Alphabet's Q2 Earnings Signal Rising Google Ads Costs

The practical read on Alphabet’s Q2 2026 earnings is not “Google Ads is weak.” It is also not “AI made ads cheaper.” For anyone updating a Q3 or Q4 media plan, the cleaner translation is this: Google Search revenue is still growing, click costs are still rising, and the auction is moving further into an AI-managed layer whether account teams are ready or not.

Alphabet reported $63.3 billion in Search ad revenue, up 17% year over year, while missing StreetAccount expectations by roughly $100 million. That miss is small in absolute terms, but CFO Anat Ashkenazi’s warning that “Q3 laps an acceleration in search from last year” matters because many advertisers are already planning into higher CPCs and less room for forecasting error.[1]

The stock market version of the story can wait. The account-level version cannot. If your Q3 forecast still assumes that Google’s automation will offset inflation with better efficiency, the burden of proof has shifted. The benchmark environment says average CPC reached $5.42 in 2026, up 12% year over year, and 87% of industries saw rising costs.[2]

What The Earnings Signal Actually Changes

Alphabet’s total ad revenue reached $81.63 billion in Q2, so this is not a demand-collapse story.[3] The issue for advertisers is that strong platform revenue, rising CPC benchmarks, and heavier automation can all be true at the same time. Google can grow, campaigns can keep converting, and your marginal click can still get more expensive.

That distinction matters for budget pacing. A 12% CPC increase does not mean every account should add 12% to budget. It does mean a flat budget has to earn its way into the same volume with better conversion rates, better query matching, better landing pages, or tighter allocation. If those improvements are not already visible, the forecast should not quietly assume them.

The earnings headline also needs one accounting caveat. Alphabet’s reported GAAP EPS included a $99 billion non-operating equity securities gain tied to holdings including Anthropic and SpaceX, while adjusted EPS was $2.85 versus a $2.89 consensus estimate.[4] That is relevant only because it keeps the operating discussion clean: advertiser planning should not be anchored to a headline earnings number that includes a large non-operating gain.

Do Not Read The CPL Drop As A Cheaper Auction

The most tempting counterpoint is the average cost per lead. WordStream and LocaliQ reported a $66.69 average CPL, the first drop in five years.[2] That is good news for some accounts, but it does not cancel the CPC inflation story. It says that, in aggregate, lead acquisition improved despite more expensive clicks.

A lower CPL can happen when targeting, bidding, and matching improve enough to make each expensive click more likely to convert. That is different from a cheaper auction. The manager who treats the CPL drop as permission to hold budgets flat may discover too late that the account’s lead volume now depends on an efficiency gain that is not guaranteed to repeat.

This is where planning should split into two questions. First, what CPC inflation should the budget absorb before volume starts to fall? Second, what conversion-rate improvement is already proven in the account, not merely available in a benchmark deck? Those questions are uncomfortable, but they are easier to answer in July than after September spend is already committed.

AI Is No Longer A Campaign Type At The Edge

Automation is now part of the auction’s default operating system. Performance Max adoption reached 71% of advertisers and accounts for roughly 62% of all Google ad clicks, according to a 2026 data compilation.[5] AI Max has been out of beta since April 2026, and Dynamic Search Ads campaigns are scheduled to begin auto-migrating from September 2026.[5]

That migration changes the campaign architecture conversation. The old question was whether to test an AI-led campaign beside a more controlled keyword structure. The new question is where control still creates value: exclusions, asset quality, conversion definitions, audience signals, landing page selection, feed hygiene, brand protection, and budget separation.

This is not an argument for fighting automation. Smart Bidding now manages 78% of Google Ads spend, and accounts with 50 or more monthly conversions see 22% lower CPA on average, according to Google data cited in 2026 market summaries.[5] The lesson is narrower: automation needs enough clean conversion volume to learn from, and the account team still has to decide which signals deserve to shape the bidding system.

For small accounts, this becomes a harsher planning problem. Accounts spending under $5,000 per month saw 18% higher CPCs and 31% lower conversion rates than the median in the 2026 benchmark set.[2] That does not mean small advertisers should avoid Google Ads. It means they have less margin for loose match types, vague conversions, bloated geographies, and campaigns that never generate enough signal for the bidding model to stabilize.

Revise Budget Assumptions Before Q3 Pacing Makes The Decision

The first change is forecasting. If the plan assumes year-over-year efficiency gains, make that assumption explicit. Name the mechanism. Is the account expecting better lead quality from PMax? A landing page conversion-rate lift? A budget shift away from expensive non-brand terms? A feed cleanup? If the model only says “AI efficiency,” it is not a forecast; it is a placeholder.

• Use higher CPCs as the base case for Q3 and Q4, especially in categories that already saw auction pressure in the first half of 2026.
• Separate volume targets from efficiency targets so a missed CPA does not automatically become a surprise budget conversation.
• Create a downside scenario where CPC rises but conversion rate does not improve.
• Hold back budget for proven pockets of marginal return instead of pre-committing every dollar to broad automated expansion.
• Review campaign learning periods before major promotional windows; a system that is still relearning in late Q3 is not a neutral risk.

The CFO’s Q3 comparison warning should also change how teams explain performance internally. A weaker YoY growth rate in Q3 may reflect tougher comparisons, not necessarily a broken account.[1] But that caveat should not become cover for poor pacing. The useful version is: comparisons get harder, so the plan needs clearer thresholds for when to cut waste, when to protect volume, and when to ask for more budget.

Campaign Structure Needs Control Points, Not Nostalgia

The account structure that worked when keyword campaigns carried most of the intent may not be the structure that protects margin in an AI-first auction. That does not mean every account should collapse into one automated campaign. It means structure should be designed around decision rights.

If branded search, high-intent non-brand, shopping feeds, remarketing, and prospecting all share the same budget pool, the platform can optimize toward the easiest reported conversions while the business loses visibility into incremental value. If every segment is split too finely, the model may never get enough volume to bid well. The useful middle is not ideological; it is based on where performance decisions genuinely differ.

• Keep brand and non-brand performance readable, even if automated campaigns contribute to both.
• Protect budget for proven high-intent demand before expanding into looser AI-discovered inventory.
• Audit search terms, placement signals, asset groups, and product-level performance where the interface still exposes useful controls.
• Treat the September 2026 DSA migration as a reason to review landing page coverage and exclusions before the migration begins.
• Avoid making major bid strategy, conversion, and budget changes at the same time unless the account has enough volume to recover quickly.

AI Overviews add another pressure point. 2026 statistics roundups report that AI Overviews appear on 47% of search queries and reduce organic click-through rates by 40% to 58%.[6] The exact effect will vary by query set, but the direction matters: if less organic traffic reaches the site, more teams may try to recover demand through paid placements, which can increase competition in auctions that were already more expensive.

Device Data Deserves More Than A Dashboard Glance

One of the more actionable benchmark findings is device imbalance. Mobile generated 65% of clicks but only 47% of conversions in the 2026 benchmark data.[2] That does not justify a blanket mobile cut. It does justify checking whether mobile is absorbing spend because it is available, not because it is equally valuable.

For lead generation accounts, this review should go beyond device-level CPA. Look at form completion rates, call quality, page speed, post-lead qualification, and downstream revenue by device where the data exists. Mobile may be creating more early-stage leads and fewer qualified opportunities. Or it may be under-credited because calls and cross-device conversions are measured poorly. Either way, the account should not let an aggregate automated bid strategy hide a device mix problem.

Measurement Is Where The AI Story Gets Expensive

Google’s AI performance claims are useful, but they are not the same thing as incrementality. On the earnings call, Google pointed to AI ad tools producing 50% more conversions at similar ROAS.[7] That may be directionally helpful for in-platform optimization. It does not prove that every reported conversion was newly created by the spend.

Independent measurement often lands in a different place because it asks a different question. Cassandra’s MMM benchmark, based on 253 models, 59 advertisers, and $383 million in media spend, found that platform ROAS typically overstates incremental value by 2x to 5x versus independent MMM; in that benchmark, PMax incremental ROI had a median of 4.64x versus 5.21x for Google Search non-brand.[8]

That does not make platform ROAS useless. It makes it a steering metric, not a boardroom answer. Platform attribution helps the bidding system decide which auctions resemble past converters. Incrementality work helps the business decide whether the spend created value that would not have happened anyway. Confusing those two jobs is how an account can look efficient while the media plan becomes more expensive.

The same caution applies to AI Max. Google has reported a 14% average conversion lift, while an independent 2026 advertiser survey from ALM Corp found that 84% of advertisers reported neutral or negative results in testing.[9] Those findings do not have to be reconciled into one universal verdict. They are a reminder that rollout timing, conversion quality, account volume, and measurement design can decide whether an AI feature helps or simply changes where credit appears.

The Stock Forecast Angle Only Matters As Operating Pressure

For advertisers, the stock forecast angle is useful mostly as a proxy for pressure, not as a trading input. Alphabet’s AI infrastructure spend is large: Q2 capex was reported at $44.9 billion, up 100% year over year, with full-year capex guidance of $195 billion to $205 billion.[10] It would be too strong to say that this directly causes higher CPCs. The ad auction is not a simple cost-plus pricing model.

The safer conclusion is directional. Google is investing heavily in AI infrastructure, investors expect that investment to monetize, and advertising remains the company’s core economic engine. In that environment, advertisers should not expect the platform’s default path to optimize for lower media costs at the expense of Google revenue. The system can improve matching and still participate in a more expensive auction.

A Q3-Q4 Operating Posture

The right response is not to pull back from Google Ads by default. Search demand is still valuable. High-intent clicks still matter. AI bidding can outperform manual management when the account has enough conversion volume, clean goals, and a measurement setup that rewards real business outcomes.

The wrong response is to treat Alphabet’s revenue growth and Google’s AI claims as a guarantee that your account will absorb CPC inflation without trade-offs. Q3 planning should assume higher click costs, test automation where signal quality supports it, preserve control where business value differs from platform-reported conversion value, and revise forecasts before the auction does the revision for you.

References

1. Alphabet Q2 2026 earnings live updates and earnings call transcript, CNBC, July 22, 2026.
2. 2026 Google Ads Benchmarks, WordStream/LocaliQ, 2026.
3. Alphabet Q2 2026 earnings summary, 9to5Google, 2026.
4. Alphabet Q2 2026 adjusted EPS and equity securities gain analysis, CNBC, 2026.
5. 2026 Google Ads AI automation statistics compilation, Hooked Marketing, 2026.
6. 2026 AI Overviews and search CTR statistics roundup, Digital Applied and Hooked Marketing, 2026.
7. Alphabet Q2 2026 earnings call commentary on AI ad tools, Google, July 22, 2026.
8. Cassandra MMM benchmark, Cassandra, 2026.
9. 2026 AI Max advertiser survey and Google Ads Blog AI Max results, ALM Corp and Google Ads Blog, 2026.
10. Alphabet Q2 2026 capex and free cash flow analysis, TradingKey and CNBC, 2026.