You can’t search for a benefit you don’t know exists. What’s missing is a way to deliver benefits to people, not just a bigger list of them.
1. The short answer: many people never start searching
City and county governments run programs that can really help with rent, a drop in income, child care, caring for a parent, home repairs, disasters, and more. But when someone first hears about a useful program, the usual reaction is: “Wait, that exists?” Once they know, they often open the official page right away, check whether they qualify, and look for the application.
This matters. People did not skip the program because it was useless. They never got to the point of searching for it.
SEO (search engine optimization, the work of making pages easier to find on Google) helps only after someone has typed a search. It can’t fix the earlier problem: if you have no idea a program exists, you have no name to type into Google.
A search box is not telepathic.
So the real problem is discovery. Public services should reach people through everyday situations they recognize, instead of expecting residents to learn government vocabulary first.
2. “I didn’t know, so I didn’t search” is a serious barrier
Asking someone to search for the official name of an unknown benefit is like asking a player at the start of a game to type the secret boss’s name into a walkthrough.
People think in situations:
- I lost my job.
- My income dropped.
- Rent is getting hard to pay.
- I moved.
- I got married.
- We had a baby.
- A parent now needs care.
- I need to repair my home.
- A disaster disrupted my life.
Governments need to organize information by program name, department, local rule, and eligibility group. Residents organize their lives by events and problems. If those two structures do not connect, “search our website” is a weak instruction because the user still lacks the query.
3. Where people drop out, step by step
Before getting help, a person goes through at least these steps.
| Stage | What the person must do | Typical failure |
|---|---|---|
| 1 | Experience a problem or life event | No link to government support yet |
| 2 | Suspect public support may exist | Cannot imagine that a program exists |
| 3 | Form a query | Everyday language differs from official terminology |
| 4 | Reach authoritative information | Titles, navigation, or structure may be weak |
| 5 | Judge likely eligibility | Income, age, household, and deadline rules are complex |
| 6 | Apply | Documents, identity checks, forms, appointments, and deadlines create friction |
| 7 | Complete or renew | People forget or abandon the process |
SEO mainly operates around stages 3 and 4. Someone stuck at stage 2 has not entered the SEO contest at all.
4. Many people do not actively search municipal services
In a July 2025 Osaka Prefecture online survey of 1,000 adult residents, 33.6% said they never looked up their municipality’s budgets, projects, or administrative services through newsletters or websites, and 31.1% said they rarely did so—a combined 64.7%. Frequently used municipal information channels included printed newsletters at 54.4%, municipal websites at 38.9%, and social media at 10.0%.[R1]
Osaka Prefecture explicitly warns that the respondents came from a private internet panel and should not be treated as a perfect miniature of the entire prefectural population. The 64.7% is therefore not a national prevalence estimate. It is still useful evidence that active searching is far from automatic.
A separate private web survey of 446 adults across Japan found that 57.8% had never independently searched for municipal services beyond information the municipality sent them. Among those who did search, 84.6% used the web. In the same survey, 17.3% said they had missed a service because they did not know it existed.[R2]
Again, this is not a statistically representative estimate for all of Japan. But the mechanism is clear: the web works well for people who search. The weak point is turning a non-searcher into a searcher.
5. Government language and life language do not line up
Administrations need program names, responsible departments, rules, and categories. Residents need answers to “I can’t make rent,” “I moved,” or “we had a baby.”
The private survey reported complaints such as cluttered websites, difficulty finding relevant information, uncertainty about whether the respondent met the conditions, difficult administrative wording, and the inability to search national, prefectural, and municipal programs together.[R2]
The answer is not to delete useful detail. It is to change the access route.
Instead of leading with an official benefit name, start with “Did your income fall and is rent becoming difficult?” Instead of “Eligibility for Program X,” start with “For people who moved, married, or renovated a home.” The official name can appear after the person recognizes the situation.
6. Are government websites bad at SEO? Partly, but that is not the whole story
There is real room to improve public-sector web structure. A.A.O.’s 2026 municipal website quality survey analyzes public pages from an accessibility and site-quality perspective.[R12]
But it is not a Google ranking study. It would be wrong to convert that evidence into “municipal sites do not appear in search.”
Google itself recommends descriptive, concise page titles and clear main headings.[R10] Weak titles and structures can therefore hurt a result’s clarity and discoverability, so SEO fundamentals still matter.
Another common shortcut—“government uses PDFs, so Google cannot see them”—is also wrong. Google explicitly lists PDF as an indexable file format.[R11] PDFs can still be poor on mobile, hard to navigate, and awkward to update, but “PDF” does not mean “invisible to search.”
SEO deserves repair. It simply cannot reach a person who never searched.
7. Awareness is not the finish line: applying creates a second barrier
Telling people about a program does not guarantee take-up.
The OECD’s 2024 review groups major barriers to social-program take-up into four categories: insufficient or complex information, application hassle, stigma, and low expected benefit.[R5] In a 27-country survey of 27,000 respondents cited in the report, 51% believed applying for public benefits would be difficult and lengthy.[R5]
So there are at least two cliffs: “I did not know” and “I know, but applying is painful.”
Fixing discovery while leaving the application process as the final boss still produces drop-off.
8. Experiments show why information plus assistance can beat information alone
A randomized experiment involving about 30,000 older Americans who were likely eligible for SNAP but not enrolled split participants into a control group, an information-only group, and an information-plus-application-assistance group. After nine months, enrollment was 6%, 11%, and 18%, respectively.[R6]
The mechanism is hard to miss. Information moves some people. Assistance moves more of them from awareness into completion.
But better messaging alone is not magic. Six preregistered EITC field experiments with one million observations varied message content, design, messenger, and delivery mode and found no evidence that these low-touch nudges increased tax filing or credit claiming.[R7]
By contrast, another randomized study of roughly 65,000 SNAP applicants found that a more flexible interview process increased approvals by six percentage points.[R8]
The product lesson is not “write a cleverer notification.” It is “reduce friction across discovery, eligibility, and application as one journey.”
9. Japan also has an application-first delivery problem
Research by Reiko Arami on Japanese COVID-era support describes how application-based administration and difficulty identifying target populations can create unequal service delivery. People in need may find access difficult, abandon claims, and effectively disappear from the population actually reached by policy.[R9]
Application requirements are not inherently irrational; programs may need consent, documentation, and case-specific checks. The problem is when the whole system assumes that eligible people can first discover the program, locate the right authority, interpret the rules, and submit everything before the deadline.
Information-search skill then becomes an unofficial eligibility filter.
10. A practical fix: stop asking people for the program name
A discovery interface can begin with only a few things:
- municipality
- age range
- broad household type
- housing situation
- recent life event
Recent events should be expressed in ordinary language: job loss, income drop, move, marriage, pregnancy or birth, schooling, caregiving, home renovation, disaster.
The result page should not be a wall of program names. It should show likely relevance, the condition that seems to match, benefit amount or service, deadline, required documents, online availability, official update date, and official application link.
Now the user can stop playing “please type the exact name of the unknown program.”
11. Then use life events to push only relevant programs
The next step is not to make people browse the database every day. Trigger discovery when programs are most likely to matter.
A move, job loss, income drop, marriage, birth, school transition, caregiving event, home renovation, or disaster can trigger a small set of relevant programs.
Do not push every municipal announcement. That turns the notification tray into a graveyard of public-relations material.
Push only high-relevance events: a newly relevant program, an approaching deadline, or a material eligibility change.
The ideal path is:
life event → candidate programs → a few eligibility questions → documents → application → deadline reminder
12. SEO still matters, but aim at everyday problems
A discovery service still needs SEO. The difference is that it should not rely only on people who already know official program names.
Useful entry queries are problem-shaped: “money after losing a job,” “moving grants,” “rent support for living alone,” “marriage and relocation support,” “home-renovation subsidies,” or “support when caring for a parent.”
The job of SEO becomes translating existing life-problem demand into public-service discovery.
13. From a list of articles to a “you may qualify for these” service
At this point the product is no longer merely a collection of articles about municipal programs.
Amazon says, “You may also want this.” YouTube says, “You will probably watch this next.” Public services should not be the one category that says, “Please tour government sites and PDFs nationwide and discover your own eligibility.”
The warehouses already exist. The weak layer is distribution.
An article can be the entry point, but it should not be the end. After answering one question, the system can say, “Based on these conditions, three other programs may also be relevant.” That turns content into a discovery engine.
14. Japan is already building some of the required infrastructure
Japan’s Digital Agency operates the Child-Rearing Support System Registry, which structures national and municipal child-rearing program data so apps can deliver relevant information at appropriate times. API access for app providers began in November 2025, the parenting app Piyolog began integration in January 2026, and expansion is planned from fiscal 2026 onward.[R3]
The Digital Agency’s user-centered public-service guide also acknowledges that provider-centered government IT has produced services perceived as harder to use and understand than private services, contributing in some cases to weak usage, satisfaction, and policy outcomes.[R4]
As of September 2026, MyNaPortal APIs can, with user consent, allow external services to obtain categories of personal administrative information including tax and income, household, and child-rearing data.[R13][R14] A separate electronic-application API supports various local-government procedures, including child-rearing and moving-related applications.[R15]
That creates a future path such as:
consented attributes and events → program matching → likely-eligibility display → prefilled data → application
These APIs are not open toys. Usage requires application, review, coordination, connection testing, and other steps.[R14]
15. Safety rules: never guess who qualifies
A useful system must avoid becoming a confident misinformation machine.
- Before official confirmation, say “likely eligible” or “needs confirmation,” not “you definitely qualify.”
- Show the official source, update date, jurisdiction, and deadline for each program.
- Track rule changes instead of judging from stale articles.
- Ask for the minimum personal data necessary at each step.
- Explain the purpose and scope of any consent-based administrative-data access.
- Keep phone, in-person, and human-support routes for people who cannot complete digital flows.
- Optimize notifications for relevance and deadlines, not volume.
Automation should not remove people from public service. It should automate easy discovery so human help can remain available for complicated cases. The OECD makes a similar point: digitalization should be accompanied by live support for people who struggle with electronic access.[R5]
16. Conclusion: the bigger gap is how benefits are found, not just SEO
If we explain invisible municipal services only as “bad government SEO,” we miss everyone who disappeared before the first query.
The hard design is asking people to discover an unknown program, know what to search, interpret the conditions, and complete the paperwork. A better route is: “a life problem occurs → relevant programs appear → a few questions estimate eligibility → the person can apply.”
SEO remains one entry channel. It should not be the eligibility test for finding help.
People need to encounter relevant support even when they do not know what they do not know.
Publishing a program is not the same as delivering it.
The next competition in public services is not only search ranking. It is distribution.
