The complete SEO guide

Three steps decide whether you appear: crawling (a page is requested), indexing (it is stored as eligible), and ranking (it is chosen for a query). Each step can fail independently.

Most site problems are step-one and step-two problems presented as step-three complaints. "We do not rank" is often "we are not indexed".

This is where the cheapest wins live, because the failures are unambiguous.

• Status codes: no errors on pages you want found; broken links repaired or redirected to the closest equivalent.
• HTTPS everywhere, with no mixed content.
• Canonical tags pointing at the page itself unless a duplicate genuinely exists.
• Robots.txt that blocks only what should be blocked — it controls crawling, not indexing.
• An XML sitemap containing canonical, indexable, live URLs only.
• Mobile rendering with a real viewport, plus reasonable performance and layout stability.

Verify this by crawling the site and reading what came back, not by assuming what the CMS does.

Produce a map, not a list: one intent per page, expressed in the audience's words. Then check the results page for each query to confirm which format wins.

• Informational: explanations, definitions, checklists.
• Commercial: comparisons, evidence, pricing context.
• Transactional: a clear path to act with minimal friction.

• A unique, specific title front-loading the distinguishing words.
• A meta description that accurately sells the page.
• One H1 stating the subject; H2/H3 used as structure, not styling.
• An opening paragraph that answers the question directly.
• Descriptive alt text on meaningful images, empty alt on decorative ones.

Completeness beats length. Cover the question, the obvious follow-up questions, and the practical detail people need to act. Add what only you have: examples, data, screenshots, worked cases.

Consolidate near-duplicates. Several weak pages on one topic compete with each other and dilute the site.

A cornerstone guide links to detail articles, definitions and questions; those link back. This shows depth and makes discovery reliable.

• No orphan pages.
• Descriptive anchor text.
• Recheck for broken internal links after any restructure.

Organization, WebSite, BreadcrumbList, Article and FAQPage cover most needs. Mark up only what is visible, keep the JSON-LD valid, and use consistent names for your entities.

Editorial links from relevant, credible sources remain a strong signal, and they follow from publishing something worth citing. Support them with visible expertise: named authors, relevant background, sources for claims, real contact details.

Do not invent credentials, awards or statistics. It is detectable and it is not recoverable cheaply.

The same foundations make pages usable by AI answer engines: retrievable HTML, one topic per page, a direct answer early, self-contained sections, accurate structured data, visible authorship.

There is no separate AI checklist worth maintaining — there is a stricter version of this one.

1. Track clicks and conversions per topic, not average position alone.
2. Recrawl after releases: redesigns reliably reintroduce broken links and stray noindex tags.
3. Review your weakest pages quarterly and decide: improve, consolidate or remove.
4. Keep the sitemap honest as pages are published and retired.

1. Remove blockers: errors, noindex, wrong canonicals, orphan pages, broken links.
2. Fix on-page basics on your commercially important pages.
3. Map intents and close the gaps with genuinely useful content.
4. Build the internal link structure between clusters.
5. Only then invest in authority and outreach.

Almost every site that feels stuck in step five has unresolved work in step one.