More than half of older adults in China are online. Why can registration, payment, authentication, refunds and help still become barriers?
Penetration measures entry to the internet. It does not measure whether a person can finish a consequential, multi-step task. The right unit of analysis is the complete service journey, not the device or the first screen.
Separate access from effective use.
CNNIC's 57th Statistical Report states that by December 2025 China's overall internet penetration had reached 80.1%, while internet use among people aged 60 and above stood at 53.7%. The 26.4 percentage-point difference is BEIIU's calculation from figures published in the same report. CNNIC's 56th report had recorded 161 million older internet users and a 52.0% penetration rate in June 2025. The increase of 1.7 percentage points in six months is material, but connectivity alone says little about the quality of task completion.
A denominator matters. The 69.8% figure does not mean that 69.8% of all older adults shop online. It shows that older people who are already connected can be active digital consumers. The market therefore contains both people who remain offline and connected users who still face failure in complex or high-risk journeys.
Policy has moved from providing an entrance to improving the quality of use.
Traditional and intelligent services in parallel
The State Council required essential offline channels to remain available and called for differentiated support by age, education, living environment and habits. The objective was not forced digitisation, but the ability to use, understand, trust and willingly adopt services.
Principles became interface and evaluation standards
MIIT issued universal design and assessment specifications for age-friendly websites and applications, covering typography, spacing, controls, prompts and compatibility. These are a baseline, not a substitute for observing complete real-world tasks.
From availability to quality
MIIT's quality programme emphasised standards, service quality and user experience. It also identified superficial remediation: a compliant home screen paired with inaccessible secondary screens, or a care mode too difficult to find. Age-friendly design cannot be a visual skin.
Accessibility became continuing governance
China's Barrier-Free Environment Creation Law brought information exchange and social services into a legal framework. Convenience, safety, practicality and broad benefit must be considered together when judging a digital service.
The digital divide is not one gap. It is a chain of four possible failures.
A person may be highly proficient in messaging and short video yet lose control during hospital authentication, an insurance claim or a refund. Segmenting people into “digitally able” and “unable” by age conceals where the service actually fails.
Access gap
Availability of a device, network, account, identity method and payment instrument. This determines whether a person can enter.
Test: weak signal, older devices, low battery, lost account and verification failureCompletion gap
Whether the user can travel from intention to result, rather than merely view a large-font home screen. Cross-page and cross-app transitions are common abandonment points.
Test: first-pass completion, steps, time, failure points and retriesTrust and safety gap
Whether the user understands price, consent, renewal, data use and warnings. Excess complexity causes fear; excessive simplification can conceal consequences.
Test: comprehension, mis-taps, consent withdrawal and incident responseOutcome and agency gap
Whether a completed click produces the intended care, journey or service, and whether the user can cancel, refund, appeal, reach a person and regain control.
Test: outcome delivery, recovery success and time to human supportConfigure support by task autonomy, not by chronological age alone.
An 82-year-old may buy rail tickets independently, while a 61-year-old may avoid mobile payment because of vision, cognitive load or fear of fraud. Effective segmentation starts with capability and confidence in the task, plus the support available at that moment.
Five everyday journeys reveal whether digital inclusion is real.
Healthcare
The journey is not “open the booking page”. It includes identity, department selection, appointment, payment, rescheduling, check-in, results and follow-up. Re-authentication or an app switch can break the chain.
Payment and consumption
Larger controls are insufficient. Total price, service scope, auto-renewal, refund conditions and payee must be understood before confirmation. High-risk actions need proportionate friction; low-risk actions should remain efficient.
Transport and travel
Tickets, ride-hailing and navigation must withstand weak signal, low battery, station changes and location error. A usable service preserves proof, direction and access to staff under degraded conditions.
Public services
Document requirements, identity checks and case status should use plain language. Digital cannot be the sole route, and proxy authorisation cannot become permanent transfer of account control.
Home and community services
An alert has value only when it is clear who receives it, who judges it, who acts and how it closes. Notification overload, false positives and ambiguous responsibility can create the illusion of safety without accountability.
Reluctance can be a rational response to material risk.
Cases published by China's national ageing authorities show how concrete those risks can be. One case involved about 1,500 phone numbers and verification codes obtained through small gifts. An online training fraud caused losses of RMB 13.09 million across 2,495 victims. Another malicious auto-charging case involving older mobile-phone users exceeded RMB 4.6 million. The collection also includes the use of AI-cloned voices to impersonate relatives.
Digital inclusion therefore cannot be reduced to producing more clicks. It must reduce information asymmetry, constrain default consent, make consequential decisions verifiable and reversible, and connect a worried user to a person quickly. Safety is not the enemy of conversion; it is a precondition for durable use and brand trust.
Move beyond “senior mode” to services that can be completed, understood and recovered.
- 01
Dual-channel service is architecture, not a patch
Online and offline, automated and human channels should share status and outcomes. Switching channels should not require the user to restart or retell the entire story.
- 02
Present one core decision at a time
Reduce competing information. Show result, price and consequence first, then reveal detail in decision order. Clarity means sequencing essential information, not deleting it.
- 03
Make high-risk steps reversible
Payment, authorisation, renewal and sensitive-data submission need a clear receipt, a withdrawal window and an incident path. A success animation is not proof of a real outcome.
- 04
Bound family collaboration
Proxy rights should be limited by task, amount and time, show who did what and remain visible and revocable to the older person. Help must not erase agency.
- 05
Collect only what completes the task
Explain what is collected, why and for how long in plain language. Provide reasonable authentication alternatives so that a single biometric method does not become a new exclusion.
- 06
Every automated journey needs a human path
Human help should appear at the point of difficulty, not at the end of a deep menu. Support staff should inherit the journey context rather than ask users to prove and explain everything again.
Judge age-friendly work by outcomes, not by the number of screens changed.
Share of users reaching the intended outcome without another person taking control.
Share finishing without duplicate submission or restarting the journey.
Share able to continue or reverse after a mis-tap, timeout or document error.
Time from an explicit help request to a person who can see the prior context.
Drop-off caused by authentication, payment, redirection or lack of comprehension.
Completion under clear, informed and bounded proxy authority.
Whether the person can later repeat the same task successfully.
Whether care mode covers detail, payment, settings and exception states.
Results should be segmented by task complexity, visual or motor ability, digital confidence, rural or urban context and availability of support. A single average satisfaction score for “older users” can hide the people most affected by failure.
Five apparently friendly choices that can preserve exclusion.
- 01
Changing font size alone. Larger type matters, but it cannot repair app switching, failed identity, inaccessible refunds or ambiguous responsibility.
- 02
Activating “elderly mode” by age. It can patronise confident users while failing to support people with real visual or cognitive needs.
- 03
Transferring control to adult children by default. Family collaboration requires consent; care is not a licence to replace every decision.
- 04
Treating one identity method as universal. Fingerprint, face and SMS can each fail under particular bodily, device or environmental conditions. Alternatives are essential.
- 05
Treating offline service as obsolete cost. For high-risk or infrequent tasks, staff and physical service are trust infrastructure and the system's fault-tolerance layer.
BEIIU view: genuine digital inclusion preserves the older person's authority.
BEIIU believes the next phase of silver-economy service will not be won by adding more features to a “senior mode”. It will be won by services that help people with different abilities complete necessary tasks, understand what is happening, know who is accountable and regain control when they are uncertain or something fails.
Every service for an ageing family should pass four tests: Can the person complete it? Understand it? Control it? Recover from failure? Only when all four answers are yes does technology extend capability instead of creating a new threshold.
Sources and methodology
This paper prioritises public material from government bodies, statutory institutions and China's national internet research authority. The 26.4 percentage-point gap and 1.7 percentage-point half-year increase are BEIIU calculations from published like-for-like figures. Cases identify risk mechanisms and do not indicate prevalence. The design and measurement framework is BEIIU analysis, not a product-performance claim.
- CNNIC: 57th Statistical Report on China's Internet Development
- CNNIC: 56th Statistical Report on China's Internet Development
- CNNIC: Digital Consumption Development Report (2024)
- State Council: Implementation plan for addressing older adults' difficulties with intelligent technology
- MIIT: Universal design specifications and evaluation system for age-friendly websites and apps
- MIIT: Work plan for high-quality age-friendly digital technology
- Law of the People's Republic of China on Barrier-Free Environment Creation
- China National Committee on Ageing: 2024 typical cases on protection of older people's rights