Google tightens CPA targets, Bing backfills: the July 3 brief

Google is about to make cheap-looking campaigns act closer to the price target you gave them. That can lift your cost per acquisition (CPA, what you pay to win one customer or lead) without a broken ad. Today is mostly about measurement: bidding targets, Performance Max diagnostics, AI reports, and whether customers still see your name.

Google tightens the budget-limited bid math

Google Ads told advertisers on July 2 that, starting August 17, 2026, campaigns marked "limited by budget" will perform closer to their target on Target CPA and Target ROAS bidding, according to Search Engine Roundtable. Target ROAS means how many dollars of sales you want for each dollar of ads.

The example is the whole money story. If a campaign has a $10 Target CPA but has recently been producing $5 CPAs, Google says it can move closer to the $10 target after the change if nothing is updated. A campaign that looked unusually efficient can get permission to spend up to the higher price you already typed into the account.

That hits CAC, your cost to acquire a customer. Budget-limited campaigns often look like bargains because the cap starves them of traffic and the machine serves only the best chances. Raise the budget and the bargain disappears. Predictable at the wrong target is still expensive.

This is the same budget-control lesson from our Google Ads waste audit: the machine can optimize only against the number you give it. A stale target is an open tab.

Performance Max gets a diagnostic window

Google Ads is testing a Channel Diagnostics feature for Performance Max inside "Channel Performance," Search Engine Roundtable reported on July 2. Performance Max is Google's all-in-one automated campaign that can place ads across Search, YouTube, Display and Gmail from one budget.

The feature is not fully live for every account. The reported screenshots show missing and rejected assets by channel. That matters because Performance Max failure is often boring: the campaign cannot use what it was given, so budget routes around the weak spot.

The money angle is waste detection. If YouTube has no useful video, Display has rejected assets, or Shopping lacks clean product data, the campaign can still spend. It just spends with fewer ways to win. Diagnostics do not give you channel budgets, but they give you a cleaner reason why one part is underperforming.

If the panel appears, use it as a repair list, not a victory report. Fix missing assets before raising spend. Then check the landing page, because the ad only buys the visit. Our landing page builder breakdown covers the post-click leak.

Bing's AI reports got a June 1 data jump

Bing Webmaster Tools showed larger AI Performance reporting numbers starting June 1, and Microsoft described the shift as regular data backfilling, according to Search Engine Roundtable on July 2. AI Performance reports show when your pages are cited in AI-generated answers.

This moves attribution, not demand. A reporting backfill can make AI visibility look like it suddenly surged, when customer behavior may not have changed. Owners see a chart jump and start writing a story around it. That story can waste budget if it turns into spend before the trend is real.

The useful read is direction over time. Treat June 1 as a reporting break. Compare week against week after the backfill settles, then connect citations to actual visits, leads and revenue.

That is the broader point from our AI Overviews briefing: AI visibility is becoming part of lead generation, but it is not the same as traffic. A citation is a chance to be considered. The sale still needs a click, a call, or a remembered name.

AI visibility keeps separating from rankings

MarTech argued on July 2 that strong Google rankings no longer guarantee visibility inside AI answers, citing research that only about 16.5% of AI Overview sources also rank in Google's organic top 10. Moz found 88% of AI Mode citations came from pages outside the organic top 10 in its 40,000-query analysis. The full piece is here.

The business consequence is pipeline risk. If buyers ask an AI assistant who to consider and your business is absent, your website rank may matter less for that buying moment.

Do not turn this into vague brand spending. Track the buying questions your customers ask, then check whether your name appears in AI answers, in search results, and on the first page for your own name. We covered that defensive layer in branded search. Recognition is not fluffy when it lowers the paid share of your pipeline.

What to watch

Google's Bid Target Adjustment Tool is expected July 6. The cost risk is stale CPA and ROAS targets being treated as permission to spend.

Performance Max diagnostics are still a test. If Google widens access, the fastest savings will come from repairing rejected assets before budget increases.

AI citation reporting is young. Treat every new dashboard as directional until it connects to leads, calls or sales. More paid-media shifts live in the paid ads hub.