Nearly three in four US consumer-brand marketers who run AI agents cannot follow a lead handled by one of those agents from the first ad click to a sale, according to the Broken Funnel Report that Invoca released on November 10, 2026. The survey of 451 marketing leaders found that only 19% of those running agents can reliably send conversation outcomes back to their ad platforms, and only 26% can tie an agent conversation to the campaign or ad that produced it. Invoca, a Santa Barbara company that sells call tracking, conversation analytics and AI agents to advertisers, also published separate editions for home services and healthcare.

In Short

Invoca asked 451 US marketing leaders at consumer brands how they measure the AI chat, text and voice agents that now answer their inbound leads, and most said they cannot see whether those conversations end in a sale. This matters to anyone buying search or social ads, because the bidding systems at Google, Meta and Microsoft learn from the results advertisers report back, and only 19% of these marketers report agent results back at all. What changes is the shape of the paid funnel: a growing share of ad clicks now ends in a conversation with software, and the survey suggests measurement has not caught up with that step.

451 marketers, 323 of them running agents

Invoca commissioned the research firm Gather to run the survey between September 14 and October 5, 2026. According to Invoca, respondents were US business-to-consumer marketing leaders at manager level and above, working at companies with 100 or more employees in five sectors: automotive, healthcare, home services, financial services and telecom.

The report itself adds detail that the announcement leaves out. Gather collected 505 responses and removed 54, leaving 451. The survey was a 20-question guided interview, 60% of respondents held director-level titles or higher, and the stated margin of error is plus or minus 4.6 percentage points at 95% confidence. The sector mix is uneven. Home services accounts for 30% of the sample, healthcare including dental 28%, financial services including insurance 25%, automotive 11% and telecom 6%.

The headline numbers about measurement do not describe all 451 respondents. Asked about AI agents, 44% said they had agents live, 27% were piloting them, 23% were evaluating, 5% had no plans and 1% were not sure. That combined 71% is the source of the announcement's line that "seven in ten brands have AI agents live or in pilot." It is seven in ten of the brands surveyed, not of US consumer brands at large. A blog post Invoca published on November 9, written by Derek Andersen, puts the base of marketers running agents at 323, and the measurement percentages apply to that group.

That base matters for reading the figures. By PPC Land's calculation, 19% of 323 is roughly 61 companies able to return agent outcomes to an ad platform reliably. The 4.6-point margin of error applies to the full 451, so the error band on the subgroup figures is wider, and the report does not state it.

Help desk metrics for a sales job

The core finding concerns what marketers count. According to the report, 27% of marketers running AI agents use revenue, bookings or conversion rate as the agents' primary key performance indicator. The other 73% use something else.

The breakdown shows how thin the revenue column is. Customer satisfaction leads as the primary measure at 20%. Qualified leads and conversation volume follow at 16% each, containment (the share of conversations the agent resolves without a human) at 10%, and cost per contact at 10%. One percent do not measure their agents formally at all. On the sales side of Invoca's line, 11% use conversion rate on agent conversations, 9% use appointments booked and 7% use revenue attributed to agent conversations. Those three add up to the 27%. Only 7% of agent-running marketers, in other words, measure the agents primarily on revenue itself.

The classification is Invoca's, and it is not the only possible one. Qualified leads, which sit closer to a sale than conversation volume or customer satisfaction, are counted on the engagement side of the split. Put them on the sales side and the share rises to 43%, by PPC Land's calculation, still well short of a majority.

Gregg Johnson, chief executive of Invoca, framed the gap in commercial terms. "Too many consumer brands measure their AI agents like a help desk and expect them to perform like a sales team," he said. "When prospective buyers start conversations with a brand, the AI agent is effectively a new sales channel. And you ultimately measure sales channels on revenue, not conversations or activity."

The announcement ties the issue to budgets. As finance chiefs look harder at whether AI spending produces business results, Invoca argues, agent deployments are at risk if marketers cannot show how the agents lift sales. PPC Land has reported similar pressure elsewhere: in August, a TransUnion survey of 100 senior US marketing and technology leaders found that only 53% reported meaningful ROI from AI-driven initiatives, even as 89% expected to increase AI-enabled marketing spend.

Where the trail goes cold

The measurement problem is not limited to what marketers choose to count. Many cannot follow the lead at all.

According to the report, 74% of marketers running agents cannot fully track an agent-handled lead from first click to conversion. Only 27% say their agents can see the buyer's digital journey, such as the ad clicked and the pages viewed before the conversation started. Only 26% can tie agent conversations to the campaign or ad that drove them. And 19% can reliably send outcomes back to their ad platforms.

The report then follows the lead past the agent. Once a lead is handed off, 56% of marketers say they lose attribution when it reaches a dealer, franchise or local branch. Text and chat handoffs break attribution for 46%, and phone handoffs for 40%. According to the report, only 21% can both see and influence leads all the way through to conversion.

Those handoffs describe how much of high-value consumer buying still works. A car is bought at a dealership; a roof replacement is quoted by a local franchise; an insurance policy may be closed on the phone. An AI agent that qualifies the buyer and books the appointment sits in the middle of that chain. If the sale happens later in a system the marketing team cannot see, the conversation that produced it shows up, at best, as an engaged lead.

Why the ad platforms care

The announcement contains a sentence that is easy to skim past. "That means AI-driven improvements in sales effectiveness may inadvertently hamper marketing efficiency," according to Invoca.

The reasoning runs through bidding. Automated bid strategies such as Google's Smart Bidding set bids for each auction based on the conversions an advertiser reports. Lead generation advertisers have for years been encouraged to feed sales and lead quality back into those systems, through offline conversion imports and value-based bidding, so that the algorithm learns which clicks produce revenue rather than which clicks produce a form fill. In August 2024, PPC Land covered a five-part Google series on value-based bidding for lead generation, which recommended judging performance over at least 50 conversions or a full month of data.

An AI agent inserted between the click and the sale changes what the ad platform can see. If a brand's agent converts more leads but the outcomes never reach Google Ads, Meta or Microsoft Advertising, the bidding system keeps optimizing toward whatever signal it does receive, which may be a chat started or a lead captured. What does a bidding algorithm learn from a lead that an agent qualified on Tuesday and a franchise closed the following week? In most of the companies surveyed, according to these figures, it never hears about the sale. The report's figure of 19% suggests that, for most of the agent-running brands surveyed, the closed loop between ad spend and agent-driven sales is not closed.

The plumbing for sending that data back has been shifting. In May, Google said the Google Ads API would stop accepting new offline conversion import adopters from June 15, pointing new integrations to its Data Manager API. In August, PPC Land reported that Google Ads attribution ignores offline conversions uploaded after seven days, a window that matters when a sale closes days after a conversation. Version 1.9 of the Data Manager API, released on November 4, added Google Ads call conversions, but only with SHA-256 hashed phone numbers. Microsoft Advertising, meanwhile, made its Conversions API generally available in late October with ten technology partners, Invoca among them.

OpenAI's ad business is building the same dependency. PPC Land reported in May that conversion-optimized ChatGPT campaigns would roll out from June 5, with early access tied to conversions configured through OpenAI's pixel or Conversions API. Every platform that optimizes toward conversions depends on advertisers reporting them, and an agent conversation is one more step where that reporting can stop.

AI answer engines at the top, agents at the bottom

The survey's most positive number concerns the start of the journey. According to Invoca, 86% of all 451 respondents can identify leads who began their research in an AI answer engine such as ChatGPT or Google AI Mode. The report qualifies this: 46% say they can do so for most such leads, which means the 86% includes many marketers who can do it only some of the time.

Invoca has made the case before that these buyers are valuable. In July, PPC Land reported Invoca data showing that calls referred by ChatGPT converted to leads at 49%, ahead of every other channel in its benchmark. HubSpot research published in June found that buyers who used AI search while evaluating CRM software were 36% more likely to purchase.

The announcement describes the ad platforms moving into the same conversational space. "From Google to ChatGPT, ad platforms are exposing AI agents in their ad products," it states, "putting the sales conversation at the forefront of the buyer journey." Google introduced Business Agent for Leads in US beta at Google Marketing Live in May. Invoca itself released Messaging Agents For Google Ads on October 28, a beta product that answers RCS chats started from Google Search ads and organic results and sends sales outcomes back to Google Ads. Google has also worked the other direction: in July 2025, it announced an AI feature that calls local businesses on behalf of US Search users to check prices and availability.

So the industry now has AI at both ends. A consumer may research with an AI system, click an ad and talk to a brand's AI agent before any human is involved. The report's argument is that the middle of that path, where the agent's work turns into revenue, is the part most brands cannot measure.

Two surveys, two answers on speed

The most striking contrast in the announcement compares two different groups of people. According to the Broken Funnel Report, 84% of marketers say they respond to web leads within an hour. According to Invoca's 2026 B2C Buyer Experience Report, only 36% of consumers say they hear back that fast.

A 48-point gap looks stark, but the two figures are not a matched pair. The marketer figure comes from the September and October survey of 451 marketing leaders. The consumer figure comes from an earlier Invoca survey, also conducted by Gather, fielded between May 8 and May 22, 2026. Invoca's report on that survey describes 1,356 consumers across the US and UK, of whom 693 were in the US, all of whom had researched a high-stakes purchase, generally above $500 or above $1,000 for travel, in the previous 12 months. The consumers were not asked about the brands whose marketers took the later survey. According to the Broken Funnel Report, 56% of consumers expected a response within an hour, which leaves a 20-point gap between what consumers expected and what they got.

The consumer study also measured what happens after a slow response. Asked how they reacted, 46% said they contacted the business again, 24% waited, 27% switched to a competitor and 3% gave up, according to Invoca. In a separate question, 79% said they would switch to a competitor that responds faster.

Marketers report losses of their own. According to the report, 25% say they lose 20% or more of inbound leads to slow or missed responses. The full distribution is wider than the headline suggests: 18% put their losses below 5%, 24% between 5% and 9%, 31% between 10% and 19%, 16% between 20% and 29%, 7% between 30% and 49%, and 2% at half or more.

What marketers want from agents

Asked to rank the capabilities they want most, respondents put context first. According to the report, 43% of all 451 marketers placed tailoring conversations to the buyer's digital journey data in their top three, and 41% chose recognizing returning customers and their history. Qualifying leads, scheduling appointments and warm transfer to a human agent each drew 39%.

The gap between wish and reality is direct. Only 27% of marketers running agents say their agents can see the buyer's digital journey today, according to the report, which is what the announcement means when it says fewer than a third can use that data to tailor a conversation.

Home services: more agents, fewer outcomes returned

The two vertical editions draw on subsets of the same survey, and their samples are smaller.

The Home Services Edition covers 137 respondents, of whom 113 run agents. Adoption is higher than in the full sample: 50% have agents live and 33% are piloting them, a combined 83%. According to the edition, SMS is the most common live channel at 82%, followed by phone at 77% and website chat at 74%.

Home services brands measure agents on sales more often than the average. According to the edition, 35% use a revenue or booked outcome as the primary measure, compared with 27% across the full sample. Their tracking is weaker, though. Some 81% cannot track an agent-handled lead from click to conversion, and only 14% send outcomes back to their ad platforms, five points below the overall figure.

The handoff problem is sharper in a sector built on franchises and local operators. According to the edition, 73% of home services marketers cannot see and influence leads once they reach franchises or branches, 70% lose visibility through text or chat handoffs and 66% through phone handoffs. At the top of the funnel, the sector scores higher than average: 95% can identify leads that began in AI answer engines, and 59% can do so for most of them, against 46% across the full sample.

On speed, 88% of home services marketers say they respond within an hour, while 57% of home services consumers expect to hear back that fast and 41% do. According to the edition, 23% of these marketers lose 20% or more of their leads.

These numbers arrive as Google changes the economics of unanswered calls in the sector. From October 1, Google began charging Local Services Ads advertisers for missed business-hours calls when the caller stays on the line for more than 20 seconds. Five days later, Invoca packaged its AI agents for home services partly around that change. For a brand paying for missed calls on Local Services Ads, an agent that answers is one fix; knowing which of those answered calls became a job is a separate problem, and the edition suggests most home services marketers have not solved it.

Healthcare: slower adoption, higher feedback

The Healthcare Edition covers 125 respondents, of whom 73 run agents. Adoption is lower: 36% have agents live and 22% are piloting, just under 60%, compared with 71% across all sectors. Another 31% are evaluating and 10% have no plans.

According to the edition, 22% of healthcare marketers running agents use a revenue or booked outcome as the primary measure, below the overall figure. Customer satisfaction leads at 23%, followed by conversation volume at 17% and qualified leads at 16%. Some 71% cannot track an agent-handled lead from click to conversion.

One figure runs against expectations. According to the edition, 23% of healthcare marketers running agents send outcomes back to their ad platforms, higher than the 19% overall and well above the 14% in home services, despite the restrictions on using patient data in US advertising. The base for that figure is 73 marketers, so 23% amounts to roughly 17 organizations, by PPC Land's calculation, and the edition does not explain the difference.

The location handoff is where healthcare struggles most: 83% lose attribution once a lead reaches a location, 71% through text or chat and 68% by phone. According to the edition's introduction, only 17% can both see and influence a lead after it has been routed. On speed, 83% of healthcare marketers say they respond within an hour, 53% of healthcare consumers expect that and 32% receive it. A quarter of healthcare marketers say they lose 20% or more of leads.

Invoca released its healthcare package, with AI agents that book appointments in Epic and four other EHR systems, on October 20.

What the survey does not establish

The Broken Funnel Report is research commissioned by a company that sells the remedy it describes. Invoca's announcement closes by explaining that its platform gives AI agents buyer journey data, ties conversations to the campaign or ad that drove them and sends outcomes back to ad platforms "to power AI-driven optimization and improve return on ad spend." The gaps the report identifies map closely onto the products Invoca released in October.

Several limits follow from the method. Every figure is self-reported by marketers about their own capabilities, and the survey has no measure of how agent-handled leads actually perform against leads handled by people. It does not say whether agent leads convert better, worse or the same. It does not name which ad platforms the 19% send data to, or define what "reliably" means. The consumer response-time figure comes from a different survey, of different people, conducted four months earlier. And the sample is US only, weighted toward home services and healthcare, with telecom represented by roughly 27 respondents by PPC Land's calculation.

What the survey does show, within those limits, is a consistent pattern across sectors: marketers are deploying conversational agents faster than they are connecting them to the revenue data that ad platforms use to decide who sees an ad and at what price.

Timeline

Summary

Who: Invoca, a Santa Barbara company selling call tracking, conversation analytics and AI agents to advertisers, which commissioned the research firm Gather to survey 451 US business-to-consumer marketing leaders. Invoca CEO Gregg Johnson is quoted. Of the respondents, 323 run AI agents that are live or in pilot.

What: The Broken Funnel Report, which finds that 74% of marketers running AI agents cannot track an agent-handled lead from first click to conversion, 26% can tie agent conversations to the campaign or ad behind them, 19% can reliably send outcomes back to ad platforms, and 27% use revenue, bookings or conversion rate as the agents' primary measure, with only 7% using revenue itself. Home Services and Healthcare editions break out results for 137 and 125 respondents.

When: November 10, 2026. The survey ran from September 14 to October 5, 2026; the consumer response-time figure comes from a separate Invoca survey fielded May 8 to 22, 2026.

Where: The United States, across automotive, healthcare, home services, financial services and telecom companies with 100 or more employees.

Why: Ad platforms optimize bids toward the conversions advertisers report. When an AI agent sits between the ad click and the sale and its outcomes are not sent back, the bidding systems at Google, Meta, Microsoft and OpenAI cannot learn which clicks produced revenue. The findings are self-reported and come from a vendor that sells tools to close that gap.