Search visibility issues: a field map for SEO and GEO problems

Name the search visibility issue before you pick a fix

Search visibility is the chance that the right person, or the right answer engine, can reach a page you meant to be found. A search visibility issue is the reason that chance fell: the URL is hard to crawl, the page answers the wrong query, or the passage is too vague to cite. Teams often report one number — fewer impressions, fewer clicks, or a brand missing from an AI answer — and then treat every SEO problem and every GEO issue as the same job.

Separate the systems before you open a tool. Classic search needs a fetchable canonical URL that matches a real query. Generative engine optimization needs a bounded answer an AI system can reuse, plus a source trail that says who wrote it and where it applies. Hong Kong adds a third layer: district, language, and the difference between a neighbourhood search and a citywide one. A page can fail in one layer while the other two still look healthy.

Write down the URL, the query, and the week the change appeared. Then ask three questions. Can a crawler fetch the canonical URL? Does that URL answer one query more clearly than the pages beside it? Can an answer engine quote a short passage and tell which organisation it belongs to? The first two questions are SEO issues. The third is a GEO issue. The issue index on this site gives each of those failures its own page, because the query a team types is already that specific.

• Record the URL, the query, and the date visibility changed.
• Keep indexing, ranking, and AI citations as three separate readings.
• Choose one issue note once you can name the failure.

SEO problems that take a page off the route

The SEO problems that remove search visibility are usually mechanical. A canonical tag points at a different URL. A noindex tag shipped on a template that should rank. The XML sitemap is full of filters, while the service page has no ordinary internal link. Google has discovered the URL and has not fetched it yet. Two pages target the same query and split the signal. The page is thin beside a competitor who answers the follow-up question.

These are search visibility issues because the page never reaches the result, or it reaches it too weakly to earn the click. Open the page indexing report and group the sample URLs by reason and by template. Coverage exclusions, discovered-but-not-indexed URLs, crawled-but-not-indexed URLs, a canonical Google replaced with a different one, an alternate Google agreed to fold, duplicates with no user-selected canonical, pages that redirect, URLs indexed though blocked by robots.txt, keyword cannibalization, thin content, and broken internal routes are different SEO issues with different fixes. A rewritten title will not help a URL that search has not fetched. A crawl request will not help a URL Google has already fetched and declined to index. When the report says Google chose a different canonical than the user, the tag did not win: the answer has to live on the URL search selected, or the two pages have to become different enough to stay apart. ‘Alternate page with proper canonical tag’ is the matching case: Google agreed, and the alternate stays out on purpose. That status is a search visibility issue only when the URL you still want found, or cited, is the one that was folded. ‘Duplicate without user-selected canonical’ is the case with no declared preference. Google still folded the copies, and it chose the URL that can rank and be cited. Name that URL yourself. ‘Page with redirect’ is a different exclusion. Google fetched the URL and was sent somewhere else, so the address itself stays out of the index. A retired URL should stay in that row. A guide, a service page, or a language version you still want found, or cited, should return 200 on the URL you publish. ‘Indexed, though blocked by robots.txt’ is a warning, not an exclusion. Google kept the URL without fetching it, usually because another page linked to it, so the result is a thin snippet. A page you still want found, or cited, has to be allowed in robots.txt before the HTML can be read. A page you want gone needs that block removed and a noindex, because a disallow does not take an indexed URL out of search.

On a Hong Kong site, add language and place to that check. An English page and a Traditional Chinese page that repeat the same intent without a clear hreflang relationship compete with each other. A district landing page that only swaps the place name is a thin SEO problem, even when the head office page ranks. Confirm the live status code, the canonical, and the robots meta on the URL you care about, then give it one internal link from a page Google already crawls.

• Read Search Console by reason, then by template.
• Leave filters, parameters, and thank-you pages excluded on purpose.
• Request a crawl only after the canonical URL is linked and indexable.

GEO issues that take a page out of the answer

GEO issues show up when the page still ranks and the citation disappears. The passage is too broad to quote. The organisation, the person, and the place are named differently on every URL. An AI crawler is disallowed in robots.txt, so the answer engine keeps an older copy or a competitor’s. The fact is stale, so another source is safer to reuse. The page explains a topic and never states the boundary: who it is for, where it applies, and what it does not cover.

Generative engine optimization problems still depend on the SEO route. A blocked, duplicated, or unlinked URL is a poor source, however polished the prose looks on screen. Once the canonical page is fetchable, write the answer as a short unit under a heading that matches the question people and prompts actually ask. Put the subject and the location next to the claim when Hong Kong context changes the answer. That is the part an answer engine can lift without inventing the missing context.

Common GEO issues in this library include citation-ready passages, entity clarity, AI crawler access, AI Overview citation loss, and the difference between GEO and SEO. Use them when the ranking holds and the answer does not. Use the SEO notes when the URL itself is missing. Search visibility returns when the route, the answer, and the source trail describe the same page.

Keep one URL aimed at one issue keyword

People search for the failure, not for a category. The queries look like canonical tag mistakes, thin content, Google AI Overviews, local pack ranking, or AI crawler access. A single overview can orient a team. It cannot rank for every SEO problem and every GEO issue at once, and it should not try. This page is the map. Each issue note is the page that matches one of those keywords and tells you what to check.

After you change the site, measure the same URL and the same query. For SEO issues, look at indexing and the query in Search Console. For GEO issues, look at whether the answer still quotes you, and whether the crawler you care about can fetch the canonical HTML. One confirmed repair is more useful than a checklist that touches every template on the site.

• Match the query to one issue page, then do that page’s check.
• Link the repaired URL from a page that is already crawled.
• Re-read the same query after the change, in search and in the answer.