In 2024, 77% of US adults were offered online access to their medical records and 65% used it at least once, according to the ASTP/ONC analysis of 7,278 respondents to the federal HINTS survey. That is the best independent measure of portal adoption anywhere, and it carries the lesson most portal projects miss: the gap between offering a portal and people using it is where the return is won or lost.
This page pulls together the adoption numbers that come from large, named surveys (healthcare, banking, legal and accounting), the self-service resolution data from Gartner, Salesforce and Zendesk, and the ROI studies vendors commission, with the definition and the sponsor next to each figure.
How many people use a portal when one is offered
There is no single cross-industry survey of client portal adoption. The strongest data sits in regulated sectors where portals are near universal and someone independent counts usage. Read across them, the pattern is consistent: when a portal is offered and tied to a task people already have to do, most of them use it.
| Sector | Measure | Figure | Source and year |
|---|---|---|---|
| Healthcare, US adults | Offered online access to medical records | 77% | ASTP/ONC, HINTS 2024 |
| Healthcare, US adults | Offered and accessed at least once in the past year | 65% | ASTP/ONC, HINTS 2024 |
| Healthcare, US adults | Accessed six or more times in the past year | 34% | ASTP/ONC, HINTS 2024 |
| Banking, US customers | Mobile app is the main way they manage their account | 54% | ABA and Morning Consult, 2025 |
| Banking, US customers | Online banking on a laptop or PC is the main way | 22% | ABA and Morning Consult, 2025 |
| Legal, consumers | Prefer a secure client portal to view or share transactional documents | 60% | Clio Legal Trends Report 2021, via 2Civility |
| Accounting, North American finance teams | Spend most of their engagement time in client portals | 70% | Suralink, vendor-commissioned, 2025 |
Sources: ASTP/ONC Data Brief No. 77 (2025), ABA Banking Journal (2025), 2Civility reporting Clio (2021), Suralink (2025)
The banking figures show where adoption ends up when a portal becomes the default. In 2017, 26% of ABA's respondents named a mobile app as their main banking method. By 2025 it was 54%, and only 9% named a branch. Even baby boomers now favour apps (38%) over laptops (35%).
The professional services figures are weaker. Clio's data is from 2021, and in the same survey 71% preferred to view documents in person and 63% by email, so the portal was one acceptable option rather than the favourite. Suralink sells client collaboration software and does not disclose its sample size, so treat its 70% as directional.
Adoption grows slowly, then quickly
The HINTS series is the only long run of portal adoption data with a consistent method. It shows a decade of slow growth, then a jump after 2020 when the 21st Century Cures Act rules and the pandemic pushed both supply and demand.
| Year | Offered online access | Offered and accessed | Share of those offered who accessed (our calculation) |
|---|---|---|---|
| 2014 | 42% | 25% | 60% |
| 2017 | 52% | 27% | 52% |
| 2020 | 59% | 38% | 64% |
| 2022 | 73% | 57% | 78% |
| 2024 | 77% | 65% | 84% |
Source: ASTP/ONC Data Brief No. 77, HINTS 4 to HINTS 7, 2025
The third column is the useful one for anyone building a portal. Offering access barely changed usage for years: in 2017, only about half of the people offered a portal used it. The share rose to 84% once portals carried things people wanted quickly, such as test results and messaging, and once clinicians started pointing patients to them.
Three other shifts in the same data matter for portal design:
Delegated access doubled. Caregivers who accessed someone else's portal rose from 24% in 2020 to 51% in 2024. In a B2B portal the equivalent is the assistant, the finance team or the agency acting for the account owner, and a portal that assumes one login per client will block them.
Apps overtook the browser. 57% of portal users accessed records through an app in 2024, up from 38% in 2020, and web-only use fell from 60% to 42%.
People have too many portals. 59% of adults had more than one health portal, and only 7% used an app to combine them. Fragmentation is a cost the client pays, not the provider.
Encouragement is the strongest lever measured
The single largest effect in the ONC data is not a feature. It is whether a person was told to use the portal.
Of the people offered access in 2024, 89% said their provider encouraged them to use it, up from 79% in 2020. Among those encouraged, 87% accessed the portal; among those not encouraged, 57% did. Encouraged users also viewed test results (92% against 78%) and clinical notes (82% against 57%) at higher rates.
The same effect appears outside healthcare in a weaker form. Suralink's survey found 82% of accounting clients describe firm requests as unclear or time-consuming, which is a sign of portals used as a document drop rather than a guided process.
What people do once they log in
Portal usage concentrates on a handful of tasks. Everything else is rarely used, however much effort went into building it.
| Task | Share of portal users |
|---|---|
| Viewed test results | 90% |
| Viewed clinical notes | 80% |
| Messaged a provider | 79% |
| Scheduled an appointment | 77% |
| Downloaded information | 32% |
| Added information | 25% |
| Sent information to a third party | 21% |
| Requested a correction | 10% |
Source: ASTP/ONC, Health IT Buzz, HINTS 2024
The top four tasks are read, check status, ask and book. The bottom four are data management. For a client portal in professional services or B2B software, the translation is direct: status, documents, messages and booking first; exports, integrations and editing later, if usage justifies them.
Frequency follows need. 34% of all users logged in six or more times a year, rising to 38% for people with a chronic condition and 54% for those with a recent cancer diagnosis. A portal for a one-off engagement should not expect weekly logins, and a dashboard that measures success by monthly active users will undersell it.
Using self-service is not the same as resolving through it
Customer service data shows the other side of adoption. People try self-service readily; it often fails them.
| Measure | Figure | Source and year |
|---|---|---|
| Customers who use self-service at some point in their service journey | 73% | Gartner, 5,728 customers, reported 2024 |
| Issues fully resolved in self-service | 14% | Gartner, reported 2024 |
| Issues customers called "very simple" that were fully resolved in self-service | 36% | Gartner, reported 2024 |
| Self-service users who could not find relevant content | 43% | Gartner, reported 2024 |
| Self-service users who felt the company did not understand their need | 45% | Gartner, reported 2024 |
| Customers who would rather use self-service for simple issues | 61% | Salesforce, State of the Connected Customer, cited 2024 |
| Customers who resolved issues completely via self-service | 9% | Gartner, 2019 |
Sources: CX Today reporting Gartner (2024), Salesforce (2024), tcworld reporting Gartner (2019)
The resolution rate rose from 9% in 2019 to 14% in 2024, which is progress at a slow pace. The failure reason is not demand: Salesforce's 61% and Gartner's 73% show people want to help themselves. It is findability and fit. The most common failure, at 43%, was not finding relevant content.
This is the gap a client portal can close where a public help centre cannot. A portal knows who is logged in, which project or account they belong to and what stage it is at, so it can show the answer to "where is my thing" without a search. That is the "very simple" category where Gartner found only about a third of issues resolved.
The email side of the comparison is worse. SuperOffice's benchmark of 1,000 companies, last updated in 2023, found 62% did not respond to a customer service email at all, and the average reply took 12 hours and 10 minutes. Only 20% answered fully in the first reply.
Ticket reduction and ROI: what the studies say and who paid for them
Most published portal ROI numbers come from studies the software vendor commissioned. They are useful for scale, not as a forecast.
| Study | Sponsor | Year | Reported result |
|---|---|---|---|
| Gartner Customer Service and Support Leader poll | Independent analyst | 2019 | Live contact $8.01; self-service about $0.10 per contact |
| Forrester Total Economic Impact of Zendesk | Zendesk | 2021 | 15% increase in ticket deflection; 40% lower cost per ticket; 286% ROI over three years |
| Forrester Total Economic Impact of Verint | Verint | 2021 | 45% of calls deflected to less expensive channels; 391% ROI; payback under six months |
| Salesforce State of Service, 6,500 service professionals | Salesforce | 2025 | 30% of cases handled by AI today; 50% expected by 2027 |
| Zendesk CX Trends, about 10,500 respondents | Zendesk | 2025 | 75% of CX leaders expect 80% of interactions resolved without a human within a few years |
Sources: tcworld reporting Gartner (2019), Zendesk (2021 study), Verint (2021 study), Salesforce (2025), Zendesk (2024 release)
Three cautions apply. The Forrester figures model a composite organisation built from interviews with the sponsor's customers. The Salesforce and Zendesk 2025 figures are expectations from service leaders, not measured outcomes. And the Gartner cost figures are leader estimates from 2019.
The cost gap still explains why the business case usually works. As an illustration, if a firm handles 1,000 status and document requests a month by email or phone at Gartner's $8.01, moving even a fifth of them into a portal at $0.10 saves more than $1,500 a month in handling cost, before any effect on retention. The arithmetic only holds if the portal resolves the request. A visit that ends in an email anyway costs both.
Measure resolution, not visits. Salesforce reports that 67% of organisations track case deflection. Zendesk's own support team used a "self-service score", help centre users divided by ticket submitters, and suggested 4:1 as an example ratio; it once reached almost 40:1 but its author warned the metric gave "a lot of benefit of the doubt". A better test for a client portal is the change in inbound emails and calls on the topics the portal covers, measured before and after launch.
What this means for a client portal build
The data points to a short list of decisions that move adoption more than any feature list.
Launch the portal inside a task clients already have, and have a person tell them to use it; that was worth 30 percentage points in the ONC data. Put status, documents, messages and booking on the first screen, because those are what 77% to 90% of users come for. Plan for delegated access from the start, since caregiver access to someone else's portal doubled to 51% in the largest dataset. Make it work on a phone, because app access has overtaken the browser. And set the success metric as fewer inbound requests on covered topics, not logins.
We build applications in Next.js, and when we scope a portal we would start from the client's inbound email and call log, sort it by topic, and build the first release around the three or four topics that make up most of it. Our web apps service page explains how we scope it.
