A title tag, a meta description, and a heading do three different jobs. The title is the label Google often uses for the blue link, and the label a retriever uses for the page. The meta description suggests a snippet. It is not a ranking score. The H1 and the headings under it are the structure a reader, and an answer engine, treat as the answer. When the three disagree, the result is an on-page SEO issue: a title Google rewrites, a weak click-through rate, or an AI citation that names a topic the page does not answer.
This guide follows common SEO and GEO issues that hurt rankings. Use the duplicate content guide when several URLs share one title because they share one intent. Use the thin and outdated content guide when the title is fine and the paragraph is empty or stale. Use this one when each URL should stay, but the title, the description, or the headings are missing, duplicated, or written in the wrong language.
Three fields, three failures
| Field | What it is for | SEO issue | GEO issue |
|---|---|---|---|
| Title tag | Names the page for the query | Missing, duplicated, or stuffed titles get rewritten or ignored | The label does not match the prompt, so another URL is retrieved |
| Meta description | Suggests the snippet | Empty or slogan-like text loses the click; it is not a ranking factor | A stale description can be the sentence a crawler stores |
| Headings | Structure the answer | No heading states the topic | Nothing to lift as the question and the answer |
Title tag issues
A title tag is a relevance hint, not a promise of the text in the result. Google generates the title link from the title element, from headings, or from the words other pages use when they link to you. It rewrites the tag when the element is missing, shared across URLs, too long to show, stuffed with repeats, or a poor match for the query. A rewrite is a reason to edit the tag. It is not, by itself, a penalty.
These are the failures that show up on Hong Kong sites:
- Missing title. The template renders an empty
<title>, or every URL inherits “Home”. The result title is then built from anchors and headings. You do not control it, and it often omits the query. - Duplicate titles. A service page, a blog post, and a district page all say “SEO Hong Kong | Brand”. Search has to guess which URL is the one. If the intent is the same, that is keyword cannibalization. If the intent differs, the titles have to differ too.
- Topic after the brand. “Brand Name | Official Site | Professional Services | SEO” is cut off before the topic. Put the query topic first. Put the brand at the end if it still fits.
- A year or a price that is no longer true. The title is the line most likely to be quoted. An old year in the title is the same class of problem as outdated content.
- The English title left on the Traditional Chinese URL. A Chinese query then matches the English page more easily than the page written for it.
There is no character cutoff that demotes a page. Title links are truncated by width. An English title that states the topic in the first fifty to sixty characters usually survives that cut. Traditional Chinese characters are wider, so fewer of them fit, and the topic still has to be among the characters that show.
Repeating the same phrase three times in the title does not make the page more relevant. It makes the tag look like a list, which is a common reason the result uses different text.
Meta descriptions
A meta description is not a direct ranking factor. Google uses it as the snippet when the text is useful, and writes its own snippet when a sentence on the page matches the query better. The SEO effect is the click. A URL that ranks and is not clicked does not get the visit. A snippet that promises a different page from the one that opens wastes the visit that does arrive.
Treat these as the issues:
- An empty description, so the snippet is a menu label, a cookie line, or a random sentence
- The same description on every URL, including both languages
- A slogan (“your trusted partner”) that never says what the page answers
- A fee, a year, or a product name the body no longer states
- A description that only repeats the title
Write one or two sentences from the opening paragraph: the answer, who it is for, and the detail that makes this URL the one to open. If you would not want an AI answer to quote the description, do not publish it. A crawler that stores the meta description as a summary will store that sentence.
Headings
A page may use more than one H1. Google has treated the count as a non-issue for years. A missing or decorative heading is still an on-page problem, because the page never states its topic where a reader looks. The usual mismatches:
- The title says “title tags and SEO rankings” and the H1 says “Welcome to our insights”
- Every H2 repeats the same keyword instead of naming a sub-question
- The only headings are “Overview”, “Details”, and “Contact”, which nobody searches and no model can quote
- The English H1 was copied onto the Chinese page, or translated into a phrase nobody in Hong Kong types
Use one clear H1 that states the same topic as the title, in the language of the page. Use H2s for the questions the page actually answers. That is also the shape AI platforms cite: a question heading and a direct answer under it. Do not add a second, keyword-only heading for crawlers. The visible heading is the one that gets retrieved.
Why this is a GEO issue
Generative engines match a prompt to a title, a heading, and a passage. A title that says “Services” matches every service prompt and none of them well. The page is retrieved for the wrong reason, or it is not retrieved. Teams read that as an AI visibility gap. It is an on-page labeling gap.
The description and the first paragraph have to agree. If the meta description still names a 2023 price and the paragraph names the current fee, a crawler can store either sentence. Schema markup has the same rule for JSON-LD: the machine-readable line cannot contradict the visible one. The title, the description, and the H1 are the three lines most likely to be stored before the body is.
Language is part of the label. A Traditional Chinese prompt is a poor match for an English title, even when the body was translated. The Chinese URL needs a Chinese title and a Chinese H1 that use the words people type. Those words are not always a word-for-word translation of the English keyword.
A practical order of fixes
- In Search Console, export pages with impressions and a weak click-through rate. Note any URL whose result title is not the title in your HTML. A rewritten title is the first edit, not a reason to add more keywords.
- Crawl for missing titles, duplicate titles, and duplicate descriptions. Split duplicates into same intent and different intent. Same intent goes to the cannibalization guide. Different intent gets a title only that page could wear.
- Rewrite the surviving title with the topic first, in the language of the URL. Keep the brand at the end if it fits. Remove a year or a price you cannot defend.
- Set the H1 to the same topic. Rewrite H2s as the sub-questions, not as repeats of the title.
- Write the meta description from the opening paragraph. If that paragraph is thin or stale, fix it first, using the content guide, then copy the answer into the description.
- Ship English and Traditional Chinese together. Do not leave the Chinese URL with the English title while the body is in Chinese.
- Confirm the
<title>is in the initial HTML, not injected after JavaScript runs. Many AI crawlers do not execute the script. The crawl guide covers empty initial HTML. - Request indexing for the URLs you changed, and check the result title after Google recrawls. Anchors on other pages can keep an old wording in the blue link for a while. The HTML title is still the line you control today.
Questions teams ask
Do meta descriptions affect SEO rankings?
Not as a direct ranking factor. They affect the snippet and the click. A missing or duplicated description gives the result nothing useful to show, so the snippet is pulled from the page, sometimes from a menu or a disclaimer. Write the description so it matches the answer. Do not expect the tag alone to move positions.
Why does Google change my title tag?
Usually because the tag is missing, shared with other URLs, longer than the result can show, stuffed with repeats, or a weak match for the query. Put the topic first and make the title unique. Google may still adjust a title to fit one query. That is different from replacing a tag that never stated the topic.
Will multiple H1 tags hurt SEO rankings?
The count of H1s is not the issue. A page with no heading that states the topic is. Use a clear H1 that matches the title’s subject, and H2s for the sections. Two H1s that name two different topics are an architecture problem, closer to cannibalization than to a tag limit.
Should the English and Chinese pages share one title?
No. Each URL needs a title and an H1 in its own language. A shared English title makes the Chinese page a weak match for Chinese queries and a competing match for English ones. The brand name can appear in both. The topic words should be the words that language uses. Hreflang connects the URLs. It does not translate the title. How those language URLs should be canonicalised is covered in the duplicate content guide.
What to review this month
Export titles and meta descriptions from the templates that generate service pages, articles, and district pages. Mark every missing title, every title used more than once, and every Chinese URL whose title is still English. Rewrite the topic-first title and the H1 together, in both languages, and set the description from the first paragraph only after that paragraph is current.
Title tags, meta descriptions, and headings are an SEO issue with a GEO consequence. The blue link, the snippet, and the label an answer engine stores are the same three lines. If they do not name the topic in the language of the query, the ranking and the citation both go to a page that does.