorithm had found whichever combination of searches, ad messages and landing pages brought in the most form fills for its cost per conversion target, while completely ignoring everything else, including whether those people were ever likely to buy. It was a textbook case of Goodhart’s law: when a measure becomes a target, it ceases to be a good measure.

How I helped

As a first step, over 100 negative keyword themes were added straight away, to cut out the free advice and helpline searches. The quality of enquiries rose significantly, but this alone wasn’t enough, as the account still had large structural problems. With just 15 campaigns and 30 ad groups, the top keyword was receiving 43.1% of all clicks, and hundreds of different search themes were all being shown the same ad.

So a plan was put together to move away from Performance Max and AI Max altogether. However, they couldn’t simply be switched off overnight, as the business needed a steady supply of appointments. Instead, the campaigns were separated out by state, and rolled out one state at a time over 6 weeks, with adjustments along the way to keep smart bidding relatively stable, and to avoid giving Google’s smart bidding algorithm too much information about how much the business was willing to pay for each enquiry. The new campaigns were built from the ground up:

• Number of campaigns increased from 15 to 322
• Number of ad groups increased from 30 to 10,150
• 10,150 ads, tailored to 60+ unique search themes and 100+ locations
• 100+ negative keyword themes

Campaign results

Clicks are now spread across a much wider range of keywords, each with an ad message more closely matched to the search, and the top keyword now receives just 8.3% of clicks, down from 43.1%. In the first full month after the transition with reliable tracking, compared to an average month under Performance Max and AI Max:

• Google Ads spend fell by 10%
• Clicks from Google Ads fell by 43%
• Appointments from Google Ads stayed about the same (down 2%)
• Appointment conversion rate rose by 71%
• Cost per appointment fell by 8%

Average cost per click (CPC) rose by 57%. This was expected, since many of the cheaper clicks which were cut out came from free advice and helpline searches, which were never likely to turn into sales.

Client results

Since the move away from Performance Max and AI Max, the quality of appointments has risen sharply, and over the last 2 months has been the highest in 2 years. Follow-through from appointment to sale is now 5% higher than before Performance Max and AI Max were switched on. Compared to an average month under Performance Max and AI Max:

• Follow-through from appointment to sale rose by 59%
• Cost per sale from Google Ads fell by an estimated 42%
• Return on investment from Google Ads rose by an estimated 56%

Total appointment volume is still lower than last year, due to market conditions. However, relative to total Google search volume, the business is now outperforming both the market and its competitors. There is also a considerable lag between first enquiry and sale, so the full impact is still flowing through. After 4 months of working together, the business has just reported one of its most profitable months of the year relative to Google Ads spend.

Calculate Marketing has since worked with the business’s developer to improve measurement and tracking and fix attribution issues, advised on new website copy positioning and psychology, and provided search volume trend data so that the business has a better informed and more up-to-date understanding of what’s happening in its market. This allows the business to share these insights internally, and to respond to market changes as they happen.

If you’ve switched on Performance Max or AI Max, and your costs have gone up while the quality of your leads has gone down, get in touch today for a free analysis and review.