Core Web Vitals · Schema · SERP preview · Rendered DOM
Find out what is wrong with your on-page SEO.
A free on-page SEO analysis of any URL: title pixel widths, heading trees,
image attributes, schema and Core Web Vitals. Every issue holding the page back, ranked by
what is worth fixing first, with the markup to fix it.
Most tools count characters, render the page in a browser and hand you an
undifferentiated list. We do none of those.
Pixels, not characters
Truncation measured in pixels
Search results cut off on rendered width, not character count.
“Illinois” and “llllllll” are both eight characters
and 20px apart on screen.
Desktop and mobile limits checked separately
Measured against Arial metrics, the font Google renders in
Preview shows where the cut falls, if it does
Google desktop: about 560px of the 580px limit
onbixo.com › blog › onboarding flow targeting rules who sees what when
Onboarding flow targeting rules: who sees what, when - Onbixo
Onboarding flow targeting rules, in practice: pick the condition not the persona, group rules with ANY and ALL, cap how often a flow fires…
! 61 characters, over the 60 usually quoted, yet it still fits
Rendered, not raw
We audit the page as it gets rendered
Google crawls the HTML your server sends, then renders it and indexes what the page
becomes. So that is what we read. A checker that stops at the served markup reports a
missing title on a React page whose title is there by the time it is indexed.
Reads the rendered DOM, the document Google indexes
Reads response headers too, including X-Robots-Tag
Falls back to served HTML when a browser is unavailable
Served vs rendered: the same page, two answers
// what the server sends, correct<title>Onboarding flow targeting rules…</title><h1>Onboarding flow targeting rules</h1>// what a rendering tool reports, wrong<title>Onbixo</title>(no h1 found)
Ranked, not listed
Ranked by impact against effort
Every finding carries both so the report opens with the fixes that buy the most for
the least work, not an alphabetical list of everything at once.
One score across five categories
Disqualifying faults cap the score outright
Unmeasurable categories are excluded, not guessed
Score breakdown: onbixo.com/blog/onboarding-flow…
Indexability & Crawling20 checks · 0 issues100
On-Page Content & Structure16 checks · 2 issues92
Core Web Vitals13 checks · 3 issues70
Schema & Rich Snippets2 checks · 0 issues100
Media & Links9 checks · 3 issues86
Overall scoreGrade A · 52/61 passed91
Coverage
Dozens of on-page factors tested
Grouped the way the report presents them.
Crawling & indexing
15+ checks
A page that cannot be crawled cannot rank, whatever else is right about
it. These run first and a fault here caps the score outright rather
than costing it a few points.
Redirect chains, loops and protocol downgrades
Robots meta and X-Robots-Tag, including headers
Canonical presence, correctness and self-reference
Also hreflang, sitemaps and mixed content
Meta tags & snippet
10+ checks
Search engines truncate on rendered width, not character count so the
title is measured in pixels and the preview shows where the cut
actually falls, desktop and mobile separately.
Title pixel width against the 580px limit
Meta description length and truncation
Open Graph and Twitter card completeness
Also favicon and the social image
Content & metadata
15+ checks
Headings are read as a tree, not a list. A skipped level or a second H1
tells a crawler something different from what the page looks like and the report shows the outline it inferred.
Single H1, heading order and skipped levels
Empty headings that carry structure with no meaning
Content depth and keyword distribution
Also language, charset and viewport
AI & answer readiness
10+ checks
Assistants lift passages so a page that answers in a self-contained
paragraph gets quoted where one that buries the answer mid-section does
not. Dates and attribution decide whether it is trusted.
Answer-shaped paragraphs (40–60 words)
Published and modified dates and staleness
Author attribution and answer schema
Also scannability and freshness
Speed & delivery
15+ checks
Core Web Vitals come from real Chrome users where that data exists and the report says which it used. A lab number and a field number are
different claims and should not be presented alike.
LCP against 2.5s, INP against 200ms, CLS against 0.1
Source labelled: field or lab, never blended
Page weight and request count by resource type
Also compression, caching and payload
Security
15+ checks
A certificate about to expire or a missing HSTS header is not an SEO
problem until it is one, at which point it is the only problem. These
are checked because nobody looks until something breaks.
Certificate validity and expiry window
HSTS, CSP and X-Content-Type-Options
Clickjacking exposure and version disclosure
Also referrer policy and version disclosure
Links, media & schema
20+ checks
Every offending element is listed, not counted: the report names the
images missing alt text and the anchors that say “click
here” so a fix does not start with finding them.
Non-descriptive anchors and dead in-page fragments
Missing alt text and unsized images that shift layout
Malformed JSON-LD and missing required properties
Also affiliate disclosure and rich results
Real impact
See the difference an on-page cleanup makes
The same page before and after acting on what the report ranked first.
Every line below is a finding the tool emits, worded as it words them.
Grade A · Very goodGrade F · Needs work
91
Score / 100
The same four faults resolved: the
title now reaches searchers whole, every image is readable, the subject
is unambiguous and the page is eligible for rich results.
✓
Title fits at 561px of the 580px limit
Onboarding flow targeting rules: who sees what, when - Onbixo
✓
Every image carries an alt attribute
21 images described so each one can be read and indexed.
✓
One H1 and no skipped heading levels
The outline reads H1 to H2 to H3 with nothing missing between.
✓
Article and FAQ JSON-LD both valid
Required properties present so the page is eligible for rich results.
!
Title is 844px wide and truncates in desktop results
264px over the 580px limit so the last four words never appear.
!
21 images have no alt attribute
Each one is listed by source so the fix starts with the file names.
!
Two H1 tags and H2 skipped to H4
A crawler cannot tell which heading states the page's subject.
!
No structured data of any kind
Nothing for a rich result to be built from so none is eligible.
Plus 4 more resolved, including render-blocking
resources and images served at one fixed size.
Plus 4 more found, including render-blocking
resources and images served at one fixed size.
Pricing
Spend less time auditing. Spend more time ranking.
Every plan runs the same checks. A paid one keeps the history, watches the
page after you fix it and turns the report into something you can send.
No. Reports run without signing up. An account is only
needed to keep history, compare a page over time or raise the daily limit.
How do you handle single-page apps built with React, Vue or Next.js?
We render the page in a real browser and audit the result, which
is what Google indexes: it crawls your served HTML, queues the page, renders it and indexes
what comes out. So a React or Vue app that fills in its title and headings on the client is
read the way a search engine reads it, not reported as empty. Worth knowing that rendering is
queued and can lag the crawl and that many other crawlers, including social previews and
most AI assistants, do not render at all so tags that only exist after JavaScript are still
a risk worth closing.
Why measure a title in pixels rather than characters?
Google truncates on rendered width, not character count. A
60-character title of wide letters can be cut where a 70-character one is not so the
report measures the pixels and shows where the cut falls.
Can I export reports with my own logo and branding?
Yes, on the Agency plan. The PDF carries your name, logo, accent
colour, contact line and footer, plus an opening and closing note and the client's name on
the cover. Severity colours stay as they are: red for critical and green for passing are
conventions your client already reads and a report that looked on-brand while reading
wrong would be worse than one that did not match.
Are there limits on how many URLs I can analyse?
One report a day without an account, three once you register and whatever your plan allows above that. A page analysed recently returns from cache and does
not count again so re-reading a report you already ran is free.
Which pages can I analyse?
Any page reachable over the public internet. Pages behind a
login, on a private network or blocked by robots.txt cannot be fetched and the report
says so rather than guessing.
