Guide

New Website SEO Guide

Last updated

Building a new website is a different problem to migrating an existing one. There is no ranking history to protect, no redirect map to build, and no baseline traffic to measure against. That freedom comes with a cost: every structural decision is made from scratch, and the wrong ones compound over time. A URL structure that made sense for twenty pages becomes a problem at two hundred. A CMS chosen for design flexibility can become a ceiling on technical SEO control.

This guide covers the planning decisions that determine whether a new site can rank, before a page is published. For the technical launch checklist, see the SEO Go-Live Checklist.

Choosing a domain

The domain name has a modest direct influence on SEO. Keyword-rich exact match domains carry a small contextual signal but no meaningful ranking advantage. A branded domain builds more long-term authority and avoids the low-quality associations that over-optimised domains carry.

The most important domain consideration for SEO is history. If you are registering a brand new name, there is nothing to check. If you are buying an existing or previously registered domain, check it before completing the purchase:

  • Run site:yourdomain.com in Google. If the domain has been active but returns no results, it may have been de-indexed for a manual penalty.
  • Check the backlink profile in Ahrefs or Semrush for spam links, link farms, or unrelated foreign-language sites.
  • Review the Wayback Machine to see what content previously lived on the domain.
  • Search for the domain name in Google to surface any spam reports or blacklist mentions.

A penalised domain passes that penalty to you. Recovery is possible but slow, and it works against the new site’s momentum from launch.

For TLD selection, .com remains the default for most brands. Country-code TLDs (.co.uk, .de, .com.au) send geo-relevance signals that help for local or national targeting.

Hyphens in a domain are not a ranking problem, whatever you may have read. Google has said its algorithms do not treat a hyphenated domain as low quality, and the “hyphen filter” is a twenty-year-old myth.1 The genuine caution is different and worth separating out: John Mueller has said he is not a fan of keyword-keyword domains, because they are associated with spam, they date badly, and they box the business in if it later changes focus.1 So blue-widgets-cheap.com is a poor choice for what it says about you, not for its punctuation. A hyphen in a genuine brand name is fine, though it is still harder to share verbally.

Full detail on domain name decisions is in choosing a domain name.

Choosing a CMS

The CMS determines your level of technical control over the site. Before choosing a platform, confirm it gives you direct control over:

Any platform that hides or restricts these controls limits what you can do with SEO. The most common options:

WordPress offers the most flexibility and the widest range of available plugins. SEO is controlled through plugins (see below). The main trade-off is maintenance: WordPress requires hosting decisions, security updates, and plugin management.

Webflow produces clean HTML output with strong native SEO controls and fast managed hosting. Well-suited to marketing sites where design control matters. Less flexible for complex functionality.

Shopify is built for ecommerce. It handles product pages and transactions well but limits content marketing flexibility and has less granular SEO control than WordPress.

Static site generators (Astro, Hugo, Next.js) deliver the fastest page loads and highest Core Web Vitals scores. They require technical knowledge to set up and maintain, and suit developer-run sites.

Platform choice should be driven by the site’s primary purpose and who will maintain it. A well-configured WordPress site will outperform a poorly-configured Webflow site, and vice versa.

SEO plugin setup (WordPress)

On WordPress, an SEO plugin controls the most important technical SEO settings site-wide. The two dominant options are Rank Math and Yoast SEO.

Rank Math has a stronger free tier (including schema types and Search Console integration that Yoast reserves for its paid version) and a lighter codebase. Its PRO plan covers unlimited websites you own; client sites need the pricier Business or Agency tiers, so the “unlimited” framing does not apply to agencies managing sites for others.2 For most new single-site projects it is the better default.

Yoast SEO has a simpler interface and more beginner-friendly documentation. If the person managing the site is new to SEO and benefits from step-by-step guidance, Yoast’s setup process is more approachable.

Either plugin handles the essentials: title tag and meta description templates, XML sitemap generation, canonical tags, robots meta, and basic schema. Set it up before publishing any content so templates apply consistently from the first page.

Planning site architecture

Site architecture is the hierarchy of pages and how they link to each other. For a new site, this can be planned before anything is built, and should be: restructuring after content has accumulated is expensive.

Keep depth flat. The most important pages should be reachable within three to four clicks from the homepage. Pages buried deeper receive less crawl budget and less internal link equity.

Plan URL structure before building. Decide on the hierarchy first: /category/sub-category/page-slug/. Use lowercase, hyphens to separate words, and descriptive slugs. Avoid dates in URLs for evergreen content; they create pressure to update the URL when the content is revised.

Plan for scale. A flat structure that works at twenty pages can break at two hundred if the category logic is not clear from the start. Map out the full topic hierarchy before building, even for pages that do not exist yet.

Group content by topic. Pages covering a theme should sit under a shared parent. This concentrates internal link equity and signals topical depth to Google.

For a full guide to site architecture principles, see Site architecture.

Establishing the site as an entity

A new site has no brand signals. Google has no record of the business, no Knowledge Graph entry, and no way to verify whether the entity is legitimate. Building those signals early reduces how long it takes the site to earn trust.

Start with a canonical description: one clear, accurate paragraph that describes the business. Use it consistently across every external profile. Then create profiles on:

  • Google Business Profile (required for local businesses; useful for any brand)
  • LinkedIn company page
  • Wikidata (an openly licensed structured record of the entity, widely used as a reference by systems that need to resolve who you are)
  • Crunchbase (for B2B and technology businesses)
  • Industry-specific directories relevant to the sector

These profiles do double duty. Corroboration across independent sources is how AI systems resolve an entity as well as how Google does, which is why citations and directory presence carry more weight for AI visibility than they do for the traditional local pack.

Add Organisation schema to the homepage with sameAs properties pointing to each profile. This tells Google that the entity on your site is the same one found on those external sources.

Set these up before or immediately after launch. Entity signals take time to register, and earlier is better. For implementation detail on Organisation schema, see Structured data.

Decide who is allowed to crawl you

A new domain arrives with crawler defaults nobody on the project chose. They are worth one deliberate look, and then not worrying about.

The decision that matters is which categories of AI crawler you allow, and for most new sites the answer is “all of them, for now”. Blocking training crawlers is a legitimate editorial choice, but it buys you nothing in search and costs you nothing to defer.

Two specifics if the site is behind Cloudflare. From 15 September 2026, newly onboarded domains get the Training and Agent categories blocked by default on pages that display ads, while Search stays allowed.3 If the site carries no advertising, the default does not touch it, and because Search crawlers remain allowed either way, AI citation visibility is unaffected in both cases.

The thing to actually avoid is the opposite mistake. If you choose to block the Training category at Cloudflare, that also blocks Googlebot, Applebot and Bingbot, because multi-purpose crawlers are judged on everything they do.3 A well-meaning decision to stay out of model training can remove a brand-new site from Google, which is the one outcome a new site cannot recover from quickly. If keeping out of training data is the goal, name the training crawlers in robots.txt instead.

Sort the crawlers by what they do, not by whose logo is on them:

  • Training (GPTBot, ClaudeBot): content used to train or fine-tune a model
  • Search (OAI-SearchBot, PerplexityBot): indexing that produces citations in AI answers
  • Agent (ChatGPT-User, Perplexity-User): fetching a page in real time on a user’s behalf

For most new sites the sensible default is to allow search and agent crawlers, and to treat training as a genuine editorial choice. The robots.txt reference covers the syntax.

Content strategy for a new site

A new site has no ranking history, which means no trust buffer and no established topical authority. The content approach needs to reflect that.

Target low-competition topics first. Competing for high-volume, high-difficulty keywords against established domains in the first six months rarely produces results. Start with specific, answerable questions where the competition is weaker and build outward once topical authority is established.

Depth over breadth. A site with ten thorough articles on one topic will outperform a site with fifty shallow articles across five topics. Google’s ability to recognise topical authority depends on seeing a cluster of related content, not isolated pages.

Do not publish stub pages. Placeholder pages with thin content set a low-quality baseline that is harder to recover from than simply not publishing yet. Publish when the content is complete.

Answer directly. New sites often earn AI citations before they earn traditional rankings. A page that answers a specific question clearly and completely can appear in AI-generated answers without a top-ten ranking. Write for direct answers, not around them.

Launch and post-launch

The technical launch steps (removing crawl blocks, submitting the sitemap, verifying canonical tags, setting up GSC and GA4, checking Core Web Vitals) are covered in the SEO Go-Live Checklist.

Once live, the first four to eight weeks are an observation window. Use URL Inspection in GSC to request manual indexing for your highest-priority pages. Do not request indexing for utility pages (contact forms, thank-you pages, login pages); they carry no ranking value and spending your quota on them dilutes the signal at the point when it matters most.

From day one, monitor GA4’s AI Assistant channel (traffic from ChatGPT, Claude, Gemini, Perplexity, etc.). New sites often earn AI citations before traditional rankings, so tracking this early gives visibility into content performance across different discovery surfaces.

Expect rankings on competitive terms to be slow. The pattern for new sites is: indexation in the first few weeks, initial rankings on long-tail queries within one to three months, competitive rankings taking six months or longer depending on domain age, content quality, and backlink acquisition.

Footnotes

  1. Google Says Hyphens In Domain Names Not Considered Low Quality — Search Engine Roundtable ↩ ↩2

  2. Rank Math pricing ↩

  3. Your site, your rules: new AI traffic options for all customers — Cloudflare ↩ ↩2