A missed phone call can cost a small business a customer before anyone on its team even knows there was a sales opportunity. That problem is becoming a commercial entry point for voice AI, but a new report suggests that making automated conversations sound human is only part of the equation.
According to The State of Voice AI for Small Businesses 2026, published by AI answering service Upfirst, 78% of surveyed US consumers said they had hired or purchased from a competitor because a business failed to answer or return their call. Meanwhile, 80% said they would want an AI receptionist to identify itself as artificial intelligence at the beginning of a conversation.
The findings point to two distinct challenges for companies developing voice agents: capturing demand that would otherwise be lost and earning enough customer confidence to handle the interaction effectively.
The report draws on surveys of 270 US consumers and 262 small and midsize business owners and managers, alongside an analysis of 677,562 calls handled for 2,967 US businesses between January 2025 and August 2026.
The business case starts with unanswered calls
For many small businesses, answering every phone call is an operational problem rather than a question of customer-service strategy. Employees are occupied with existing customers, owners are managing daily operations, and hiring a dedicated receptionist may not make financial sense.
The data suggests that the gap is particularly visible among companies with 10 to 49 employees. Upfirst found that 91% of businesses in this group missed at least one potential-customer call during a typical week, while 76% missed four or more. Nearly eight in ten said they had already lost a customer because of an unanswered call.
The pattern helps explain why voice AI is attracting interest beyond companies looking to reduce staffing costs. Upfirst’s survey found that 77% of businesses missing at least four calls a week were open to using an AI receptionist, compared with 32% of businesses that reported missing none.
Adoption also varied by company size. Only 2.4% of solo operators reported using an AI receptionist, compared with 41.5% of businesses with at least 50 employees.
These figures suggest that demand is tied to the volume and complexity of incoming calls, rather than company size alone. A business with enough customer inquiries to overwhelm its existing staff has a clearer reason to automate call handling than an operator who can answer every call personally.
Cost is also becoming less of a barrier. Upfirst estimates that the cost of running a voice agent fell roughly threefold to fivefold over the preceding year, as providers including OpenAI, Google and xAI introduced voice capabilities or reduced prices. Lower operating costs could make automated reception more accessible to businesses that previously could not justify dedicated phone coverage.
The commercial opportunity, however, depends on what happens after the call is answered.
Customers will accept AI for some tasks, not all of them
The report found that 43% of consumers were comfortable with a small business using an AI receptionist, while 29% were uncomfortable. The remaining respondents were neutral, meaning most did not express outright opposition to the technology.
But acceptance varied considerably depending on the task.
Seventy percent of respondents said they would be comfortable letting an AI answer questions about business hours, services and pricing. Another 62% would allow it to take a message. The figure fell to 45% for booking or changing appointments, and to 16% for taking payments. Only 15% were comfortable with an AI handling a complaint.
For businesses deploying voice agents, these differences matter more than whether a model can sustain a convincing conversation. Answering a question about opening hours requires a different level of judgment and accountability from resolving a billing dispute or handling a dissatisfied customer.
The findings also challenge the assumption that voice AI must replace human interaction to deliver value. Among businesses already using AI receptionists, 79% reported having human coverage for calls outside business hours, compared with 48% of businesses that were not interested in the technology.
Upfirst’s call records offer another indication of how the systems are being used: 16.4% of calls included an attempt to transfer the caller to a human. Rather than completing every interaction independently, voice agents can act as an initial point of contact, collecting information and routing callers to employees when needed.
That distinction has implications for how startups build and sell these products. Systems that answer routine questions, capture leads and recognize when a conversation requires human intervention may offer a more practical proposition than platforms promising to automate every customer interaction.
Transparency could matter more than a human-sounding voice
The most striking finding concerns disclosure. Four in five consumers surveyed by Upfirst wanted an AI receptionist to identify itself at the start of a call. Among respondents who were neutral or uncomfortable with the technology, 55% identified disclosure as a factor affecting trust, while only 19% pointed to a natural-sounding voice.
Upfirst also cited earlier research covering 450,702 calls across 503 AI receptionists. Systems that disclosed their AI identity were associated with roughly 20% fewer hang-ups. The company cautioned that the analysis was observational and does not establish that disclosure alone caused the difference.
Still, the results offer a useful signal for an industry that has invested heavily in making synthetic speech sound increasingly natural. Customers may not need to believe they are talking to a person. They need to understand who is answering, receive useful information and have a clear way to reach someone who can resolve their problem.
The findings should also be read within the limits of the research. The consumer and business surveys used opt-in online panels rather than probability samples, and the consumer sample skewed toward higher-income households, which Upfirst identified as a factor that may inflate acceptance estimates. The call records, meanwhile, cover businesses using its answering service and cannot establish missed-call rates across the wider US economy.
Even with those caveats, the report points to a practical test for voice AI companies: whether their products help businesses recover opportunities without introducing new friction into customer interactions.
For startups in the sector, the next stage of competition may depend less on producing the most realistic voice and more on building systems that know what they can handle, when to disclose their identity and when a human needs to take over.
