WebEngine

0%
SEO & Analytics

SEO Checklist 2026: 60 Checks a Site Needs Before It Ranks

A complete SEO checklist — indexation and robots.txt through Core Web Vitals, structured data and content. Every item explained by what it actually changes.

Pavlo9 min read

Most sites fail to rank for boring reasons — five or six basics nobody ever checked. Below is a 60-point checklist in six blocks, the one we run every project through before launch and every six months after. Each item is phrased so you can answer yes or no in under a minute, with no separate audit required.

60

checks across six blocks

~80%

of real problems sit in the first two

2 days

for a full pass on a 100-page site

6 mo

recommended re-check interval

How to run this list, and in what order

The order of the blocks is deliberate. Polishing headings on a page blocked from indexing achieves nothing, and building links to a site where half the pages duplicate each other achieves less than nothing. Work top to bottom and do not move on while the current block still holds a "no".

OrderBlockWhen to checkCost of getting it wrong
1Indexation and crawlingBefore launch, after every releasePages absent from search entirely
2Keywords and structureAt wireframe stage, then yearlyURL rebuild and lost rankings
3On-pageFor every new pageWeak CTR despite decent positions
4Technical and speedMonthly, on field dataMobile rankings slipping
5Structured data, mobileAfter launch, then quarterlyNo rich results in the SERP
6Content, links, analyticsContinuouslyGrowth flattens into a plateau

Block 1. Indexation and crawling

The most expensive mistakes live here, and nearly all are release accidents. The most common one we see: Disallow: / travelled from staging to production and went unnoticed for three weeks.

Block 1 — 10 checks

  • No Disallow: / in robots.txt left over from staging
  • Domain added to Search Console, sitemap submitted and read without errors
  • sitemap.xml lists only canonical 200 URLs — no redirects, no 404s
  • Every page carries a single self-referencing canonical
  • www/non-www, http/https and slash variants all 301 to one version
  • Cart, filters and internal search are noindex, not just robots-disallowed
  • Pagination has unique title tags and no canonical to page one
  • No growth in the "Discovered — currently not indexed" bucket
  • No redirect chains longer than a single hop
  • A missing page returns a real 404, not a 200 saying "nothing found"

Checking by hand is slow, so the first eight items close out with the Pages report and URL inspection — how to read both is in our Google Search Console guide.

Blocks 2 and 3. Structure and on-page

Structure decides what you can compete for; on-page decides whether anyone clicks. The first is expensive to change, the second takes an hour — do not confuse the priorities.

Block 2 — keywords and structure

  • A keyword set with volumes and intent, not nice-sounding phrases
  • Each keyword cluster maps to exactly one target page
  • URLs readable, with no dates and no ?id=274 parameters
  • Any page reachable within three clicks of the homepage
  • Breadcrumbs on every page except the homepage
  • Navigation mirrors how demand is shaped, not the org chart
  • Every service has its own page, not a bullet in a shared list
  • Commercial and informational intent separated: services apart from blog
  • No orphan pages without a single internal link
  • Catalogue filters with real demand resolve to indexable static URLs

Block 3 — on-page optimisation

  • title unique, 50–60 characters, primary term near the front
  • description 140–158 characters giving a reason to click, not a keyword pile
  • One h1 per page that does not repeat the title word for word
  • h2/h3 levels descend without skipping
  • The opening paragraph answers the query within two sentences
  • Every meaningful image has a descriptive alt, not "img_2043"
  • Images served as WebP or AVIF with explicit width and height
  • Internal anchors descriptive — never "here" or "read more"
  • No hidden text, no keyword stuffing, no tag cloud at the bottom
  • The update date is real and visible on time-sensitive topics

The one-cluster-per-page rule is broken most often. Two pages targeting the same intent do not double your chances — they halve your signals, and Google picks one, usually the wrong one. Separating clusters at the research stage is covered in our keyword research method.

Blocks 4 and 5. Speed, structured data, mobile

Measure field data from real users here, not a lab run on a developer laptop with gigabit fibre. The gap is typically a factor of two.

Block 4 — technical foundation and speed

  • LCP under 2.5s on mobile, measured on field data
  • INP under 200ms on the heaviest interaction — usually a filter or form
  • CLS under 0.1: banners, ads and fonts all have reserved space
  • HTTPS with a valid certificate and no mixed content
  • Brotli or gzip compression enabled
  • Static assets on a CDN, cached for a year with a hashed filename
  • Homepage JavaScript under 200KB compressed
  • Fonts self-hosted, font-display: swap, main weight preloaded
  • No console errors on the key templates
  • TTFB under 600ms while the site is under load

Block 5 — structured data and mobile

  • Organization or LocalBusiness on the homepage with correct contacts
  • BreadcrumbList on every internal page
  • Product with price and availability for stores, Article for the blog
  • FAQPage only where the questions are genuinely visible to the user
  • Markup passes the Rich Results Test with no errors
  • No horizontal scroll at a 360px viewport
  • Tap targets at least 44×44px
  • Body text 16px or larger, contrast at least 4.5:1
  • The mobile version carries the same content as desktop
  • Overlays do not cover content immediately on load

If this block is red, do not fix everything at once: LCP and CLS usually take a day or two, while INP means reworking logic. What each metric measures is in our Core Web Vitals breakdown.

This block is never "done". It is also where the difference lives between a site that plateaus at 300 visits a month and one still growing in its second year.

Block 6 — 10 checks

  • Every page solves one user task and carries one primary call to action
  • Length is dictated by the query, not by a "3,000 words" rule of thumb
  • No two pages competing for the same query
  • An About page with real people, a real address and real experience
  • Each article sends out 3–5 contextual internal links
  • The link profile grows steadily, with no 200-domain spikes
  • A Google Business Profile exists if there is a physical address
  • GA4 and Search Console linked, goal set on enquiries not pageviews
  • Rankings tracked by query group, not individual phrases
  • Broken links and new 404s reviewed at least quarterly

The steady-link-profile item is the only one on this list you can satisfy while actively harming the site. Buying 200 domains in a month technically produces growth, and adds real risk. Which links still move rankings, and how to vet a donor before paying, is covered in our guide to safe link building.

A checklist does not do SEO. It only shows how many places you are currently losing to a competitor who ran the same list earlier.

Frequently asked questions about the SEO checklist

How long does a full pass through all 60 points take?

For a site under 100 pages, roughly two working days, assuming access to Search Console, analytics and the codebase. For a store with several thousand URLs it is a week or more, because indexation and structure need a crawler rather than hand-checking. Fixing what you find always takes longer than finding it.

Where do I start if almost everything is a "no"?

Start with indexation: a page outside the index earns no traffic however good the copy is. Then close out duplicates and canonicals, then mobile speed. Touch content and links last — they only compound on top of a healthy technical foundation.

Is running the checklist once enough?

No. Indexation and speed break during releases, so re-check them after every site update. Structure and keyword mapping are reviewed annually, while content, links and analytics need continuous attention. A practical rhythm is a full pass every six months plus a quick check after each release.

Can I work through this list without a developer?

About half the items can be handled from the site admin panel: titles, descriptions, alt text, internal links, content. Indexation, speed, caching, structured data and server settings need code access. The sensible split is the owner covering content and a developer covering the technical side.

How soon do fixes show results?

Indexation errors pay off fastest: pages usually appear in search within one to three weeks of a recrawl. Better titles and snippets move CTR within two to four weeks. Structural changes and new content work more slowly, generally three to six months.

In short

  • Order beats completeness: indexation first, then structure, then everything else.
  • Around 80% of real problems are closed by the first two blocks — twenty items.
  • Measure speed with field data from mobile users, never a single lab run.
  • Blocks 1 and 4 break during releases — re-check them after every deploy, not once a year.
  • Four or more "no" answers in the technical block means you need a full technical audit, not the rest of the list.
Share
  • SEO
  • checklist
  • audit

Related reading