The first time you see Google’s “Have AI check pricing” feature work, it is impressive. A consumer describes a service, provides a few details, and lets Google call multiple businesses to collect pricing and availability. Google then consolidates the answers and emails the consumer a summary. From the consumer’s point of view, the immediate reaction is easy to understand: Wow. Google just saved me from making five phone calls. But after the initial “wow” wore off, I started thinking about what was happening on the other side of those calls. Every business Google contacts must dedicate resources to answering questions, understanding the request, calculating pricing and explaining the service. A salesperson, customer-service representative, dispatcher or call-center agent is doing real work. But is the business speaking with a real lead? Not exactly. Google introduced AI-powered calling as an agentic feature in Search. When it appears for an eligible search, the user can select “Have AI check pricing,” describe what they need and submit the request. Google then calls multiple businesses to ask about pricing and availability. According to Google, the information is returned to the consumer by email. Google also acknowledges that AI-generated responses can contain mistakes. Most importantly, Google’s current documentation says the process does not create a booking. After receiving the results, the consumer must contact the selected provider separately. That distinction matters. Google describes the feature as creating new opportunities for businesses to book customers. However, the initial AI call does not give the business a normal opportunity to sell, qualify or follow up with the consumer. The business supplies the information. Google controls the relationship. As a digital marketer, one of my first questions is how these calls will be classified and measured. The process begins with a Google search, so will these calls appear as organic calls? Will they be included in Google Business Profile call reporting? Will call-tracking platforms identify them as Google organic leads? If they are counted as calls but cannot convert during the interaction, reporting could become misleading very quickly. Imagine that a local business previously received 100 Google-generated phone calls and converted 30 of them. Now it receives 130 calls, but 30 are Google AI research calls with no consumer on the line and no direct way to follow up. Call volume increased by 30%, but the business still closed 30 customers. On paper, the conversion rate fell from 30% to 23%. Did the sales team suddenly get worse? No. Google changed what qualifies as a phone call. These interactions need their own classification. They are not traditional leads, and they should not be mixed with calls from consumers who are ready to speak with a business. AI research works well when it collects structured information that already exists online. Store hours, addresses, service areas, product specifications and published starting prices are all reasonable examples. An AI system can retrieve that information without requiring an employee to stop working and answer another phone call. Verbal pricing is different, especially for complex services. In the video I recorded, I asked Google about a full-service, three-bedroom move between two ZIP codes. Google found information online, but it also indicated that live pricing and availability were being confirmed by phone. A moving quote is not a simple commodity price. How much furniture is involved? Are there stairs, elevators or long carries? Is packing required? Are specialty items involved? How many crew members are needed? What day is the move? How much driving time is involved? What equipment will be required? The same problem exists in home services, automotive repair, healthcare, legal services and many other local industries. A plumber may have an hourly rate, but that rate does not tell the consumer what a job will cost. An automotive shop cannot accurately quote every repair without inspecting the vehicle. A mover cannot responsibly price a complicated relocation based only on the number of bedrooms. When an AI agent requests one number, the conversation can lose the context that makes the number meaningful. Google explicitly warns users that AI responses may include mistakes. That disclaimer may protect the platform, but it does not solve the problem for the business or the consumer. What happens if a salesperson tells the AI that rates “start at $150 per hour,” but Google summarizes the answer as “the rate is $150 per hour”? What if the price applies only to a two-person crew, a weekday appointment or a limited service area? What if the AI combines a price found on the website with information collected during the call and produces a quote that the company never actually authorized? By the time the consumer contacts the business, the AI-generated number may already be treated as a promise. Now the salesperson is no longer explaining the correct price. The salesperson is battling an expectation established by Google. Maybe the actual rate should be $185 per hour based on the crew, schedule, equipment or scope of work. The customer may hear that as a price increase even though the original $150 figure was never appropriate for the job. The AI summary becomes the anchor, and the business is placed in the position of defending its own pricing. There is also a major difference between making research easier and removing nearly all of its friction. Traditionally, a consumer who calls five businesses invests some effort in the process. That effort signals at least a moderate level of intent. With an AI agent, a user can send pricing requests to multiple businesses with a few clicks. The consumer may be seriously shopping-or may simply be curious. Once this feature becomes familiar, what prevents a person from repeatedly sending agents to research prices without any near-term intention of buying? What prevents competitors from using automated research to monitor rates? What prevents AI systems from repeatedly contacting the same businesses as they try to refresh their data? One consumer can now create work for several companies without speaking to any of them. The consumer’s effort approaches zero. The businesses’ combined effort does not. Large companies may be able to develop special routing, scripts and reporting for AI-generated calls. They can train call-center employees to recognize the calls and provide controlled responses. A small business may have the owner answering the phone. That owner could be on a job, helping a customer or trying to close a legitimate lead. An AI pricing call consumes the same time as a real inquiry but offers much less opportunity to convert the interaction. Businesses are effectively being asked to contribute labor and proprietary pricing information to improve Google’s consumer experience. They receive no guarantee that the consumer will contact them. They may not receive usable consumer information. They may not know how Google summarized the conversation. They may not even know whether the information they provided influenced the recommendation. The platform receives the data. The consumer receives the convenience. The business absorbs much of the cost. Today, Google says this feature gathers information but does not complete the booking. What happens tomorrow? Will the AI select the provider? Will it schedule the appointment? Will it agree to a price? Will it provide a deposit or accept contract terms? Are we moving toward a process in which an AI agent books a complicated service and everyone hopes the details work out when a real person finally arrives? Automation can remove friction, but some friction is useful. A short conversation can uncover problems, set expectations and prevent misunderstandings. It gives a trained sales professional the opportunity to understand what the customer actually needs. Removing every human interaction does not automatically improve the transaction. Sometimes it merely delays the human interaction until there is a disagreement. I do not think businesses can ignore this development. They need to prepare for it while Google determines how far agentic calling will expand. Businesses should: I understand why consumers will use this feature. The first experience feels almost magical. But innovation should not be measured only by how much effort it saves the person pressing the button. We also need to measure how much effort it creates for everyone on the other side. Google needs clearer attribution, controls against excessive requests, better qualification of consumer intent and a way for businesses to review exactly how their information was summarized. Most importantly, AI research calls should not be treated as normal leads. The future of search may be agentic, but businesses should not become free data providers for AI systems with no visibility into the customer, the attribution or the final outcome.
How Google’s AI Pricing Calls Work
Is This an Organic Lead That Can Never Convert?
original Google-generated calls
calls after 30 AI research calls
apparent conversion rate
Service Pricing Is Not Always a Simple Number
What Happens When the AI Gets the Price Wrong?
Consumers Can Research Without Showing Real Intent
Small Businesses and Call Centers Will Carry the Cost
What Comes Next?
What Businesses Should Do Now
AI Convenience Needs Business-Side Guardrails
Joe Laratro is CEO and founder of Tandem Interactive in south florida.