Here's a stat that should make every CMO sit up straight: calls referred by ChatGPT convert to qualified leads at 49%, roughly 10 percentage points higher than any other marketing channel. That's according to Invoca's Lead Conversion Benchmarks Report 2026, which analyzed over 70 million phone conversations across 10 industries.
Let that sink in. The channel you probably aren't tracking properly is outperforming the channels you've been optimizing for years.
The Buyer Who Already Did Their Homework
The finding makes intuitive sense once you stop thinking about ChatGPT as a search engine and start thinking about it as a research assistant. When someone asks ChatGPT to compare health insurance plans in Austin with low deductibles that cover pediatric specialists, they're not browsing. They're narrowing. By the time they pick up the phone, they've already eliminated the options that don't fit.
Invoca's press release notes that 41% of consumers now research major purchases with generative AI. These aren't tire-kickers. They're buyers who arrive "more informed and closer to a decision." The lead rate reflects that intent.
But here's where it gets interesting: once those calls are answered, the conversion rate drops to 40%, which is actually slightly below the all-channel average of 42%. The leads are better. The close rate is average.
The Gap Between Lead Quality and Conversion
What explains the disconnect? A few possibilities.
First, sales teams may not be calibrated for these callers. Someone who spent 20 minutes with ChatGPT comparing options doesn't need the standard discovery script. They need answers to specific questions, and they need them fast. If your reps are still running through qualification checklists while the prospect is ready to buy, you're creating friction where there should be momentum.
Second, the report reveals a broader problem: 64% of businesses don't ask leads to buy or book appointments during the call. That's not a ChatGPT problem. That's a sales enablement problem. You can have the highest-intent leads in the world, but if nobody asks for the business, intent doesn't convert.
Third, there's the answer rate issue. Across all industries, 44% of callers never reach a person. Nearly half of the demand you're paying to generate evaporates before anyone picks up. For calls lasting more than 30 seconds, the answer rate climbs to 71%, but that still means three out of ten engaged callers are hitting voicemail.
The Attribution Blind Spot
Now for the uncomfortable part: most marketing teams have no idea how many leads ChatGPT is actually sending them.
Search Engine Journal's analysis of the report points out that Invoca doesn't publish the raw volume of ChatGPT-referred calls, only noting that the overall volume "remains very low." When a rate is derived from a small base, it's less reliable than the same rate calculated from paid search's much larger volume. This is the first year Invoca had enough data to measure generative AI as its own channel at all.
The attribution mechanics are also murky. The report doesn't detail how Invoca labels a call as ChatGPT-referred. Did the caller click from ChatGPT? Use a tracked number? The methodology section is thin on specifics. Gemini, Claude, and Perplexity aren't included in the channel breakdown, which Invoca frames as a measurement limitation rather than a judgment on those platforms.
This matters because attribution is already breaking down across the board. Customer journeys are fragmented, privacy regulations have reduced observable signals, and identity breaks across devices. Adding AI-assisted research to the mix creates another layer of opacity. The buyer who asked ChatGPT for recommendations, visited your site, left, came back via organic search, and then called from a different device? Good luck tracing that path.
What This Means for Your 2026 Playbook
The strategic implications are clearer than the measurement challenges.
Optimize for AI visibility, not just search visibility. If ChatGPT is recommending your competitors and not you, you're losing high-intent buyers before they ever see your brand. This means thinking about how your content, reviews, and structured data appear to large language models, not just to Google's crawler. ChatGPT processes 2.5 billion prompts per day. Some of those prompts are about your category. The question is whether your brand shows up in the answer.
Treat phone calls as a conversion channel, not a support channel. The report's finding that 64% of businesses don't ask for the sale during the call is damning. If your contact center is staffed and trained for customer service rather than revenue generation, you're leaving money on the table. The highest-intent leads in your funnel are calling you, and you're not closing them.
Fix the answer rate before you fix the ad spend. Invoca's data shows that answer rates vary by industry, but the cross-industry average of 56% means nearly half of all calls go unanswered. Before you pour more budget into demand generation, audit how much of that demand is actually being engaged. AI-powered voice agents can handle after-hours calls, qualify intent, and schedule callbacks. The technology exists. The question is whether you're using it.
Build measurement infrastructure for a channel that barely exists yet. ChatGPT call volume is small today. It won't stay small. ChatGPT reached 894 million unique users in April 2026, and consumer behavior is shifting toward AI-assisted research for high-consideration purchases. The brands that build attribution and tracking now will have a data advantage when the volume scales.
The Bigger Picture
Marketing has always been about meeting buyers where they are. For two decades, that meant search engines. For the last decade, it meant social platforms. Now it means AI assistants.
The Invoca report is a signal, not a conclusion. The sample size is small, the attribution is imperfect, and the conversion rate is average. But the lead quality is undeniable. People who research with ChatGPT and then call are further along in their decision than people who click a paid search ad.
That's not a reason to abandon your existing channels. Paid search still drives the most volume. Google Business Profiles still dominate organic. But it is a reason to start treating generative AI as a legitimate source of demand, with its own optimization strategies and its own measurement requirements.
The buyers are already there. The question is whether you're ready to answer.
