Gemini AI Resolves 75% of Google Ads Support Queries: Alphabet

What happened

Alphabet has disclosed that Gemini-powered AI agents now autonomously resolve 75% of customer support queries related to Google Ads. The figure was shared during the company's Q2 2025 earnings call, marking one of the most specific, publicly reported benchmarks yet for AI-driven support resolution at scale.

The disclosure positions Gemini as central to Alphabet's customer-facing operations, with the company framing the automation of Ads support as evidence of the model's practical utility beyond consumer-facing chat applications. Reporting from CX Today notes that Alphabet is treating this as a proof point for broader Gemini deployment across its business lines.

Why it matters

A 75% autonomous resolution rate, if sustained and independently verified over time, is a meaningful data point for any organisation weighing how far generative AI can realistically shoulder first-line support. Ads support queries tend to involve account-specific, rules-based troubleshooting — exactly the kind of structured, high-volume interaction where AI resolution is most tractable, and a useful contrast to messier, emotionally charged service categories.

For CX and service-design leaders, the number matters less as a headline and more as a template: it shows a large platform operator willing to publish a concrete accountability metric for AI performance, rather than vague claims of "efficiency gains." That transparency — however self-reported — gives other operators a comparable reference point when building their own AI deflection targets and human-escalation thresholds.

By the numbers

• 75% of Google Ads support queries are now resolved autonomously by Gemini-powered AI agents, according to Alphabet's Q2 2025 earnings disclosure.

The Renascence take

The headline number is impressive, but the more interesting question is what happens to the remaining 25% — and whether Alphabet's definition of "resolved" matches what customers would call a good outcome.

A resolution rate is only as trustworthy as its denominator. Ads support is a narrow, rules-heavy category — not a proxy for how AI performs on ambiguous, high-stakes or emotionally loaded issues. Before quoting this figure as a benchmark, operators should ask what counts as "resolved," whether customers were offered an easy path to a human, and how satisfaction — not just deflection — was measured. The real signal here isn't the percentage; it's that a company of Alphabet's scale is now willing to publish a hard number at all. That sets a precedent worth following, but the metric that will actually protect customer trust is escalation quality, not automation volume.