📋AI Meta Description Rewriter
Meta descriptions are the most undervalued piece of on-page SEO real estate on the web. Google has stated publicly, and repeated for over a decade, that meta descriptions are not a direct ranking signal. That has led most writers to either skip them entirely, copy the first sentence of the page, or hand them to an intern who knocks out a dozen at the end of a Friday afternoon. The problem is that meta descriptions remain the single biggest driver of organic click-through rate once a page already ranks, and click-through rate is itself a behavioural ranking signal Google factors into how aggressively to surface a result. A page sitting at position five with a 12 percent click-through rate often outranks a page at position three with a 4 percent click-through rate inside 90 days, because Google reads sustained click-through behaviour as relevance. Skipping the meta description is throwing away the highest-leverage compounding win in on-page SEO. The FixTools Meta Description Rewriter takes your existing weak description, or a paragraph summary of the page if you do not have a description yet, then generates five fresh variants each under 155 characters, each emphasizing a deliberately different angle: benefit-led, curiosity-led, urgency-led, authority-led, and direct-CTA. Each variant is character-counted, each is checked against the 155 limit before being emitted, and the output includes a recommendation block that flags which variant suits commercial-intent queries, which suits informational queries, and which pair of variants to run as an A/B test on live traffic. The Notes block calls out the CTA verb each variant uses, flags any variant that brushes against the character ceiling, and suggests a learnable split test pairing. The free tier handles inputs up to 600 characters, which fits most page summaries comfortably, and the paid tier extends to 5,000 characters for richly contextualised pages or detailed product briefs.
Why meta descriptions still matter even though Google says they are not a ranking signal
Google's official position, restated as recently as 2024, is that meta descriptions do not directly influence rankings. That statement is technically true and strategically misleading. Meta descriptions do not feed the ranking algorithm directly, but they feed click-through rate, and click-through rate feeds the ranking algorithm through Google's user-interaction signals. Multiple independent studies from Backlinko, Ahrefs, and Search Engine Journal have shown that pages with rewritten meta descriptions see organic click-through rate lifts of 15 to 40 percent within four weeks, and pages that sustain elevated click-through rate frequently move up one to three positions on competitive queries within 90 days. The mechanism is indirect but reliable: a better description earns more clicks, more clicks signal relevance to Google, relevance signals push the page up the SERP, and the higher position earns even more clicks. The loop compounds over months. Treating meta descriptions as a low-priority chore costs you the easiest compounding win in on-page SEO, and it costs it on every page you publish, not just one. A content team shipping 30 posts a month and skipping the meta description is forfeiting roughly 25 to 35 percent of the click-through rate it could otherwise earn, which on a healthy site translates into tens of thousands of sessions lost per quarter.
The 155-character limit is the binding constraint, and it is stricter than most writers realise. Google's desktop SERP snippet wraps to two lines at roughly 158 characters, and the mobile SERP, which now accounts for over 60 percent of all Google traffic, truncates at around 120 characters on narrow viewports like older iPhones held in portrait. Writing to 155 characters as the safe ceiling means your description displays in full on the dominant desktop layout while losing no more than the closing call-to-action on mobile, which is the right tradeoff because mobile users skim faster and weigh the first line more heavily than the second. Descriptions that run over 160 characters get truncated mid-thought with an ellipsis, which suppresses click-through rate by an estimated 8 to 12 percent according to research from Moz. The rewriter counts characters including spaces and punctuation, refuses to emit any variant that exceeds the limit, and tags each variant with its exact count so you can pick the one that fits your specific layout priorities. The rewriter targets 145 to 150 characters as the sweet spot, leaving 5 to 10 characters of headroom so Google is less likely to rewrite the description on the fly.
Angle diversity beats single-shot generation by a wide margin. Most AI writers, asked to produce a meta description, return one polished line written in the same neutral marketing voice every time, optimised for nothing in particular. Five variants emphasizing five distinct angles let you pick the one that matches your audience and your query intent rather than settling for a generic line that suits neither. A benefit-led description ("Cut your PDF size by 90 percent in 10 seconds, no signup, no watermark") wins for commercial-intent searchers who already know what they want and just need confirmation that this page delivers it. A curiosity-led description ("The compression trick that shrinks PDFs without anyone noticing the quality drop") wins for top-of-funnel readers still researching and weighing options. An urgency-led description wins for seasonal, time-sensitive, or limited-availability topics where scarcity drives action. An authority-led description wins when the searcher needs trust signals (named expert, cited research, established brand) before clicking. A direct-CTA description wins when the page is a tool, signup flow, or checkout rather than a piece of content. Producing all five in parallel surfaces the angle you would not have considered if you wrote one variant and stopped.
The recommendation block is where the tool earns its keep. Generating five variants is useful, but picking the right one without context is the hard part, and the part most writers get wrong. The model reviews all five variants together, infers the likely search intent from your input, and tags the best overall pick, the best variant for commercial-intent queries where the searcher is ready to act, and the best variant for informational queries where the searcher is still researching. It also flags an A/B test pairing, naming the two variants that differ enough in framing and emotional appeal to produce a learnable signal if you split-test them on live traffic over two to four weeks. The output includes a Notes block calling out which CTA verb each variant uses, whether any variant brushed up against the 155-character ceiling and might be at risk on edge-case rendering, and confirmation that no clickbait phrases slipped into the output. This block turns a brainstorm into a decision, which is the difference between a content team that ships and one that endlessly debates. Decisions, not options, are what move organic traffic.
How to use AI Meta Description Rewriter
- 1
Paste your existing meta description or a page summary
Paste the current meta description if you have one, or a one-paragraph summary of the page if you do not. The richer the input, the better the variants. Include the primary keyword, the target audience, and the page's core promise. For an existing weak description, paste it verbatim so the rewriter can preserve the parts that already work while rewriting the parts that do not. The free tier accepts up to 600 characters of input, which fits most page summaries comfortably.
- 2
Optionally specify search intent and CTA preference
If you know whether the page targets a commercial-intent query (someone ready to buy or sign up) or an informational query (someone researching), mention that in your input. Also mention any CTA verb you want to favour, for example "try," "download," "compare," or "learn." The rewriter incorporates these constraints into the variants and weighs them in the recommendation. Omitting these is fine, the model will infer reasonable defaults from your input, but explicit hints produce sharper output.
- 3
Run the rewriter
Click Generate and the model produces 5 variants tagged with their angle and character count, in roughly three to five seconds. Variants stream in one by one so you can read the first while the others are still being written. Every variant is verified under 155 characters before being emitted, and the model regenerates any variant that goes over rather than truncating it.
- 4
Read the recommendation and Notes blocks
After the five variants, the output includes a Recommendation block naming the best overall pick, the best variant for commercial-intent queries, and the best variant for informational queries. The Notes block lists the CTA verb each variant uses, flags any variant within five characters of the 155 limit (which may render differently on edge-case devices), and names two variants worth pairing in an A/B test. This is where most of the value compounds.
- 5
Ship the winner and split-test if traffic warrants
Replace your current meta description with the recommended variant and request reindexing via Google Search Console. For pages with significant traffic, set up a 30-day A/B test using the recommended pairing, alternating descriptions weekly and tracking click-through rate in Search Console. The data will tell you which angle resonates with your specific audience, which is information no model can infer from a description alone.
Real-world use cases
E-commerce product detail page
A direct-to-consumer cookware brand has 140 product pages, each with a meta description written by the product manager during the launch sprint two years earlier. The descriptions read like SKU summaries rather than search-optimized snippets ("Stainless steel sauté pan, 12 inch, brushed handle"), and the brand's product-name queries underperform compared to the volume of monthly impressions, suggesting searchers are seeing the listings but choosing competitors instead. The SEO lead exports the current descriptions, runs each product summary through the rewriter, and chooses the benefit-led variant for hero products marketed on the cooking outcome and the direct-CTA variant for accessories priced under fifty dollars where the path to purchase is short. Within six weeks Google has reindexed the catalog and aggregate click-through rate on product-name queries has climbed from 2.1 percent to 3.7 percent according to Search Console, a 76 percent relative lift that translates into roughly 9,000 additional sessions per month on the same impression base. The brand reinvests the project budget into a quarterly rewrite cadence for the rest of the catalog.
Blog editor refreshing evergreen posts
A B2B SaaS content team has a backlog of 60 evergreen blog posts written between 2022 and 2024, most of which still rank in the top 20 for their target keywords but have suspiciously low click-through rates compared to the SERP average for their positions. The editor runs each post's opening paragraph through the rewriter, picks the angle that matches the inferred search intent for the keyword (curiosity-led for top-of-funnel terms, authority-led for terms where competitors are weak), and schedules a republish with the new meta description and a refreshed publish date. The team batches the work into two-hour sessions, processing roughly ten posts per session, and ships the entire backlog inside three weeks. Three months later, 38 of the 60 posts have climbed at least two positions in Search Console, with the median page moving from position 12 to position 7, and aggregate organic sessions to the refreshed set are up 41 percent. The editor attributes most of the lift to the meta description refresh rather than the date change, because date changes alone in earlier experiments produced minimal movement on similar pages.
SaaS landing page conversion optimization
A pre-seed SaaS startup has one landing page targeting the keyword "team retrospective tool" and the founder wrote the meta description in five minutes the day before launch, copying the hero subhead verbatim. The page ranks at position eight but the click-through rate is below one percent, suggesting the description is not earning clicks even though searchers are seeing the listing. The founder runs the page summary through the rewriter, gets five variants, and decides to ship the curiosity-led variant for two weeks followed by the direct-CTA variant for two weeks, using the rewriter's recommended A/B pairing. Click-through rate climbs to 3.4 percent with the direct-CTA variant, which leads with "Try retros that actually surface root causes" rather than the original passive description. The page moves to position four within six weeks as Google reads the elevated click-through rate as a relevance signal, and signups from organic search double over the same period. The founder treats the variant pairing as a recurring quarterly exercise on every public landing page and builds it into the team's SEO playbook.
SEO consultant auditing a client site
A solo SEO consultant takes over a client account where the previous agency wrote 220 meta descriptions over the past two years, half of which exceed 160 characters and truncate on desktop SERPs with ellipses, and a third of which read like duplicate boilerplate copied across category pages without customisation. The consultant exports the descriptions to a spreadsheet alongside the page H1 and opening paragraph, runs each row through the rewriter, and produces a CSV of recommended replacements tagged by inferred query intent. They prioritize the 50 pages with the highest impressions but lowest click-through rates for the first batch of updates, since those are the pages where the rewrite will produce the largest visible lift, and they request reindexing through Search Console immediately. After tracking the cohort in Search Console for four weeks to confirm the lift, they roll out the remaining 170 descriptions in waves of 30 per week to avoid overwhelming the crawl budget. The consultant turns what would have been a tedious manual rewrite job into a one-day project and bills the client for the click-through rate lift over the following quarter rather than for the hours spent at the keyboard, which is the higher-margin engagement model.
Pro tips
💡 Lead with the value proposition, not the brand name
Many writers default to "BrandName helps you do X" as the opening of the description. That wastes the first 15 characters on a phrase that does not influence the click. Lead with the benefit, the outcome, or the number, and mention the brand at the end if at all. The exception is when the brand itself is the search query, in which case leading with the brand name is the right call because it confirms the searcher landed on the right page.
💡 Match the CTA verb to the page type
A blog post should not say "Buy now" and a checkout page should not say "Learn how." Use "Read," "Discover," or "Compare" for informational content. Use "Try," "Start," or "Get" for product pages and signup flows. Use "Shop," "Order," or "Add to cart" for commerce. The rewriter's direct-CTA variant adapts the verb to the inferred page type, but if you want a specific verb, mention it in your input and the model will prioritise it.
💡 Stay 5 to 10 characters under the 155 limit for safety
Google occasionally rewrites meta descriptions on the fly when it judges the page content a better fit for the query. Descriptions that sit right at the 155 character ceiling are more likely to be truncated or rewritten than descriptions that leave 5 to 10 characters of headroom. The rewriter targets 145 to 150 characters as the sweet spot, which gives you the full benefit of the snippet without flirting with the edge case where Google decides to rewrite.
💡 Split test the top two angles for at least 14 days
The recommendation tells you which variant the model thinks is best, but real user behaviour reveals niche-specific patterns the model cannot infer from a description alone. Pair the recommended overall winner with the variant tagged for the opposite intent (commercial vs informational) and run them in alternating weeks for at least 14 days each. Track click-through rate in Search Console at the query level, not the page level, because aggregate page CTR masks per-query differences.
Frequently asked questions
Does Google actually use meta descriptions for ranking?
Not directly. Google has stated for over a decade that meta descriptions are not a ranking factor. However, meta descriptions strongly influence click-through rate, and click-through rate is a behavioural signal Google factors into how aggressively to surface a result. A better meta description leads to more clicks, which leads to a position lift over time. The effect is indirect but reliable, and it compounds over weeks rather than days.
Why exactly five variants instead of three or ten?
Five corresponds to the five distinct angles that cover the most common search intents: benefit, curiosity, urgency, authority, and direct CTA. Three variants would not cover the full angle space. Ten variants produce decision fatigue and most writers default to the first one they like rather than evaluating each. Five is the smallest number that gives you genuine angle diversity while still being readable in one screen.
How strict is the 155-character limit?
Strict. The model self-checks every variant against the limit before emitting it and regenerates any variant that runs over. 155 characters is the safe ceiling for Google's desktop SERP snippet, which displays two lines wrapping at roughly 158 characters. The rewriter targets 145 to 150 characters as the sweet spot, leaving 5 to 10 characters of headroom so Google is less likely to rewrite the description on the fly.
Can I rewrite descriptions for non-English pages?
Yes. Paste your input in English but specify the target language, for example "rewrite this in French for a French-Canadian audience" or "produce variants in German for a B2B industrial buyer." The rewriter produces variants in the target language, applies the 155-character limit correctly (including for multibyte languages like Japanese and Korean, where the visible character count differs from the byte count), and adapts the CTA verbs and idioms to language-appropriate conventions rather than translating English phrasing word for word.
How is this different from asking ChatGPT to rewrite my meta description?
A general chat tool will return one polished line in the same neutral voice every time and will not check the character limit. The rewriter runs a tested prompt that always produces five distinct angles, always counts characters, always rejects variants over 155 characters, and always includes a structured recommendation block. It is purpose-built for this one task, so you do not have to remember to specify character limits, angle diversity, or query-intent recommendations every time.
Will the rewriter use clickbait phrases to boost click-through rate?
No. The system prompt explicitly forbids phrases like "you won't believe," "this one trick," "doctors hate this," and similar clickbait. Clickbait initially lifts click-through rate but suppresses dwell time and increases pogo-sticking back to the SERP, which Google reads as a negative behavioural signal. The rewriter optimizes for sustained click-through rate by matching honest value propositions to query intent, not by pattern-matching to clickbait formats.
Should every page on my site have a custom meta description?
Not necessarily. High-traffic pages, money pages (signup, pricing, top blog posts), and pages targeting competitive keywords absolutely should. Low-traffic utility pages (privacy policy, individual archive pages, search result pages) can rely on auto-generated snippets. As a rule, the top 20 percent of pages by impressions produce 80 percent of the click opportunity, and rewriting those first delivers the highest return on the time invested.
How much does each generation cost in credits?
One credit per generation, regardless of input length. Day Pass users get 100 credits, Monthly users get 1,500 credits per month, and Free tier users get 5 generations per day. There is no cost difference between a short paste and a long page summary, only the character limit per request changes between free and paid tiers. The credit is deducted on successful generation only, not on failed or aborted runs.
Related tools
AI SEO Title Generator
Pair your new meta description with 10 title variants under 60 characters, each tagged with intent angle and ready for SERP testing.
Try it
Meta Description Generator
Build a meta description from scratch when you do not have an existing one to rewrite, sized to the same 155-character SERP snippet limit.
Try it
Open Graph Generator
Generate the og:title and og:description tags for social sharing, reusing the angle and CTA from the meta description you just rewrote.
Try it