The score is a HEURISTIC, not a forecast. It rewards the characteristics that research consistently associates with more clicks — the title fitting inside Google’s pixel limit and the ~50–60-character sweet spot, a number, brackets/parentheses, the year, one or two power words, an emotional word, the keyword front-loaded, and a meta description in range with a call-to-action — and blends them into a 0–100 number so you can compare variants on a level field. But your actual click-through rate is driven mostly by things this tool cannot see: your ranking POSITION (position 1 gets vastly more clicks than position 8), the exact query and intent, what your COMPETITORS’ snippets look like, rich results, and real user behaviour. Google also rewrites a large share of titles and often auto-generates the description per query. So use the tester to shortlist the one or two strongest variants, then run a real A/B test and watch the CTR column in Google Search Console before deciding.

The signals this tool checks come from widely-cited CTR research. In the TITLE: stay inside Google’s ~600px desktop pixel limit (roughly 50–60 characters) so it is not cut; front-load the keyword so searchers see the words they typed; add a number or digit (numbers are linked to notably higher CTR and set a clear expectation); use brackets or parentheses like “(Guide)”, “(With Examples)” or “[2025]” for a specificity signal (Google keeps parentheses more often than square brackets); include the current year for evergreen or “best-of” pages; and add one or two power words (proven, ultimate, essential) plus an emotional word and a clear positive or negative tone — but avoid stacking hype. In the META DESCRIPTION: aim for ~150–160 characters so it is not truncated (mobile is tighter at ~120), echo the keyword so Google can bold it, and include a call-to-action (“Discover…”, “Learn how…”, “Compare…”). Above all, keep it accurate — clickbait that mismatches the page loses trust and rankings.

A practical workflow is to enter your current title and meta as a baseline, then two to four rewrites that each push a different angle — one leaning on a number and brackets, one on a power/emotional hook, one that front-loads the keyword and tightens the length. The tester scores and ranks them side-by-side with per-signal chips, so you can see not just which wins but WHY. Testing a small set (the tool supports up to eight variants) keeps the comparison meaningful; testing dozens at once tends to blur the differences. Then take the one or two highest-scoring variants and A/B test them for real, because the heuristic can only compare characteristics — only live data in Google Search Console tells you which actually earns more clicks at your ranking position.

The CTR Rewrite Tester is 100% client-side. The titles, meta descriptions, URL and keyword you type are measured with the Canvas measureText() API (in the same fonts Google renders the SERP in) and matched against local word banks, all on your device; nothing is fetched, uploaded, logged or stored on a server, and no URL is ever crawled. That makes it safe for unpublished pages, draft rewrites and staging titles you do not want to expose to a third-party tool. It works offline once cached as a PWA, supports dark mode and mobile with a sticky action bar, and closing the tab discards everything you entered.