This article is about how consumer digital health companies are found and chosen. It is not health advice, and nothing here says any app or program works.
The short version
- In Rock Health’s survey of 8,000 US adults (opens in a new tab), fielded in December 2025, 32% had used AI chatbots for health information, up from 16% a year earlier, and 64% of those users ask health questions weekly or more.
- General assistants dominate: 74% of AI health users (opens in a new tab) turned to tools like ChatGPT, against 5% for chatbots offered by providers. The same people trust health apps far more than non-users do (55% vs. 25%).
- The shelf is crowded: IQVIA counts (opens in a new tab) 337,000 digital health apps, and US digital health startups raised $7.4 billion across 244 deals (opens in a new tab) in the first half of 2026, with mental health and weight management drawing the most money.
- Each new user is hard won: in RevenueCat’s 2026 benchmarks (opens in a new tab), the median health and fitness app converts 2.9% of downloads to paying users, and 68% of the category’s subscriptions are annual plans.
- OpenAI now connects apps such as MyFitnessPal, Weight Watchers and Peloton inside ChatGPT Health (opens in a new tab), which it says fields health questions from over 230 million people a week.
Who are a digital health company’s customers, and what is one worth?
Mostly engaged adults who track their health and pay for an annual plan or a program, after a trial.
The consumer side of digital health covers wellness and fitness apps, mental health and sleep apps, weight management programs, condition management apps and wearable companions. Rock Health’s 2025 survey describes the people most likely to use AI for health as engaged trackers: 43% of AI users track sleep, against 28% of non-users, and they follow an average of four health metrics. They are the natural audience for health apps.
The commercial engine is a subscription or a paid program. RevenueCat’s 2026 report, which covers apps using its billing platform, gives the clearest public numbers for the health and fitness category:
| Health and fitness apps (RevenueCat 2026, medians) | Value |
|---|---|
| Downloads that become paying users | 2.9% |
| Trials that convert to paid | 37.7% |
| Trials started on the day of download | 82.1% |
| Share of subscriptions on annual plans | 68% |
| Realized lifetime value after one year, per payer | $35.64 |
Two things follow. First, the decision happens fast: most trials start the day the app is installed, so the research that matters happens before the download. Second, the category depends on annual commitment, which favors buyers who arrive already convinced. Our inference is that a user who was named a product by an assistant, checked it and then downloaded it is the kind of user that annual-plan economics need. No public data yet compares the lifetime value of users referred by AI with others.
Direct-to-consumer care programs, such as coaching and weight management programs, sell enrollment rather than an app subscription, at higher prices. We found no public benchmark for their customer value and will not invent one. Virtual care providers face their own questions on price and insurance; see how telehealth companies get chosen.
Where do AI assistants sit in a health app user’s research?
At the start, for a growing third of adults, and mostly inside general assistants rather than health system tools.
Rock Health’s survey is the most direct evidence. Among US adults, ChatGPT was the most used chatbot for health information (23%), followed by Gemini (15%). Fierce Healthcare’s summary (opens in a new tab) notes that people use AI for questions to ask at appointments, for mental health needs and for “looking for specific providers and clinics.” After a chatbot answer, 42% searched for more information and 40% consulted a provider. Millennials (48%) and Gen Z (45%) lead adoption.
The platforms are building for this. When OpenAI launched ChatGPT Health in January 2026, it documented that people can connect Apple Health and wellness apps such as Function and MyFitnessPal, with Weight Watchers and Peloton among the listed connections. OpenAI says apps in Health must meet its privacy and security requirements and pass an additional security review, and that ChatGPT may suggest a connected app when helpful. It also states that Health “is not intended for diagnosis or treatment.” Those are statements documented by the platform; how often ChatGPT suggests a particular app is not public.
App discovery in general is shifting too. In AppTweak’s survey (opens in a new tab) of 1,000 US adults across all app categories, 10% first learned about their most recent download from an AI assistant, level with Google, and 59% still went to the store to check reviews, ratings or screenshots after an AI recommendation. We cover the wider app picture in how consumer apps win subscribers from AI search.
What do people ask AI assistants before they download a health app?
Which app fits a goal, how two programs compare, whether it protects data and whether it has evidence behind it.
We wrote these sample prompts ourselves to show how people shop for a health app or care program. None comes from a real user, and none asks for a diagnosis or treatment.
| Stage | Illustrative prompt |
|---|---|
| Category | “What’s a good sleep tracking app that works with my Apple Watch?” |
| Program choice | “Compare structured weight management programs that include coaching” |
| Alternatives | “Cheaper alternative to my meditation app with offline sessions” |
| Privacy | “Does this period tracking app share my data with advertisers?” |
| Evidence | “Which stress apps have published studies behind them?” |
| Cost and coverage | “Is this program covered by my employer or insurer?” |
| Legitimacy | “Is this app legit? What do reviews say about canceling?” |
Privacy questions deserve special weight in this industry. Rock Health found AI users are about twice as willing as non-users to share health data with health tech companies (23% vs. 11%), but the regulators are watching how apps handle it. The Federal Trade Commission (opens in a new tab) says its July 2024 amendments make clear that makers of health apps and connected devices must comply with the Health Breach Notification Rule, with civil penalties of up to $53,088 per violation. Our inference: an assistant asked whether an app protects data can only answer well if the company has said so clearly, in text, somewhere it can read.
How does an AI answer turn into a paying user?
Through a short list, a store check, a same-day trial and an annual plan.
- Named for a goal. The assistant lists a few apps or programs for “sleep tracking with a wearable” or “weight management with coaching.”
- Checked. The person reads store ratings, privacy labels and reviews, or asks the assistant whether the company is legitimate. When our brand reputation study asked assistants whether brands were legitimate, 88.0% of the answers cited a review or complaint platform.
- Downloaded and trialed. In RevenueCat’s data, 82.1% of health and fitness trials start on the day of install.
- Converted. About 37.7% of trials convert at the median, mostly to annual plans. That is when the recommendation becomes revenue.
For care programs, step three is an enrollment form or an eligibility check rather than an app trial, and the stakes of step two are higher because people are sharing more personal information.
There is also a newer path that skips the store. If an app is connected inside ChatGPT Health, OpenAI documents that the assistant can reference that app’s data or suggest the app during a conversation. We infer that being a well-described, trusted connected app may matter more over time, though no data yet shows how many users arrive this way.
What decides which health app an assistant names?
Evidence the assistant can verify about the product and the company; the platforms do not publish their selection rules.
What is documented: OpenAI requires apps in ChatGPT Health to meet privacy and security requirements, and Google says (opens in a new tab) SEO best practices remain relevant for AI Overviews and AI Mode, with no additional requirements or special optimizations needed to appear. Apple’s App Review Guidelines (opens in a new tab) add a gate that shapes what gets into the store at all: under guideline 1.4.1, apps must disclose data and methodology to support accuracy claims about health measurements, and medical apps may be reviewed with greater scrutiny.
What has been observed:
- Health answers favor institutions. In an audit of 615 sources cited by ChatGPT for consumer health questions, Jacques and colleagues (opens in a new tab) found commercial health platforms earned 12.4% of citations. Those that were cited stated a medical review on 71.1% of pages. We explain the pattern in what commercial health sites cited by ChatGPT have in common.
- Lists move. In our consistency study, only 25.2% of the brands ChatGPT named appeared in all five repeats of a question, so a single check proves little.
What we infer: in this industry, the trust factors are unusually concrete. They include an accurate statement of regulatory status, published evidence that is described honestly, plain-language privacy terms, clinical or expert involvement that is named, a strong store rating and a clean record on cancellations and billing.
Why is it hard for a health app to describe itself to an assistant?
Because the rules limit health claims, so the proof must be specific, accurate and plainly worded.
The Food and Drug Administration revised its general wellness policy (opens in a new tab) on January 6, 2026, superseding the 2019 version. As law firm Covington & Burling summarizes it, the update lets more non-invasive wearables that output measures such as blood pressure stay in the low-risk wellness category if their use is strictly wellness. But wellness products still may not claim clinical equivalence, clinical accuracy or medical grade. Covington notes the tension this creates for companies that want to show their product is validated without crossing into restricted claims. Wearable makers work under the same revised policy, as covered in how wellness brands get recommended for sleep and recovery.
That tension applies directly to AI answers. Anything an assistant says about an app, it first had to read somewhere. If an app’s site overstates its evidence, the company takes on legal risk; if the site says nothing specific, the assistant has nothing to repeat. IQVIA reports that more than 360 software-based digital therapies are now on the market, 140 of them prescription digital therapeutics, so many consumer apps sit beside products with formal clearances. Our inference: the winning description says exactly what the product is, whether it is cleared and for what, what the evidence shows and where it is published, and what the app does not do.
How does generative engine optimization work for a digital health company?
Generative engine optimization (GEO) makes your app easy for assistants to find, describe accurately and verify through outside evidence.
For a consumer health company, the work usually covers:
- One consistent identity. The same product name, category, platforms and plan details across your site, both app stores and any connected-app listings, so assistants do not confuse your app with a similarly named one.
- Regulatory status in plain words. Whether the product is a general wellness product or cleared, with a link to the FDA listing where one exists. Never imply clearance you do not have.
- Evidence pages written for readers. Summaries of published studies with links, who ran them and their limits, reviewed by named experts, without claims beyond what the evidence and your regulatory status allow.
- Privacy that answers the question. A readable page on what data you collect, what you never share for advertising and how people delete it, matching your store privacy labels.
- Reputation where people check. Strong ratings and fast responses in the app stores, clear cancellation terms and honest handling of complaints. Avoid any review tactic regulators would treat as deceptive; fake reviews also distort AI answers.
- Independent coverage. Editorial reviews, clinician and dietitian mentions, research partnerships and employer or insurer listings that describe you accurately. Supplement makers lean on outside proof too, as how supplement brands win AI customers shows.
- Measurement. Ask a fixed set of illustrative category, comparison, privacy and legitimacy questions across ChatGPT, Gemini, Perplexity and Google’s AI features, repeatedly, and track who is named, what is said and which sources are cited. Our guide to designing AI visibility tracking explains how.
No one can promise that an assistant will recommend a particular app or program. What this work does is give assistants an accurate, well-sourced account of your product, so the description they pass on to a would-be user is the right one.
Which questions about health apps and AI remain unanswered?
How many downloads and enrollments AI answers create, and how assistants weigh clinical evidence, are still unknown.
- Health questions are not app choices. Rock Health and OpenAI measure health questions in general. We found no public data on how often people ask an assistant which health app or program to choose.
- Surveys and vendors. Rock Health is a venture fund that invests in digital health, AppTweak and RevenueCat sell to app makers, and surveys rely on what people report.
- Evidence weighting is unmeasured. No study we found tests whether published clinical studies make an assistant more likely to name an app. The Jacques audit looked at cited pages, not app recommendations.
- Connected apps are new. ChatGPT Health launched in January 2026; how often it suggests a connected app, and how it chooses among them, is not public.
Where should a digital health company start?
Start by asking the questions your future users ask, then see which apps assistants name and what they say.
That first check usually shows whether your product is named for the goals you serve, whether assistants describe your regulatory status and privacy practices correctly, which reviews and articles shape the answer and which rivals appear instead. From there, the work is to publish proof that is accurate and compliant, strengthen your store reputation and keep every description consistent.
If your growth depends on trials that become annual subscribers or enrollments in a paid program, talk to us about a review of your app’s AI visibility. We will map where your product appears in AI answers, why competitors are named instead, and which changes are most likely to bring more qualified people to your download and sign-up pages. Our generative engine optimization service page lays out how that work runs for a health company, with claims kept inside your regulatory status and privacy terms written plainly.
Frequently asked questions
Do people use ChatGPT to find health apps?
Many use it for health questions, and some for app choices. Rock Health found 32% of US adults used AI chatbots for health information, and OpenAI lets people connect apps such as MyFitnessPal inside ChatGPT Health.
Can a health app pay to be suggested in ChatGPT Health?
OpenAI does not describe paid placement in Health. It says connected apps need explicit permission, must meet its privacy and security requirements and pass an additional security review.
Should a wellness app talk about clinical accuracy to get recommended?
No. FDA’s revised general wellness policy says wellness products should not claim clinical accuracy, clinical equivalence or medical grade. Describe validation honestly and within your regulatory status.
Does privacy affect whether an assistant recommends an app?
We infer it does when people ask about privacy, because the assistant can only repeat what it finds. The FTC also enforces breach notification for health apps not covered by HIPAA.
Sources
- Fierce Healthcare (2026-03-24), AI chatbot use for health information up 16% from 2024: Rock Health survey (opens in a new tab)
- HIT Consultant (2026-03-23), Rock Health: 32% of Consumers Now Use AI for Health Information (opens in a new tab)
- HIT Consultant (2026-07-13), Rock Health H1 2026 Digital Health Funding Recap (opens in a new tab)
- Drug Store News (2024), IQVIA releases Digital Health Trends 2024 report (opens in a new tab)
- RevenueCat (2026), State of Subscription Apps 2026 (opens in a new tab)
- OpenAI (2026-01), Introducing ChatGPT Health (opens in a new tab)
- AppTweak (2026), AI app discovery survey (opens in a new tab)
- Federal Trade Commission (2024), Complying with FTC’s Health Breach Notification Rule (opens in a new tab)
- Covington & Burling (2026-01-08), FDA Issues Revised Guidance on General Wellness Products (opens in a new tab)
- Apple (2026), App Review Guidelines (opens in a new tab)
- Google Search Central (2025), AI features and your website (opens in a new tab)
- Jacques and colleagues (2026), Authority Signals in AI Cited Health Sources: A Framework for Evaluating Source Credibility in ChatGPT Responses (opens in a new tab), arXiv:2601.17109.
- Underneath (2026), “Is this brand legit?” How AI assistants build a reputation
- Underneath (2026), Ask an AI the same question 5 times: do the brands change?