Calls referred by ChatGPT convert into sales leads at a 49 percent rate, the highest of any marketing channel measured and roughly 10 percentage points above the cross-channel average, according to a benchmarks report Invoca published today drawing on more than 70 million phone conversations.

The finding sits inside the Invoca Lead Conversion Benchmarks Report 2026, a study built from anonymized call data tracked across the company's platform. It marks, according to Invoca, the first year the company had enough measurable call volume from a large language model to include generative AI as its own referral category alongside paid search, organic search, Google Business Profiles, paid social and other established channels. Conversion rates from those same ChatGPT-referred calls came in close to average, at 40 percent, about 3 percentage points below the top-performing channels in that specific metric. The company frames the combination as evidence that people who pick up the phone after researching with ChatGPT arrive further along in a buying decision, even if the raw number of such calls remains small next to paid search or local listings.

What the report measures and how

Invoca built the report from more than 70 million calls and over 600 million minutes of tracked phone conversation, spanning 10 industries and seven marketing channels. The company describes all figures as averages pulled from its customer base rather than a market-wide census, and it says the calculator and benchmark tables accompanying the report are illustrative estimates rather than guarantees tied to any single business. Because the underlying calls run through Invoca's own call tracking and analytics infrastructure, the dataset captures a slice of commercial phone traffic large enough to support cross-industry comparison, though it remains bounded by whichever companies use Invoca to tag and route their calls.

Three benchmarks anchor the study before it narrows to the generative AI finding. Across all industries, 56 percent of calls to businesses get answered by a person; that share climbs to 65 percent when isolating calls lasting longer than 15 seconds, and to 71 percent for calls over 30 seconds, filtering out misdials and immediate hangups. Of the calls that are answered, 38 percent qualify as leads. Of those leads, 42 percent convert on the call itself. Invoca also reports that 64 percent of businesses fail to ask callers to buy or book an appointment at all, a gap the report treats as a coaching opportunity separate from any channel-level question.

Where ChatGPT fits against other channels

Paid search continues to drive the highest volume of calls, leads and conversions of any paid channel in the dataset, and Google Business Profiles play the equivalent role among organic sources for multi-location businesses. Invoca is explicit that channel efficiency and channel scale are separate questions, and that percentages alone do not capture which channel actually delivers the most business. Within that context, ChatGPT's 49 percent lead rate outperforms Google Business Profiles, the next-closest channel, by 6 percentage points; Google Business Profiles posted a 43 percent lead rate in the same comparison. Every other channel in the study falls further behind on this specific measure.

The report's own caveat carries weight here. Invoca states plainly that call volume attributable to generative AI referral remains very low across its dataset, and that ChatGPT is currently the only large language model producing measurable call volume at all. No comparable figures exist yet for Gemini, Claude, Perplexity or other assistants, not because their referral quality differs, but because the underlying call counts have not reached a size Invoca considers reportable. The company's own conclusion, stated in the methodology section of the report, is to treat the ChatGPT figure as an early signal worth tracking rather than a channel that currently justifies a budget shift.

Context from prior AI referral research

The Invoca figures land inside an already active body of research on how AI-referred web traffic converts compared with traditional channels, though nearly all of that prior work has measured clicks, sign-ups and web sessions rather than phone calls. <cite index="20-1">Ahrefs research published in June 2025 found that visitors from AI-powered search platforms convert at rates 23 times higher than conventional search engine visits</cite>, even though <cite index="20-1">AI search visitors accounted for just 0.5 percent of total website visits during the measurement period</cite>. Microsoft Clarity extended that line of inquiry across a much larger sample. <cite index="19-1">Its analysis of more than 1,200 publisher and news websites found AI referral traffic expanding 155 percent over an eight-month period, while AI-sourced visitors converted to sign-ups at 1.66 percent compared with 0.15 percent from organic search, an 11-fold advantage</cite>, even as <cite index="19-1">AI-driven referrals accounted for less than 1 percent of total website visitors across the analyzed dataset</cite>.

Not every study in this space points the same direction, and the divergence matters for how the Invoca figures should be read. A separate examination cited in that same Microsoft Clarity coverage found <cite index="19-1">that ChatGPT traffic underperformed Google in an e-commerce study examining 973 websites with 20 billion dollars in combined revenue, with ChatGPT referrals showing conversion rates inferior to every channel except paid social media</cite>. That result suggests the AI-referral conversion advantage is not universal; it appears concentrated in certain content types and purchase categories rather than holding uniformly across e-commerce.

Criteo has published the most detailed commercial figures on this pattern to date, tracking conversion performance for its ChatGPT Ads integration with OpenAI since March 2026. <cite index="15-1">When Criteo first disclosed the integration, it reported that users referred from large language model platforms were converting at approximately 1.5 times the rate of other referral channels, based on a sample of 500 US retailers observed in February 2026</cite>. That figure has moved since. <cite index="15-1">By May 2026, specific retail categories, including consumer electronics, lifestyle and wellbeing, and home and garden, were seeing conversion rates approaching twice those of traditional search</cite>, and <cite index="14-1">by June 2026, more than 80 percent of ad-driven ChatGPT Ads traffic was reported to come from new customers rather than returning shoppers</cite>.

What none of that prior research addressed is the phone. Web analytics platforms including Google Analytics track referral sources for site visits, downloads and form submissions, but a call placed after someone closes a browser tab and dials a number leaves no clickstream for those tools to follow. That gap has structural roots. <cite index="22-1">ChatGPT sends approximately 190 times less traffic to websites than Google, despite handling roughly 12 percent of Google's query volume</cite>, and <cite index="22-1">the platform focuses on providing complete answers within its own interface rather than directing users elsewhere</cite>, a design choice that narrows the pool of trackable referral clicks before a phone call ever enters the picture. Invoca's report addresses that specific blind spot by measuring the call itself as the unit of analysis, independent of whether a website visit preceded it.

The measurement problem this data addresses

Marketing teams already know that a meaningful share of AI-influenced commercial activity never appears as a clean referral in standard analytics. <cite index="18-1">A Similarweb study found that 55.9 percent of AI-influenced website visits arrive as branded organic search rather than through a trackable referral link</cite>, which means most of the value generated by an AI recommendation stays invisible to any measurement approach that only counts direct clicks from known assistant domains. Phone calls compound that same problem rather than solving it. A caller who researched a purchase inside ChatGPT, closed the app, and later typed a business's name into Google or simply remembered the number would show up in a company's phone system with no digital trail connecting the call back to the AI conversation that prompted it, unless that business runs the kind of call tracking infrastructure Invoca sells.

Platform-level tools have started to close part of that gap on the web side. <cite index="16-1">Google Analytics added a native AI Assistant channel to its Default Channel Group on May 13, 2026, automatically classifying traffic from ChatGPT, Gemini, Claude and other AI assistants without requiring manual configuration</cite>. That update addresses attribution for web sessions. It does nothing for a phone call, since Google Analytics has no visibility into what happens after someone picks up a handset. Google has moved in a parallel direction for phone data specifically: Google Analytics added a direct integration with Google Business Profiles on June 8, 2026, pulling seven local interaction metrics, including phone calls placed through a Business Profile listing, into centralized Analytics reporting. That integration connects call volume to a business's organic local presence, but it does not distinguish a call driven by a ChatGPT recommendation from one driven by any other discovery path.

On the paid media side, Google has been consolidating offline conversion data, including phone calls that close into sales, through a single pipeline. Google shifted new offline conversion imports to its Data Manager API starting June 15, 2026, a change the company frames as unifying how phone-based and other outside-the-browser conversions flow back into its ad platforms. The infrastructure exists for advertisers to report that a call led to a sale. What remains unresolved, both in Google's own tooling and in the Invoca report itself, is the referral source that started that chain when the first touchpoint was a conversation inside an AI assistant rather than a search query or an ad click.

Why the volume caveat matters

The 49 percent figure invites a reasonable question: how many calls does it actually represent? Invoca's report does not publish an exact count for ChatGPT-referred call volume, stating only that it remains low relative to established channels and that ChatGPT is the sole large language model currently producing measurable volume in the company's dataset. That caveat should temper how the figure gets used. A 49 percent lead rate calculated from a small sample carries wider statistical uncertainty than the same figure calculated from paid search's much larger call base, and Invoca's own framing, that this is a signal to watch rather than a channel to fund, reflects that limitation directly rather than glossing over it.

The pattern nonetheless fits a broader trend documented across multiple platforms tracking AI referral behavior over the past year. <cite index="17-1">ChatGPT's share of global generative AI website traffic stood at 76.4 percent in June 2025 before falling to 66.8 percent by December 2025 and further to 52.7 percent by June 2026, as Gemini and Claude gained ground</cite>, according to Similarweb data. That decline in relative share does not mean ChatGPT's absolute traffic shrank; it means competing assistants have captured a growing portion of a market that itself continues expanding. Against that backdrop, ChatGPT remaining the only assistant with measurable phone call volume in Invoca's dataset suggests either that ChatGPT users are more likely to complete a purchase-adjacent action like a phone call, or simply that ChatGPT's larger installed base has crossed a volume threshold competitors have not yet reached. The Invoca report does not attempt to distinguish between those explanations, and the underlying data does not yet exist to settle the question either way.

Industry and methodological context

Invoca's report breaks its lead and conversion figures out by industry and by channel, covering the same 10 industries and seven marketing channels represented in its overall dataset. The company states that all figures represent averages across its customer base, calculated from calls and conversation minutes tracked through its platform between an unspecified start date and the report's July 2026 publication. Because the dataset is proprietary to Invoca's own customer base rather than pulled from a market-wide phone carrier or a neutral third-party source, the benchmarks describe how Invoca's specific customers perform rather than the phone-based advertising market as a whole. That distinction matters for interpretation: a company using a different call tracking vendor, or none at all, would not necessarily see comparable channel splits, though the scale of the sample, over 70 million calls, gives the cross-industry patterns more statistical weight than a smaller study could offer.

The report also separates answer-rate quality from lead-conversion quality as distinct problems, which is a useful distinction for any team trying to diagnose why a channel underperforms. A channel that drives calls nobody answers looks identical, in a raw call-volume report, to a channel driving calls that get answered but rarely qualify as leads. Invoca's breakdown by industry, by channel and by call-duration threshold gives businesses a way to separate those two failure modes rather than treating "call volume" as a single undifferentiated metric.

Summary

Who: Invoca, a call tracking and conversation analytics company serving marketing, sales and contact center teams across 10 industries, published the findings; the report cites no named company spokesperson beyond Invoca itself as the study's author.

What: A benchmarks report analyzing more than 70 million phone calls and over 600 million minutes of conversation found that calls referred by ChatGPT convert to sales leads at a 49 percent rate, the highest of any channel measured and about 10 percentage points above the cross-channel average, while conversion rates from those same leads remained close to the overall average at 40 percent.

When: Invoca published the report today, July 14, 2026.

Where: The findings apply to Invoca's own customer base across 10 industries and seven marketing channels, with no geographic restriction specified in the report itself.

Why: The figures give marketing and measurement teams the first large-scale read on how generative AI research translates into offline, phone-based conversion, a category that web analytics tools have historically struggled to capture, while Invoca's own volume caveat, that ChatGPT remains the only large language model with measurable call volume, argues for treating the finding as an early indicator rather than grounds for reallocating budget.

Timeline

  • June 2025: Ahrefs publishes research finding AI search visitors convert 23 times higher than organic search visitors.
  • August 2025: Microsoft Clarity introduces AI traffic filters, separating visits from known AI referrer URLs into dedicated analytics categories.
  • November 2025: Research shows AI traffic converting at 3 times higher rates than traditional channels across more than 1,200 publisher and news sites.
  • December 18, 2025: Microsoft Clarity publishes its Agentic Web analysis, documenting 155 percent AI referral growth over eight months alongside an 11-fold sign-up conversion advantage.
  • March 2, 2026: Criteo discloses its ChatGPT Ads integration with OpenAI, reporting AI-referred users converting at 1.5 times the rate of other referral channels.
  • May 5, 2026: Criteo reports AI-referred conversion rates approaching twice those of traditional search in specific retail categories.
  • May 13, 2026: Google Analytics adds a native AI Assistant channel to its Default Channel Group, automatically classifying ChatGPT, Gemini and Claude traffic.
  • June 8, 2026: Google Analytics integrates directly with Google Business Profiles, importing phone call data among seven local interaction metrics.
  • June 15, 2026: Google requires new offline conversion imports, including phone-based conversions, to move through its Data Manager API.
  • June 22, 2026: Criteo reports more than 80 percent of ad-driven ChatGPT Ads traffic coming from new customers.
  • July 14, 2026: Invoca publishes the Lead Conversion Benchmarks Report 2026, finding ChatGPT-referred calls converting to leads at 49 percent, the highest rate of any channel measured.