"We are on page one, so why is nobody calling?" Open Search Console, filter to the page: 4,200 impressions, 38 clicks, average position 6.4. Google shows the page four thousand times a month and people click elsewhere. Then you read the title actually sitting there: "Example Ltd - Welcome - Services - Home".
The decision happens in the search result, not on your page. Two text fields of barely 200 characters decide whether anyone sees your page at all, and they are the only lever that brings clicks without winning a position. This piece explains where Google really takes title and description from, when it overwrites your field, rewrites three real snippets, and ends with the measurement method that will not lie to you.
What a search result shows and where the text really comes from
A Google result has three text parts: a line with favicon, site name and URL path, the clickable title (Google's term: "title link"), and the grey description text (the "snippet").
The title normally comes from the <title> element in your <head>, but not exclusively. Google's documentation lists further sources: the visible main heading, elements such as <h1>, og:title, prominently styled text, anchor text of links to the page, WebSite structured data.
Which names the most common mix-up: the title in the search result is not your H1. The H1 talks to somebody who has arrived, the title to somebody still weighing eight other results. In WordPress, Shopify and Craft these are separate fields: change the heading only and the result stays as it was.
The description comes from one of two places:
<title>Electrician Manchester: 24h callout and rewiring | Harding</title>
<meta name="description" content="Rewiring, fuse board upgrades and fault callouts across Manchester.">
Either from that meta name="description" or from the page copy, and Google decides per query, not per page. The same URL can show your description for one search and three stitched clauses from paragraph four for the next. Not a bug, the default.
How long should a title be? Google counts pixels
Google gives no character limit for either field. For the description the documentation says outright that there is no length limit, the snippet is simply truncated to fit the device width. That is the point: truncation happens by pixel width, not by character count.
Every character figure is a rule of thumb. "Illustration" and "wallpapering" both have twelve characters, but the second needs more room: w, a, p and e are wide letters, i, l and t narrow ones. Capitals cost most of all.
As a working basis, measured off the display and not confirmed by Google:
| Field | Rule of thumb, desktop | Rule of thumb, mobile |
|---|---|---|
| Title | around 55 to 60 characters | narrower column, less fits |
| Description | around 150 to 160 characters | around 110 to 120 characters |
Since most searches happen on a phone, the smaller number matters more. Hence the only length rule that always holds: most important thing first. Everything past character 110 is a bonus, so a promise sitting at the end is up for deletion. Too short is a state too: 45 characters give away space a competitor uses.
Why Google shows a different title than the one you wrote
It happens often and feels arbitrary. It is not. Google documents which titles it replaces:
- Half-empty titles such as "| Example Ltd".
- Inaccurate titles whose words are not backed up by headings and body copy.
- Obsolete titles like "Price list 2023" on a page showing current prices.
- The same boilerplate across forty pages, distinguished by one word. The case that undoes templates.
- No recognisable main title: three headings of equal visual weight.
- Language mismatch and a site name said twice, when it already sits above the title.
Two more triggers: stacked keywords ("Electrician Manchester, electrical contractor Manchester, electrical installation Manchester") and a length that leaves nothing standing.
The more useful insight is in the consequence. A rewrite is not a bug and not a penalty, it is a hint: your title failed its job of describing this page for this query. There is no switch that forces Google and no meta tag against it. Fix the cause and your own title usually comes back. Check it in the result itself: Search Console has no report saying "title was replaced".
Is the meta description a ranking factor?
No. Google does not use the content of meta name="description" as a ranking signal, and has said so for years. It is a click factor: no positions gained, more people collected at the position you hold.
And now the part the usual guides leave out: a hand-written description is not equally worth it everywhere. It depends on how many queries land on the page.
Pages with one dominant query are the candidates. The service page of an electrical firm gets most impressions from variants of "electrician manchester", so you know what somebody wants and the hand-written line wins nearly every time.
Pages with a hundred different entry queries are the others. A long guide collects three hundred queries in a month, none worth more than three percent. There the extract Google assembles per query is often better than anything fixed, because it adapts and your sentence cannot. Write one as a safety net, keep it broad, move on.
What the description cannot do is make your result bigger. That takes structured data: review stars, price, availability and a breadcrumb line instead of a bare URL path take extra lines and make your result more noticeable than a competitor's on the same spot. See structured data.
Three snippets rewritten: before, after, why
Case 1, service page of an electrical firm.
before 72 chars Harding Electrical Ltd - Welcome to the services page of our family firm
after 58 chars Electrician Manchester: 24h callout and rewiring | Harding
before 101 chars Welcome to the website of Harding Electrical Ltd. We are pleased
about your interest in our services.
after 142 chars Rewiring, fuse board upgrades and 24 hour fault callouts across
Manchester and Stockport. Family firm since 1998, fixed quote
within 48 hours.
The before holds nothing anybody searches for: not the trade, not the town. The after leads with the typed query and makes a checkable promise, the same mechanics as local visibility.
Case 2, guide article on a decorator's site.
before 63 chars Mould in the bedroom - News from Ahrens Painters and Decorators
after 54 chars Mould in the bedroom: 5 causes and what actually helps
before 0 chars (no field set, Google takes the first paragraph)
after 119 chars How to tell condensation from a leak, the five causes we find
most often, and when opening a window stops being enough.
The old title gives 43 of its 63 characters to a section label and a company name nobody searches for. The after swaps both for a number and a promise, and the description picks up the real question: where is the damp coming from.
Case 3, category page in a sports shop.
before 62 chars Products - Category: Running shoes - Page 1 - Northwind Sports
after 51 chars Road running shoes: 240 models, free 60 day returns
"Products", "Category" and "Page 1" are system vocabulary, not search terms. The after narrows the category the way people search it (road, not trail) and answers the two questions before the click with numbers.
What pulls in all three cases: the query in the word order it gets typed, a number or a concrete promise, a local reference on local services, a reason to hurry only when it is real. A 24 hour callout is real, "today only" on a page untouched since 2019 is not. A year works on anything that can go stale, but commits you to keeping it current.
What hurts: superlatives with no evidence ("the best", "leading"), "Welcome to", exclamation marks, capitals, stacked keywords, and the plugin suffix eating the same 40-odd characters on every page.
Title templates and how to find duplicate titles
Almost every system builds titles from a template, usually %%title%% %%sep%% %%sitename%%. The only sensible question: how many pages hang off it, and do they share one job?
For shop filters and category pages the template is right. Nobody hand-writes five hundred filter combinations, and Road running shoes: {count} models, free 60 day returns does automatically what the line above did by hand.
For service pages it is wrong. A trade business has eight to twelve, each with its own dominant query. If all eight read "Service X | Firm Y - Your specialist in Manchester", Google sees near-identical titles and rewrites them, and you lose the space where the difference belongs. Eight titles by hand take an hour, the cheapest hour in search optimisation.
Finding duplicates got harder once the old "HTML Improvements" report vanished from Search Console. Two ways that work:
- Crawl. A desktop crawler such as Screaming Frog lists titles and descriptions with character counts; the free version covers small sites. Sort by the title column and duplicates line up in blocks.
- The
site:search.site:yourdomain.com "Your specialist in Manchester"shows how many indexed pages carry that suffix. The count is an estimate, but it settles "three or three hundred".
Traffalyzer records title and description of every crawled page with their lengths, flagging missing or overlong fields.
Measuring click-through without fooling yourself
The mistake is always the same: clicks before against clicks after. Raw clicks are useless: they hang on position and search volume, both of which move without your help.
Compare click-through rate at the same average position. Performance report, filter to the one URL, switch on the comparison period, show all four columns. If average position moved more than half a place, the comparison is void: visibility changed, not your snippet. Cleaner still: filter to the dominant query too.
The arithmetic: 4,200 impressions at a 0.9 percent click-through rate is 38 visitors a month, the same page at 2.4 percent brings 101. That is 63 extra visitors with no position gained, for half an hour on two fields. Your site-wide rate can still fall while you do everything right, because more result pages now open with an AI Overview.
Two tests of patience remain. Google must recrawl the page and names no deadline; in practice, days to a few weeks. Then you need impressions: at 200 a month every difference is noise. Which pages are worth the effort is a question for impressions but no clicks. If click-through is already healthy, the next lever is not the snippet but positions four to fifteen.
Where to start
- Search Console, last three months, the Pages view, sorted by impressions. Note the ten pages with many impressions and a rate under two percent.
- For each, check which query dominates. That wording is your raw material.
- Search for that query and look at your result, on phone and desktop. Is your title showing, or another one? If another, work the trigger list above.
- Rewrite the title, under 60 characters, query first, company name last or gone. Rewrite the description with the important part in the first 110 characters.
- Note the date, then check rate and position against the comparison period after four to six weeks.
Ten pages, one afternoon. Nowhere is the gap between effort and visitors shorter.
Frequently asked questions
How long should a title tag be?
Google gives no character limit, because truncation happens by pixel width. On desktop 55 to 60 characters is the working rule, with wide letters and capitals cut sooner. Writing longer is pointless: what falls off the end goes unread.
How many characters should a meta description be?
Around 150 to 160 characters on desktop, closer to 110 to 120 on a phone, so everything decisive belongs in the first 110. There is no official limit: per Google's documentation it may be any length, truncated to the device width.
Why does Google show a different title than mine?
Because in Google's view your title does not describe the page well for that query. Documented cases: half-empty titles made of the company name, obsolete or inaccurate titles, the same boilerplate across many pages, no recognisable main title, mismatched languages, a duplicated site name. Plus stacked keywords and excess length. Fix the cause and your title usually returns.
Is the meta description a ranking factor?
No. Google does not use the description text to decide positions. It only affects whether somebody clicks, and so your visitor count at an unchanged position. No reason to ignore it, just a reason not to expect a better position.
Do I need a unique description for every page?
If you set one at all, then yes: identical descriptions across many pages are worse than none, because Google discards them anyway.
The effort is not equal everywhere: with one dominant query the hand-written sentence usually wins, with hundreds of entry queries the assembled extract is often better.
What is the difference between the title and the H1?
The title sits in the head of the HTML document and shows in the result list and the browser tab; the H1 is the visible main heading on the page. They may match but need not, and Google may fall back to the H1 if your title field gives nothing.
Should the company name go in the title?
For known brands yes, it earns clicks. For everybody else it belongs at the end or nowhere: it costs characters nobody searches for, and Google shows the site name above the title anyway. The exception: the name is part of the query.
What happens if I set no meta description?
Google assembles one from sentences on the page, picked to match the query. Not an error state, and no position lost. It is unpredictable: sometimes the extract lands perfectly, sometimes mid-clause or on your cookie banner.
How quickly does Google pick up a title change?
Only after the next crawl, and Google names no deadline: days on well-visited pages, weeks on rarely visited ones. URL Inspection and "Request indexing" nudge a crawl, not force one. The measurement then needs four to six weeks.