Google this week published a new help document introducing AI-qualified call conversions, a feature that changes how call campaign performance is measured inside Google Ads. The announcement, dated April 21, 2026, was first spotted by Hana Kobzova and reported by PPC News Feed before being confirmed in a dedicated Google Ads Help Center page. The change affects how phone calls are counted as conversions and, by extension, how Smart Bidding algorithms receive and process call signal data.
For years, Google Ads measured a phone call as a conversion primarily by its duration. If a call exceeded a minimum number of seconds set by the advertiser, it counted. Simple enough - but the logic contained a structural flaw. A robocall, a misdial, or a spam interaction could last long enough to meet the threshold and be counted as a lead. The new system sets out to address that directly.
What AI-qualified call leads actually do
According to the Google Ads Help Center documentation, AI-qualified call leads use Google AI to evaluate call recordings and determine the actual quality of the conversation. The system looks for signals of intent - "such as a customer inquiring about specific services, scheduling a consultation, or showing readiness to purchase," according to the documentation. Duration alone, Google states, does not always reflect lead quality; a long call could still be a wrong number or a robocall.
The practical mechanics work through a tiered classification approach. When call recording is active, all calls are recorded and evaluated by AI. Only qualified calls count as conversions. If a call cannot be recorded for any reason, the system falls back to the secondary signal: call duration. And if a Google forwarding number is not available - an edge case - conversions are classified based on ad interaction data alone. The hierarchy is designed so that the system never misses a conversion signal entirely, even when technical conditions prevent the primary method from operating.
This is a meaningful departure from the previous single-metric model. The old approach treated call duration as a proxy for quality because it was the only scalable signal available. AI classification introduces a qualitative layer that duration could never provide.
Call recording is now on by default
One of the most operationally significant aspects of this update is the default-on status of call recording. According to Google's documentation, call recording is now turned on by default for most users to immediately provide enhanced insights and AI-qualified call leads. Advertisers who previously disabled call recording will retain their off setting. Similarly, businesses operating in healthcare or financial services will not have call recording turned on automatically - Google identifies these sectors as exceptions and excludes them from the default.
For everyone else, the implication is straightforward: call recordings are being made, and those recordings are being analyzed by AI to determine whether each call constitutes a qualified lead. Advertisers who object to this can disable call recording at any time through the Admin menu, under Account settings, then Call ads, then Call recording, selecting Off and saving. The process is reversible.
The geographic scope of the feature is currently limited. Call recording is available only when both the dialing and receiving phone numbers are located in the United States or Canada. That restriction is not incidental - call recording laws vary significantly by jurisdiction, and Google appears to have scoped the initial rollout to markets where its legal framework is established.
Transparency tools: summaries and hashtags
The feature introduces a transparency layer that is worth examining in detail. Call Details reports inside Google Ads will now include a concise AI-generated summary of each call, alongside specific hashtags. The documentation gives two examples: "#HighIntent" and "#ConsultationScheduled." These tags are generated automatically based on the AI's assessment of the conversation.
This reporting change matters because it gives advertisers something they have historically lacked: a structured, at-a-glance explanation for why a call was or was not counted as a conversion. Previously, when a call met the duration threshold, it was a conversion. There was no mechanism to review what had actually been said. The new system surfaces that information directly within the reporting interface, making the AI's reasoning visible rather than treating it as a black box.
The hashtag system in particular reflects a design choice to make categorization legible without requiring advertisers to listen to every recording. A tag like "#ConsultationScheduled" communicates the outcome of a call in a way that is immediately useful for lead quality analysis. Whether the taxonomy of tags is comprehensive, and whether the AI classification is consistently accurate, are questions that only emerge at scale over time.
How settings are adjusted
Advertisers who want to modify the call duration threshold - the secondary signal that applies when recording is unavailable - can do so through the Goals menu, then Summary, then Phone call leads, then AI-qualified call leads, then Edit, then Lead qualification, entering a minimum length in seconds, and selecting Done. This threshold still matters because it governs how the system handles calls that fall outside the primary recording pathway.
The interaction between these settings has practical consequences for lead volume reporting. Raising the duration threshold makes the secondary signal more selective. Lowering it makes it more permissive. Given that the primary AI-based classification is the preferred signal, the threshold setting functions more as a safety net than as the main control mechanism - but it remains adjustable.
Why this matters for Smart Bidding
The implications for automated bidding are direct. Smart Bidding relies on conversion signals to optimize bids at auction time. If the conversion signal contains calls that were, in practice, spam or misrouted, the bidding algorithm receives corrupted input. It cannot distinguish between a genuine consultation call and a robocall that happened to last 45 seconds. The result is that bid optimization is partly guided by noise.
AI-qualified call conversions filter that noise before it enters the bidding system. According to the documentation, by feeding higher-quality conversion data into Smart Bidding, Google Ads can optimize spend to find more customers who are likely to convert. PPC Land has documented how Smart Bidding quality depends fundamentally on the quality of conversion data it receives, a principle that applies here with particular force: the garbage-in, garbage-out dynamic in call conversion tracking has been a known problem, and AI classification is the structural fix Google has chosen.
The connection to lead generation optimization is also relevant. Google's guidance on value-based bidding for lead generation has consistently emphasized the importance of high-quality conversion actions. Counting robocalls and spam as conversions has been the antithesis of that guidance.
Context: a pattern of measurement improvements
This update does not arrive in isolation. Google has spent the past 18 months systematically tightening its measurement infrastructure across multiple fronts. Enhanced conversions for web and leads are being merged into a single unified toggle starting June 2026, announced April 10, 2026. Google Analytics introduced cross-channel budgeting, enhanced conversion management, and attribution analysis in January 2026. Each of these changes shares a common direction: making conversion data more reliable so that automated optimization systems have better inputs to work with.
Calls from location extensions can already be counted as conversions, with the default conversion window set at 60 seconds - a metric that predates AI qualification. AI-qualified call conversions effectively extend that measurement logic into a qualitative dimension, adding content analysis alongside interaction-based tracking.
The data usage terms governing this feature are also notable. According to Google's documentation, recordings are used to evaluate lead quality, monitor spam and fraud, and improve the accuracy of conversion reporting. Use of the call recording feature is governed by the Call Ads Supplemental Terms. Advertisers operating call campaigns should review those terms, particularly given the default-on status of recording for most accounts.
What it means for lead generation advertisers
Lead generation campaigns built around phone calls represent a substantial share of Google Ads spend in sectors like legal services, home improvement, healthcare-adjacent services, financial planning, and real estate. In these categories, call quality - not call volume - determines whether a campaign is profitable. A law firm receiving 100 calls per month where 40 are spam or wrong numbers is not measuring the same thing as a law firm receiving 60 genuine consultations.
The introduction of AI-qualified call conversions is a structural acknowledgment that call duration was always an imperfect proxy. The signal was adopted because it was the only machine-readable indicator of call quality available at scale. AI evaluation of call content is a fundamentally different kind of signal - one that requires the infrastructure of call recording and language model analysis, but that has the potential to align reported conversions much more closely with actual business outcomes.
The rollout is currently limited to the United States and Canada. Advertisers in other markets continue to rely on call duration as the primary conversion signal. Whether and when the feature expands geographically depends on the regulatory framework in each market.
Data, privacy, and the recording default
The decision to enable call recording by default raises practical questions that go beyond feature settings. Call recording intersects with consent law, customer expectations, and sector-specific compliance requirements. Google's choice to exclude healthcare and financial services from the default - rather than requiring all advertisers to opt in - reflects a calibrated approach to sectors with stricter data handling obligations. But many other industries serving sensitive personal situations, such as legal services, addiction treatment, or family law, fall outside those two carved-out categories. Advertisers in those areas should assess whether the default-on recording setting aligns with their own compliance posture and customer communications.
The data usage terms state that recordings are used to evaluate lead quality, monitor spam and fraud, and improve the accuracy of conversion reporting. The Call Ads Supplemental Terms govern the feature. Advertisers should note that recordings processed to evaluate lead quality are being handled by Google's AI systems, and the call summaries and hashtags generated from those recordings are surfaced back to the advertiser in reporting. This creates a chain of data handling that spans the call itself, the recording, the AI evaluation, and the reporting output - each step governed by the terms advertisers accept when using call assets in Google Ads.
For the marketing community, the shift surfaces a question about historical data comparability. Advertisers who have been running call campaigns will see their conversion reporting change as AI classification replaces or supplements duration-based counting. A period of metric discontinuity is likely during the transition, and campaign managers should account for that when evaluating performance trends.
Timeline
- 2019: Calls from location extensions introduced as countable conversions, with 60-second default duration threshold
- August 2024: Google publishes five-part video series on value-based bidding for lead generation, emphasizing conversion quality
- February 2026: Google ends call-only ads, shifting advertisers to responsive search ads with call assets
- March 11, 2026: Google Ads product managers publish Smart Bidding best practices for 2026, emphasizing conversion data quality as bidding fuel
- April 10, 2026: Google announces enhanced conversions for web and leads will merge into a single unified toggle starting June 2026
- April 21, 2026: Google publishes Help Center documentation introducing AI-qualified call conversions; call recording enabled by default for most US and Canada accounts; feature spotted by Hana Kobzova via PPC News Feed
- April 22, 2026: Search Engine Roundtable covers Google's new AI-qualified call conversion documentation
Summary
Who: Google, through its Google Ads platform, with initial coverage credited to Hana Kobzova of PPC News Feed and subsequently reported by Search Engine Roundtable's Barry Schwartz.
What: Google has introduced AI-qualified call conversions, a new feature that replaces call duration as the primary conversion signal with AI analysis of call recordings. The system uses a tiered classification approach - call recording first, duration as fallback, ad interaction as a last resort - and surfaces AI-generated summaries and hashtags in Call Details reports. Call recording is now enabled by default for most accounts.
When: The feature was published in the Google Ads Help Center on April 21, 2026.
Where: The feature is currently limited to calls where both the dialing and receiving numbers are located in the United States or Canada. Settings are managed through the Google Ads interface under Goals and Admin menus.
Why: Call duration has historically been an imperfect proxy for lead quality, allowing robocalls, spam calls, and misdials that met the duration threshold to count as conversions. AI classification of call content removes that noise before it enters Smart Bidding, with the goal of producing more accurate conversion reporting and higher-quality optimization signals for automated bidding systems.