On-page SEO optimisation

The pages that are already close

Nobody buys this because they read about heading tags. They buy it because their pages exist, they are indexed, and they sit on the second page of results where nobody goes. Or because the position looks respectable and the clicks still are not arriving. The feeling is always the same one — the content is there, so why is nothing happening.

The useful thing about that situation is that most of the hard part is already done. A page with impressions has been found, understood and judged relevant enough to be shown. It is not being ignored; it is being narrowly beaten. That is a much cheaper problem to solve than building relevance from zero, and it is why on-page work is usually the fastest return in the whole discipline.

The other half of the job is less obvious. Pages very often rank for something other than what their owner intended. A service page picks up a question, a blog post picks up a product term, and nobody notices because nobody reads the query report. Strengthening a page towards what it is actually earning impressions for tends to beat forcing it towards what it was named after.

How the work runs

1. Pull the real data and pick the targets. Search Console query and page reports for the last three months, filtered to exclude branded searches — searches for your company name inflate every average and tell you nothing about whether the work is landing. From what is left we build two lists: pages in the striking-distance band around positions eight to twenty, and pages with high impressions and unusually low click-through. Half a day to a day.
2. Read the SERP for each target query by hand. Two minutes per query, in a clean browser window, looking at what Google is actually assembling — whether an AI summary sits above everything, whether a map pack or shopping row occupies the top, and what kind of page fills the first five results. This decides whether the page needs to become a guide, a comparison, a product page or something else, and no tool reports it. One to two days across the target set.
3. Rewrite the on-page elements page by page. Titles under about sixty characters aimed at the dominant real query, descriptions under about a hundred and fifty-five with an actual reason to click, one clear H1, and a heading structure that covers the sub-questions visible in the query list and in the related-questions box. A few days to two weeks, entirely depending on page count.
4. Add internal links, with the anchor text chosen deliberately. Every target page gets links from three or four genuinely related pages, using anchor text that describes the destination rather than saying more or read here. This is the least visible part of the job and frequently the part that moves position. One to three days.
5. Run the cannibalisation check and re-measure. Confirm no two pages are now aimed at the same query, then hold the before-and-after query data side by side after the changes have had time to register. Half a day, plus a follow-up.

What moves the price

The range is $300–$2,500 across 1–3 weeks, and it is set by three things rather than by any judgement about how competitive your market is.

How many pages are in scope. This is the main driver and the reason the range is wide. Ten important commercial pages is a contained piece of work. Two hundred product pages is a production process with templates, batching and quality checks, and it is priced as one.

Whether the writing is included. Restructuring what exists is on-page work. Producing new sections, new copy or a full rewrite is content work, and it belongs in a different line of the scope with its own fee. Blurring the two is the most common source of disagreement on this service, so it gets settled before anything starts.

Whether we can edit directly. With CMS access we apply and verify each change. Without it we produce a specification for somebody else to implement, which adds handover time, adds a round of checking, and adds the risk that half of it is never applied.

What usually goes wrong

The page is optimised for the keyword somebody wants rather than the one it earns. An owner decides the page is about a term, the page is rewritten around that term, and the query report was never opened. Very often the page had been quietly collecting impressions for something adjacent and more commercial, and the rewrite walks away from the one thing that was working.

The keyword gets repeated two or three times in the title. It reads like a warehouse label, real people scroll past it, and the click-through rate falls. Repetition does not buy position, and the lost clicks are a real cost. A title that a human would choose beats a title built to satisfy a checker.

A high position with no clicks gets treated as a ranking failure. A page can sit near the top and still collect almost nothing, because the result page above it answers the question outright. Search Console has no filter that reveals this, so the query has to be searched by hand and looked at. The right response is usually to keep the page and move effort to queries where clicking is still how people buy — not to keep rewriting a title that was never the problem.