Draft click-focused descriptions for a real page
Start with a primary keyword, brand, page type, and tone, then review five editable descriptions instead of publishing the first generated sentence.
Meta Description Generator
Free, no signup. Draft five meta descriptions with AI, then compare their measured widths with approximate desktop and mobile SERP-preview guidance. Every variant remains editable and includes a matching title draft.
The AI drafts; you approve. It can overstate what a page offers, so verify every variant. The meters update live, and editable templates remain available when AI generation is unavailable.
We fetch the page and extract its text in your browser, then fill the Paste box. JS-only pages may extract little — paste instead.
Bulk mode uses titles-only, limited context: the optional topic is capped at 300 characters per row. Use single Paste/Fetch mode for up to 6,000 characters of page context.
Your input is sent to the AI model to generate a result — it is not stored afterward. Anonymous run-level outcome counters may be used for aggregate research; URLs, domains, IPs, and identifiers are never included, and no statistic is released below 100 runs.
Sample report
Say your keyword is meta description generator. You generate options and pick this one;
the tool returns the copy-ready tags:
<title>Meta Description Generator - Free Pixel Checker</title>
<meta name="description" content="Generate meta descriptions with AI, compare measured widths with approximate SERP-preview guidance, score each option, and copy your preferred draft.">…and each option is scored like this (illustrative output — your AI-drafted wording will differ):
Illustrative example — AI-drafted wording varies per run; the pixel meters use this tool's real canvas-measurement engine, the same one behind the SERP Snippet & Truncation Checker
Copy
154 chars ▮ 883/920px desktop ▯ 712/680px mobile ✓ keyword ✓ call to action
- The preview bands can disagree — this option fits the 920px desktop preview default but exceeds the tool's 680px mobile preview default. That is an editing cue, not a prediction that Google will truncate the same text.
- Keyword and call-to-action are checked independently of length — this option passes both, so the only fix needed is trimming (or reordering) the tail to clear mobile too.
- Compare the truncation point itself against a real SERP with the SERP Snippet & Truncation Checker.
How to use it
- Type your primary keyword — the phrase the page should rank for. Add a brand / site name if you want it woven in. These two fields feed every mode.
- Pick a mode. Quick works from a one-line topic; Paste content and
Fetch a URL let the AI write from the page's real text (recommended, so it doesn't
guess); Bulk takes up to 10 rows of
keyword | brand | topicat once. - Press Generate descriptions. You get five description options plus five matching title tags, each scored for length in pixels, keyword presence, and a call to action.
- Toggle Desktop / Mobile in the live SERP preview to compare each draft with the configured preview widths. Edit any option inline — the approximate width meters and preview update as you type.
- Mark the description and title you want with Use, then Copy tags
for a ready-to-paste
<meta>+<title>block, or export everything as CSV.
Site passport Local context for this saved site
Local data
Saved targets, named lists, and recent check summaries remain only in this browser.
Reflects the two variants marked “Use”. Select a different description or title below to swap.
Meta descriptions
Matching title tags
Rate this tool
Before
Meta description generator. The best free tool for SEO descriptions and metadata.
After
Generate and pixel-check SEO meta descriptions, compare practical variants, and copy the one that best fits your page.
| Signal | Working guidance |
|---|---|
| Pixels | Keep the important promise before the configured preview width |
| Characters | Use as a rough screen only |
| Content | Match the page; Google may choose different snippet text |
What you get
Each option carries a row of pills; the colour of a pill is the verdict:
- Green — within the configured preview band, with the keyword and call to action present.
- Amber — a warning: over the length band, over one configured preview width, or a missing call to action / keyword you may want to add.
- Red — well beyond a configured preview width or keyword-stuffed (a repeated word), review before use.
The specific pills you'll see:
- “N chars” — raw character count, for reference alongside the measured width.
- “▮ N/920px desktop” and “▯ N/680px mobile” — measured width versus this tool's current preview defaults, not published Google cutoffs.
- “✓ keyword” / “✗ no keyword” — whether your primary keyword appears.
- “✓ call to action” / “add a call to action” — whether the description prompts an action.
- “⚠ repeats …” — flags an over-used word (keyword stuffing), shown in red.
A Template mode banner above the results means the AI is off or over its daily budget and these are rule-based drafts — see How it works.
How it works
When you generate, the tool sends your keyword, brand, and any page content to a small server endpoint that drafts variants with a language model (Llama 3.1 8B running on Workers AI). It drafts wording from the supplied inputs, but it can still invent or overstate details. You approve every option, which is why the Paste and Fetch modes, which give it your real page text, produce the most accurate descriptions.
Everything after drafting happens in your browser. Each variant is measured pixel-for-pixel with a canvas font-metrics engine (the same one behind the SERP Snippet & Truncation Checker, so numbers agree across tools), scored for keyword, call to action, and stuffing, and rendered into the live SERP preview. Editing a variant re-runs all of that instantly — no round trip.
The AI drafting is metered and runs on a daily free budget. When that budget is spent for the day — or if AI is switched off or unavailable — the endpoint says so and the tool falls back to a rule-based template generator that builds descriptions from your inputs with fixed patterns. That fallback is labelled honestly with the Template mode banner; the pixel measurement, scoring, and preview are unchanged, only the drafting is more formulaic. A network error triggers the same graceful fallback. Nothing you type is stored.
Features
- Five description options per run, plus five matching title tags, from one keyword.
- Measured-width desktop and mobile preview meters, presented as guidance rather than Google cutoffs.
- Live Google SERP preview with a desktop / mobile toggle and keyword bolding.
- Per-option scoring: keyword presence, call to action, length band, and keyword-stuffing flag.
- Quick, Paste-content, and Fetch-a-URL input modes so the AI writes from real page text.
- Bulk mode for up to 10 keywords at once, with CSV export.
- Copy-ready
<meta>+<title>block reflecting your two chosen options. - Graceful template fallback so the tool works even when AI is off or over budget.
Limitations
The AI drafts wording, not verified facts — always read each option and correct anything that doesn't match your page, especially in Quick mode where it hasn't seen the content. Because drafting runs on a daily free budget, at busy times you may land in Template mode, which produces more formulaic copy until the budget resets the next day. The Fetch a URL mode extracts text in your browser, so JavaScript-rendered pages may return little — paste the content instead. And Google can create query-dependent snippets from page content, so a description that scores well here is a candidate, not a guarantee of what appears in any result.
Frequently asked questions
What is the ideal meta description length?
There is no exact character count or fixed Google cutoff. Proportional text means a line full of wide letters takes more space than one full of narrow letters, and Google can choose a different snippet for each query. This tool compares measured widths with approximate desktop and mobile preview guidance, so keep the important part early instead of treating 155 or 160 characters—or any pixel value—as a guarantee.
Does Google actually use the meta description you write?
Not always. Google primarily creates snippets from page content and may use the meta descriptionThe meta description is an HTML head tag — `<meta name="description" content="…">` — that suggests a short summary of the page for the search snippet. It's not a Google ranking factor, and Google rewrites it the majority of the time, but a good one can still lift click-through. when it better describes the page for a particular search. A useful description is still worth writing as an accurate, concise candidate, but it is neither the default for a main query nor a display guarantee.
Is the AI going to make up facts about my page?
The generator can invent or overstate details, so review every option before using it. Paste content or Fetch a URL gives the model more real page text than a keyword alone, but it does not verify the output. Every variant is editable and the preview meters update as you correct it.
Why does the tool sometimes say "Template mode"?
The AI drafting runs on a metered model with a daily free budget. When that budget is used up for the day, or AI is switched off, the tool falls back to a rule-based template generator so it never stops working. Template mode is clearly labelled with a banner; the suggestions are more formulaic but the pixel measurement, scoring, and SERP preview are identical. Come back the next day for AI drafts, or edit the templates directly.
What is the difference between this and the SERP Snippet & Truncation Checker?
This generator creates and scores several new description options from a keyword. The SERP Snippet & Truncation Checker takes a title and description you already have and compares their measured widths with configurable preview guidance. Use the generator to draft, then the Snippet Checker to review your final choice in context — they share the same measurement engine, so their preview numbers agree.
Feature requests for Meta Description Generator
Upvote what you want most. New ideas can be submitted from the floating Feedback menu; requests appear here once approved, and the most-wanted rise to the top.
You won't be emailed about that request anymore.
Loading…
➕ Request a feature
New requests are reviewed before they appear here.
Where this tool helps
Common use cases
Ground the wording in the page's actual content
Paste page copy or extract a public URL so the drafts reflect what the page genuinely offers and are easier to check for overstatement.
Compare keyword, call-to-action, and measured width signals
Use the colored pills to spot missing topic language, weak click guidance, repetition, and approximate desktop or mobile preview overflow.
Choose a matching title and copy both tags together
Edit any option, preview the selected title and description as one search result, then copy the paired HTML tags for implementation.
Create a small review sheet without hiding fallback mode
Generate up to ten rows, export CSV, and keep the visible template-mode label when AI is unavailable so reviewers know how the wording was produced.
Watch the full workflow
Meta Description Generator walkthrough
Read the transcript
Meta Titles & Descriptions
This beginner walkthrough covers both free metadata generators. We will define title tags and meta descriptions, choose the right input mode, run one complete fixed example, interpret the preview and scoring pills, edit and select a pair, extract page text, create a small bulk review sheet, compare the title-first and description-first pages, and finish with privacy, fallback behavior, limitations, and next steps. Every generated sentence is a deterministic walkthrough fixture, not live model output.
Step 1
A title tag is the page title in the H T M L document. Search engines often use it as the blue title link, but they may rewrite it. A meta description is a short page summary that can become the text below that link, but it can also be replaced with a query-specific excerpt. These tools create five editable options for the primary field and five matching options for its companion, so you review a complete pair rather than isolated sentences.
Step 2
Start here when a new page needs several title directions, when you want the drafts grounded in real page copy, when you need separate desktop and mobile width guidance, or when a reviewer needs a complete title-and-description handoff. The small bulk mode is useful for an editorial review sheet. It is not an unattended publishing system; each option remains a draft that must match the page.
Step 3
Quick mode uses a keyword, optional brand, tone, page type, and short angle. Paste Content gives the drafter the page's real text. Fetch a U R L retrieves a public page and extracts readable text in the browser. Bulk accepts up to ten rows written as keyword, vertical bar, brand, vertical bar, topic. Competitor titles are optional comparison context. Use the smallest mode that still supplies enough evidence to verify the wording.
Step 4
The fixed example is a beginner guide to trail running shoes for a fictional Northstar Outdoor site. The primary keyword is “trail running shoes.” The professional tone and guide page type frame the job, while the page angle adds lightweight options, fit guidance, and free returns. These details give the drafter useful boundaries, but they do not prove that every generated claim is true.
Step 5
Choose “Generate titles.” The fixed fixture returns five title options and five matching descriptions. In normal use, A I wording can vary, so generation is only the start of the review. The page opens a live search-result preview, selects the first title and description, and builds a copy-ready tag block from that pair. Nothing is silently published.
Step 7
The first title has forty-five characters, measured widths for desktop and mobile, the full keyword, and the keyword near the front. Green means the value sits inside this tool's working guidance. These pixel values come from the browser's measurement engine. They are more useful than character count alone because wide and narrow letters occupy different space, but they are still guidance rather than published Google cutoffs.
Step 8
The long final option keeps the keyword, but its desktop width is red and its rewrite-risk pill says it exceeds the pixel guidance. The mobile meter is different because the mobile preview uses a different working width. This is a reason to edit or reject the candidate, not proof that one exact title will always be truncated. Search engines can choose a different title link for a query.
Step 9
Edit the selected title directly. The pills, preview, and copy-ready code update as the text changes. Then choose “Mobile.” The narrower visual treatment is shown separately, and the keyword remains highlighted. Editing is the important step: keep the wording accurate, make the main topic easy to recognize, and remove claims or modifiers that the page cannot support.
Step 10
The matching description is also editable. Its pills report character count, measured desktop and mobile width, keyword presence, and whether it includes a call to action. Here the mobile width is red even though desktop is green, so the two views should not be mixed. A call-to-action signal only says the sentence invites the next step; it does not prove that the wording is persuasive or accurate.
Step 11
In Fetch a U R L mode, enter a public address and choose “Fetch and extract.” The fixed page is retrieved, navigation and scripts are removed, and the remaining text is placed into Paste Content for review. This makes the evidence visible before generation. JavaScript-only pages may return little, and extraction can omit context, so inspect the text and paste better source copy when necessary.
Step 12
Bulk mode accepts no more than ten rows. Each row starts with the keyword, followed by optional brand and topic fields separated by vertical bars. The fixed example has three unrelated page ideas so each row remains easy to review. Bulk uses less page context than the single-page modes, so use it for drafting and triage, not for replacing a careful page-level content check.
Step 13
Choose “Generate all.” The fixed table returns one paired title and description for each of the three rows, plus measured title and description widths. “Export C S V” creates a spreadsheet-friendly review handoff in the browser. Preserve the keyword, brand, topic, generated wording, and reviewer decision together. A green width does not turn unverified wording into approved copy.
Step 14
Meta Description Generator uses the same engine and inputs, but it leads with descriptions. Choose this page when the main task is improving the summary beneath a search title, adding a clear next step, or checking whether the candidate actually describes the page. It still creates matching title tags because the two lines should be reviewed as one search-result candidate.
Step 15
The description-first use cases emphasize click-focused summaries, real-page grounding, keyword and call-to-action signals, paired implementation, and honest fallback mode. The central question is not “Did A I produce a sentence?” It is “Does this sentence accurately summarize the page, help the right searcher decide, and still work with the selected title?”
Step 16
Choose “Generate descriptions.” The same deterministic set now appears with “Meta descriptions” as the primary section and “Matching title tags” underneath. This is not a different model or a disconnected workflow. It is the same paired output presented in the order that matches the reviewer's task.
Step 17
The third description uses related wording but omits the exact primary keyword. The amber pills flag “no keyword” and “rewrite risk: primary keyword absent.” That does not automatically make the sentence unusable; natural language and query context matter. It does tell the reviewer to check whether the page topic is still obvious and whether a clearer description would better match the selected title.
Step 18
The live preview and copy-ready code still reflect exactly one selected description and one selected title. Copying the pair together reduces the chance that implementation receives a title from one review round and a description from another. The C S V export keeps all candidates for comparison, while the tag block is the concise handoff for the chosen pair.
Step 19
Both pages include five primary options, five companion options, measured-width preview guidance, inline editing, a desktop and mobile search-result preview, keyword and repetition checks, a call-to-action signal for descriptions, four input modes, paired tag copy, and C S V export. If A I is unavailable or the free daily budget is spent, editable templates keep the local scoring and preview workflow available.
Step 20
A I drafts wording, not verified facts. Quick mode has not seen the page content. Fetched pages can lose JavaScript-rendered text. Template fallback is more formulaic. Width bands are product guidance, not Google rules. Google can create a different title link or description for each query and device. Inputs used for generation are sent to the endpoint and are not stored for reuse; a chosen public U R L is sent to this site for fetching, while scoring, editing, preview, and export run in the browser.
Step 21
Before implementation, read the title and description aloud, compare every claim with the page, remove repetition, confirm that the main topic is clear, and check desktop and mobile separately. Copy the selected tags, then use the S E R P Snippet and Truncation Checker for another approximate preview. After the page is indexed, monitor actual search queries and click behavior. Generated metadata is a hypothesis that improves through evidence and revision.
Draft. Verify. Edit. Preview. Then publish.
Treat the selected pair as an implementation candidate, not an automatic answer. Verify every claim against the page, keep the important topic clear near the front, read the desktop and mobile previews separately, and test the final result in the search snippet checker. After publishing, inspect real search performance and revise when the page, query intent, or offer changes. The tool accelerates drafting and review; the final judgment remains yours.
Common issues & how to fix them
- Error Generated description is empty Fix: Provide a primary topic and meaningful page summary, then generate a nonempty description that reflects the page.
- Warning Generated description is too short Fix: Expand the description toward 140–160 characters with one clear benefit and a useful differentiator.
- Warning Generated description is likely to truncate Fix: Trim the description toward 140–160 characters while preserving the primary topic, benefit, and soft call to action.
- Warning Generated description omits the primary topic Fix: Include the page’s primary topic naturally near the start of the description without repeating it mechanically.
- Information Generated description duplicates another candidate Fix: Rewrite the duplicate candidate with a distinct benefit, audience, or proof point instead of publishing interchangeable descriptions.