How to Do an SEO Analysis of Your Website (Step by Step, With a Free Tool)
Type “SEO analysis” into Google and most of what comes back is either a vague checklist (“write good titles!”, “make your site fast!”) or a tool that spits out a single score with no explanation of what it actually measured. Neither one tells you what’s actually wrong with your site.
A real SEO analysis is narrower and more useful than that: it’s a specific check of three things — what’s on the page, how fast it loads, and whether search engines can actually crawl and index it — each producing a concrete, individually verifiable finding. This guide walks through exactly what those three pillars cover, how to read the results without a technical SEO background, and where to run one for free.
Pillar 1: On-Page Signals
On-page signals are the easiest part of an SEO analysis to check and the most common place sites lose easy rankings, because the mistakes here are usually invisible unless you go looking for them.
| Signal | What it tells you |
|---|---|
| Title tag length | Titles over ~60 characters get truncated in search results; missing titles mean Google has to guess what the page is about |
| Meta description | Doesn’t directly affect rankings, but a missing or poorly written one hurts click-through rate from the search results page |
| Canonical tag | Tells search engines which URL is the “real” version when the same content is reachable through more than one address |
| Viewport tag | Confirms the page is actually configured to render correctly on mobile, not just visually similar to it |
| H1 count | Exactly one H1 per page is the standard convention; zero or multiple H1s is a structural signal worth fixing |
Declared language (lang attribute) | Missing this can affect how search engines and screen readers interpret the page’s content |
None of these are exotic checks. They’re exactly the kind of thing that silently breaks after a theme change, a CMS migration, or a plugin update — which is why re-checking them periodically matters more than getting them right once.
Pillar 2: Core Web Vitals & Performance
This is the pillar most SEO checklists wave at vaguely (“make your site fast”) without explaining what’s actually being measured or why it matters beyond a vague sense of user experience.
Core Web Vitals break performance into specific, individually measurable metrics:
- LCP (Largest Contentful Paint) — how long until the largest visible element finishes rendering. Good: under 2.5s.
- CLS (Cumulative Layout Shift) — how much visible content jumps around while the page loads. Good: under 0.1.
- TBT / interactivity — how long the main thread stays blocked and unresponsive to clicks and taps during load. This is the metric behind the 25-second Total Blocking Time case study we published — a page that looked fully loaded but couldn’t register a single click for 25 seconds.
- TTFB (Time to First Byte) — how long the server itself takes to start responding, before the browser has anything to render at all. This one is about your hosting and server configuration, not your front-end code.
A proper analysis reports each of these individually, with the actual published Google thresholds (not an arbitrary “score out of 100” that hides what’s actually slow) — because “your performance score is 61” tells you nothing actionable, while “your TTFB is 1.8 seconds against a 0.8 second threshold” tells you exactly what to fix and roughly how much room you have.
Pillar 3: Technical Crawlability
This is the pillar that’s easy to forget because, when it’s broken, the site still looks completely normal to a human visitor — it just becomes invisible or partially invisible to search engines.
- Robots.txt — controls which parts of your site search engines are allowed to crawl at all. A missing file isn’t necessarily fatal, but a misconfigured one can accidentally block your entire site.
- XML sitemap — gives search engines a direct map of every page you want indexed, instead of relying entirely on them discovering pages by following links.
- Broken links (4xx/5xx) — every broken internal or external link wastes crawl budget and, if the destination page itself is one of yours, can quietly remove it from the index entirely.
- Duplicate content and canonical conflicts — this is where faceted navigation, CMS pagination, and manufacturer-boilerplate product copy do the most damage. We’ve covered this in depth for two specific platforms: Shopify’s faceted URL duplication problem and WordPress’s plugin-conflict and archive-duplication issues — the underlying principle (one canonical version per piece of content) is the same regardless of platform.
How to Read the Results
Every metric in a real SEO analysis should fall into one of three bands, and understanding what they mean matters more than the raw number:
| Band | What it means |
|---|---|
| Good | Meets the published threshold — no action needed on this specific signal |
| Needs Improvement | Functional but measurably below the threshold — worth fixing, not on fire |
| Poor | Below the threshold by enough that it’s actively costing you rankings or conversions |
A site with everything in “Good” except one metric in “Poor” has a clear, specific next action. A site with everything at a vague “72/100” has no next action at all — just a number to feel anxious about.
Automated Scan vs. Manual Audit: What’s the Difference
An automated tool and a manual technical SEO audit answer different questions, and conflating them is the single most common mistake people make when interpreting their own results.
An automated scan is genuinely good at exactly the things covered above: on-page tag presence, live Core Web Vitals, TTFB, broken links, robots.txt/sitemap presence. These are objective, binary-or-numeric checks — either the title tag exists or it doesn’t, either LCP is 2.1 seconds or it isn’t. There’s no judgment call involved, which is exactly why a script can check them reliably in under a minute.
A manual audit picks up where that objectivity runs out. Whether your content actually satisfies search intent, whether your internal linking structure reflects real topical authority, whether a backlink profile looks natural or manipulated, whether a specific faceted-navigation setup needs full noindexing or selective indexing for high-demand filters — these all require a human making a judgment call informed by the specific business, not just a threshold comparison.
The practical takeaway: run the automated scan first, always. It’s free, it’s fast, and it will catch the majority of what’s actually wrong on most sites — thin titles, missing meta descriptions, slow servers, broken links. Reserve manual audit time for the sites where the automated scan comes back clean but rankings still aren’t moving, because that’s the signal the problem is one only a human read can catch.
Mistakes People Make Reading Their Own Results
A few patterns show up constantly when site owners run their first SEO analysis and try to act on it alone:
- Chasing the “Poor” scores in the wrong order. A single “Poor” TTFB reading is usually a hosting/server problem worth fixing immediately; a single “Poor” CLS reading on one page is often just a lazy-loaded ad slot and far lower priority. Not every red flag deserves the same urgency.
- Fixing the symptom instead of the cause. Rewriting a title tag that’s too long treats the symptom; the more common underlying cause is a CMS template auto-appending the site name to every single page title regardless of length, which will keep breaking new pages until the template itself is fixed.
- Re-running the scan on the homepage only. The homepage is often the most polished page on a site by far. App-injected script bloat, missing meta descriptions, and duplicate content overwhelmingly concentrate on product pages, category pages, and blog posts — pages that never get manually reviewed as carefully.
- Treating a clean scan as “done forever.” A site that passes every check today can fail several of them next month after a routine plugin update or theme change, which is why this is a recurring process, not a one-time task.
Run Your Own Free SEO Analysis
We built exactly this — a free tool that checks all three pillars above against a real, live fetch of your site, not a cached estimate: on-page signals, real Core Web Vitals via Google’s PageSpeed Insights, real server response time measured directly, and technical crawlability including broken-link detection and robots.txt/sitemap checks.
Run a free SEO analysis of your website →
It takes under a minute, requires no signup, and shows you the same Good/Needs Improvement/Poor bands covered in this guide — not a mystery score.
Need the Fixes, Not Just the Findings?
An analysis tells you what’s wrong. Fixing it — migrating hosting, rewriting duplicate content, eliminating render-blocking scripts — is where the actual ranking gains come from. That’s the part we do as technical SEO services, built around real before-and-after numbers rather than a generic checklist.
Get a free technical SEO audit →
Published by the Runkexpert Engineering Team. Last updated 2026-07-15.