Invoca on October 20, 2026 released a data and AI product for healthcare providers that puts AI voice and messaging agents on patient calls, texts and web forms, books appointments directly in electronic health record systems including Epic, and sends de-identified booking results back to advertising platforms for bid optimization. It was the Santa Barbara company's second vertical package in two weeks, following a home services version on October 6.

In Short

Invoca, a company that tracks phone calls from ads, now sells AI assistants that answer patients' calls and texts at any hour and put appointments straight into hospital and clinic booking systems. Hospitals and clinics that advertise on Google or Meta are the target, because those ad systems usually cannot see whether an ad led to a booked visit. Invoca says it strips sensitive details before sending results to the ad platforms, though its materials do not explain how that works or what exactly gets sent.

What Invoca put on sale

The announcement describes a single package rather than separate tools. According to Invoca, it joins marketing campaigns, websites, AI agents and patient access centers into one system, and it is built for health systems, cancer centers, dental networks, specialty practices and senior care organizations. The product is available now, according to the company. No price appears in the release, the accompanying blog post or the product page at healthcare.invoca.com.

The release lists five capabilities. The first is a set of AI voice and messaging agents that respond around the clock to calls, texts and web forms, answer common questions, qualify new patients and route "clinical or complex conversations" to staff with the conversation history attached. Invoca describes these agents as HIPAA-compliant. The release does not say how that status was assessed.

The second is booking. According to Invoca, the agents book, reschedule and cancel appointments autonomously in Epic, athenahealth, eClinicalWorks, NextGen Office and ModMed, along with unnamed "leading dental and dermatology practice management systems." They also create and update patient records and log interactions in customer relationship management software such as Salesforce Health Cloud and Microsoft Dynamics. That is a step beyond logging a call in a CRM: the agent writes directly into the scheduling system a clinic runs on. The product page names a slightly different set of scheduling connections - Epic, NexHealth, ModMed, Dentrix and Calendly - so the exact list depends on which Invoca document is read.

The third is drop-off analysis. Invoca says its data shows where a patient journey breaks down, citing unanswered after-hours calls, slow follow-up on web forms and abandoned online scheduling as examples. The fourth is attribution: every call, text and form arriving from search, AI discovery, social and TV campaigns is tied to the appointment it produced. The fifth, which the release labels "Protect patient privacy," is a single sentence stating that the platform is supported by a business associate agreement and "automatically redacts sensitive information the moment it's shared."

Three worked examples

The release describes how three kinds of organization would use the product. In a health system, an orthopedics inquiry driven by a campaign would be routed to the right service line in the access center and queued ahead of routine rescheduling requests. For dental groups and dental service organizations, AI agents would answer insurance questions at any hour, book cleanings and consultations into the practice schedule, and flag implant and orthodontic inquiries - which Invoca calls "high-value" - for staff. In senior living and in-home care, where Invoca says families tend to research in the evenings and at weekends, agents would answer questions, help determine the level of care required and book community tours or care consultations.

The orthopedics and implant examples carry an obvious commercial logic. Both involve higher-revenue procedures, and both describe the system prioritizing patients by the value of the service they ask about rather than by the order in which they called. Invoca frames this as expanding access to care. Whether queue-jumping based on service line raises questions for patient access teams is not addressed in the material.

The measurement gap Invoca is selling into

The problem the product targets is a familiar one in paid media. Ad platforms bid toward the conversions they can observe. A click that leads to a phone call, and a call that leads to a booked appointment three days later, often never reaches the platform at all, so automated bidding learns from form fills and page visits instead of from patients who actually show up.

According to Invoca, de-identified conversion results flow back to Google, Meta, Adobe Advertising and The Trade Desk "so bidding optimizes toward the patients who actually book and show." Feeding confirmed bookings into Smart Bidding and similar systems is the stated purpose. The blog post accompanying the release names booked appointments feeding Google Smart Bidding and Performance Max specifically.

Timing matters for that loop. PPC Land reported in August that Google Ads attribution ignores offline conversions uploaded more than seven days after the event, while standard reporting columns still count them. An appointment booked by an AI agent minutes after a call falls easily inside that window. An appointment that is later attended, or a "show" in Invoca's wording, might not, depending on how far out the visit is scheduled. The release does not say whether Invoca sends the booking event, the attendance event, or both.

Which platforms receive the data

The three Invoca documents linked from the announcement do not agree on where results go. The release names Google, Meta, Adobe Advertising and The Trade Desk. The blog post lists Google Ads, Salesforce (including Agentforce Health), Microsoft Dynamics and Adobe Experience Platform as conversion destinations. The product page shows a longer list of ad integrations: Google Ads, Search Ads 360, Google Analytics, Adobe, The Trade Desk, MNTN, Meta, TikTok, Reddit Ads, StackAdapt and tvScientific. None of the three describes the fields transmitted, the API used for each platform or the delay between booking and upload.

Privacy: what the documents say and what they leave out

Healthcare is the vertical where sending conversion data to ad platforms carries the most legal exposure in the United States. Under HIPAA, a vendor that handles protected health information for a covered entity has to sign a business associate agreement that carries the law's obligations downstream. The Department of Health and Human Services issued guidance on online tracking technologies in December 2022, revised it in March 2024, and saw part of it vacated by a federal court in June 2024. Hospitals have faced years of litigation over pixels sending browsing data to Meta. On October 8, the Washington Supreme Court ruled in one of those cases, and PPC Land reported the next day that parents lost their Meta Pixel wiretap case against Seattle Children's Hospital because the state statute covers communications between individuals, not clicks answered by a server.

Against that background, Invoca's privacy claims are brief. The release says the platform is "BAA-supported" and redacts sensitive information automatically. The product page adds that sensitive patient data is "redacted the moment it's shared, never stored," and lists SOC 2 Type 2 and PCI DSS credentials. The blog post says compliance and legal teams get controls over call recording, data redaction and data access.

Several things are not stated in any of the linked materials:

  • De-identification method. HIPAA recognizes two routes to de-identification: the safe harbor method, which removes 18 categories of identifiers, and expert determination, in which a statistician documents that re-identification risk is very small. Invoca does not say which, if either, it uses for the results sent to ad platforms.
  • What is sent. Matching a conversion to an ad click normally requires some identifier, such as a click ID or a hashed contact detail. The release does not say what identifier accompanies a "de-identified" booking, or how a platform could attribute a booking to a campaign without one.
  • Consent. None of the documents describes how patients are told that a call may be answered by an AI agent, recorded or analyzed, or how consent is collected. Washington, the state in the Seattle Children's case, is one of 11 all-party consent states for recordings.
  • Platform rules. The release does not address whether Meta or other platforms place limits on health-related conversion events, or how Invoca handles them.

None of this means the product is non-compliant. It means the claims are made at the level of a badge rather than a mechanism, and the healthcare marketers the product is aimed at would need the detail from Invoca directly.

The benchmark behind the pitch

Invoca's case rests on its 2026 Healthcare Lead Conversion Benchmarks Report, which the release cites for two figures: only 54% of calls to healthcare organizations reach a person, and lifting answer, lead and conversion rates by five points each would produce 35% more booked appointments.

The report page, published in July 2026, gives more. Answer rates rise to 65% for calls longer than 15 seconds and 72% for calls longer than 30 seconds. Across six sub-industries - behavioral health and rehab, dental, eye care, hearing, hospitals and physical therapy - answer rates range from 45% to 68%. According to the report, 43% of answered calls are leads, 45% of those leads convert on the call, and 43% of healthcare organizations do not ask patients to book an appointment. ChatGPT-referred calls show a 47% lead rate and a 41% conversion rate.

The report has limits that the release does not mention. The data comes from calls tracked on Invoca's own platform and is averaged across Invoca's customer base, not across the healthcare market. The page cites the overall dataset - more than 70 million calls and 600 million minutes across 10 industries and seven channels - but gives no healthcare-only sample size and no collection period. It also uses two denominators for the same figure: the highlights say 43% of calls from digital marketing are leads, while the body says 43% of answered calls are leads.

The 35% figure is modeled, not observed. It is the arithmetic result of compounding three five-point improvements, not the measured result of any deployment. The release says the improvement "would produce" more booked appointments; the pitch email accompanying it drops the conditional and says it "produces" them.

Compared with home services

The healthcare numbers sit close to the figures Invoca published for its first vertical. When PPC Land covered the home services launch, it reported that 52% of home services calls reach a person and that a five-point lift on each rate implies about 38% more conversions. Invoca's cross-industry figure from earlier in the year was 56%, and its July data showed ChatGPT-referred calls converting to leads at 49%, the highest of any channel it measured. Healthcare, by Invoca's count, answers slightly more calls than home services and converts a similar share.

Customer claims

The release names University Hospitals, Nemours Children's Health, Ochsner Health, AtlantiCare, CHRISTUS Health and Amplifon as healthcare customers that use Invoca to tie marketing to patient outcomes. It does not say how many of them use the new AI agents, as opposed to Invoca's existing call tracking.

One customer is quoted. "We see AI agents making it easier for every patient to get help when they need it," said Matt Eaves, VP of Digital Marketing at University Hospitals. "Our Invoca AI agent recovered 25 appointments on its first night, and it also knew its limits. When a patient texted for help from inside our hospital, it directed him to a staff member right away. That combination of speed and judgment is what will earn patients' trust."

The 25 appointments are a single night's count with no baseline. The release does not say how many conversations the agent handled that night, how many appointments staff would have booked the following morning from the same inquiries, or how many of the 25 were kept. "Recovered" implies those appointments would otherwise have been lost, which is the vendor's and customer's interpretation rather than a measured counterfactual.

The product page carries further claims without methodology or time periods: a 300% increase in conversion rate at University Hospitals, a 43% conversion rate increase at Senior Helpers, a 45% increase in marketing-driven leads at Nemours Children's Health, and a 20% reduction in cost per acquisition at eHealth. It also states operational figures for the platform: a target of under 800 milliseconds for 95% of AI voice responses, 99.999% call uptime and 253 million calls handled over the trailing 12 months. All are Invoca's own numbers.

What the chief executive says

Gregg Johnson, Invoca's chief executive, positioned the product against simpler chatbots. "In healthcare, the moment a patient reaches out is the moment that matters most," he said. "Most AI agents stop at answering basic questions. Invoca's agents go further, uniting AI with data that shows which marketing drove each conversation and which appointment it produced. That coordinated system is how providers grow revenue while expanding access to care."

The quote makes the commercial argument explicit. The differentiator Invoca claims is not the agent itself, a category now crowded with vendors selling agentic AI for customer service, but the attribution data the company has collected from call tracking for more than a decade. The agent books the appointment, and the call-tracking history tells the marketer which campaign paid for it.

The company

Invoca describes itself as an "AI-powered revenue execution platform" connecting marketing, commerce and contact center teams. According to the company, it has raised $184 million from investors including Upfront Ventures, Accel, Silver Lake Waterman, H.I.G. Growth Partners and Salesforce Ventures, and its customers include Mayo Clinic, Mutual of Omaha and Verizon. The release says the team will attend the Healthcare Internet Conference (HCIC) from October 25 to 28.

Why this matters for healthcare marketers

The healthcare package follows a pattern Invoca set on October 6. The home services release arrived days after Google began charging Local Services Ads advertisers for unanswered calls longer than 20 seconds from October 1, and it leaned on that billing change. The healthcare version has no equivalent trigger. Its argument is the slower one about bidding signals: if half of patient calls go unanswered and the ad platform never learns which clicks became visits, automated bidding is working from incomplete data.

That argument has weight in a category where most of the value is offline. A hospital's paid search account can record a click and perhaps a call, but the visit, the procedure and the revenue sit in an electronic health record the ad platform never sees. Vendors have spent the past year building pipes between health data and media. DeepIntent, for example, built a HIPAA-ready data cloud called Helix in March for pharmaceutical and healthcare advertisers. Regulated healthcare advertising is also growing on the supply side: LegitScript reported in September that telemedicine certification applications rose 101% in a year.

Invoca's product moves in the opposite direction from DeepIntent's. It does not bring health data into an ad platform for targeting. It takes outcomes from inside a provider's own scheduling system and sends a stripped-down version back out to optimize bids. That narrower flow is easier to defend, but only if the stripping is done in a way that holds up. On that question, the material published on October 20 offers a description rather than an explanation.

Two other questions remain open. One is how ad platforms will treat a conversion signal generated partly by an AI agent rather than a human receptionist; the release does not say whether Invoca labels agent-booked appointments differently. The other is cost. Invoca has published no pricing for either vertical package, which leaves the economics of replacing after-hours staff or answering services with AI agents impossible to assess from the outside.

Timeline

Summary

Who: Invoca, a Santa Barbara call tracking and conversation analytics company that has raised $184 million, and healthcare providers including University Hospitals, Nemours Children's Health, Ochsner Health, AtlantiCare, CHRISTUS Health and Amplifon.

What: A healthcare data and AI package with AI voice and messaging agents that answer patient calls, texts and forms, book appointments directly in Epic, athenahealth, eClinicalWorks, NextGen Office and ModMed, and send de-identified booking results to Google, Meta, Adobe Advertising and The Trade Desk for bid optimization. Invoca's benchmark says 54% of calls to healthcare organizations reach a person. The materials do not explain the de-identification method, the data fields sent to ad platforms, patient consent, or pricing.

When: Released October 20, 2026, two weeks after Invoca's home services package on October 6.

Where: United States healthcare providers, with the team at HCIC from October 25 to 28.

Why: Ad platforms optimize toward conversions they can see, and in healthcare the booked appointment usually happens in a call, text or form outside their view. Invoca argues that answering more inquiries and feeding confirmed bookings back to bidding systems would raise booked appointments, a claim based on its own modeled benchmark and customer-reported results.