Website optimisation: how to scale organic growth

"Website optimisation" covers three quite different jobs that get treated as one: making the site findable, making the content worth finding, and making the visit convert. They interact, and the order you tackle them in determines how much of the budget is wasted.

The order that works

1. Technical. If search engines cannot crawl, render and understand the site, nothing else has any effect.
2. Content and structure. Once found, the pages have to be the best available answer.
3. Conversion. Once read, the visitor needs an obvious next step.

Most programmes run these in reverse — redesigning the call-to-action button on a page that is not indexed. The sequencing is the single cheapest improvement available to most sites.

Technical foundations

Indexing before anything else

Check Search Console's index coverage first. Every site we look at has surprises: pages excluded that should be included, staging environments indexed, parameter URLs creating duplicates, orphaned pages nobody links to.

This is dull work with an outsized effect, because a page that is not indexed earns nothing regardless of how good it is.

Core Web Vitals

Google's thresholds for "good" are LCP at or under 2.5 seconds, INP at or under 200 milliseconds, and CLS at or under 0.1 (web.dev). Measure on field data rather than lab scores — real users on real connections are what counts.

The usual culprits are unoptimised images, render-blocking scripts, and layout shift from elements loading without reserved space. Setting explicit width and height on images fixes more CLS problems than any other single change.

Mobile

Most B2B research now happens on mobile even when the purchase does not. Check the mobile experience on an actual mid-range phone rather than a desktop emulator; the difference is routinely alarming.

Structured data

Schema.org markup tells search engines and AI systems what your pages are about explicitly rather than by inference. Organisation, Service, Article, FAQPage and BreadcrumbList cover most needs. It is also the foundation of AI visibility.

Content and structure

Topical authority through clusters

One thorough page per question beats five thin pages competing with each other. Group related queries into clusters, build a substantial page for each, and link them deliberately to the commercial page they support. See our keyword research guide.

Find the gaps before writing anything

Compare what your buyers search for against what your site answers. The gaps that matter are the ones with commercial intent, not the ones with the biggest volume.

Internal linking

The most underused lever on most sites. Links from strong pages to the pages you want to rank pass authority and tell search engines what matters. It costs an afternoon and is frequently worth more than a month of new content.

Conversion

Once the traffic arrives, three things determine whether it converts:

Clarity. Can a visitor tell within five seconds what you do and whether it applies to them? Most B2B homepages fail this.

Friction. Every form field costs conversions. Ask for what you need to respond usefully and nothing else.

Proof. Specific outcomes with real numbers. Vague testimonials persuade nobody.

On low-traffic sites, skip formal A/B testing — you will not reach significance — and do the qualitative work instead: session recordings, watching a real person attempt the task, and removing the steps that visibly confuse them.

Measuring it properly

Track three things monthly:

• Index and technical health. Coverage, Core Web Vitals, broken links. Regressions here are cheap to fix and expensive to miss.
• Cluster visibility. Aggregate position and clicks per cluster, not individual keyword bounce.
• Conversions by landing page. Which pages earn their existence.

Then act on the pattern: pages at positions five to fifteen are the cheapest next month of work, and technical regressions get fixed before anything else.

Making it continuous

A one-off optimisation project decays. Redirects accumulate, pages get orphaned, a developer change breaks something six months later and nobody notices until rankings drop.

What works is a standing process: continuous technical monitoring, a monthly review of what moved, and a prioritised queue that gets worked through rather than a report that gets filed. That is the difference between a project and an organic growth engine.