Copywriting

SEO copywriting

Content that ranks and content that reads well are not in conflict. The work is finding the phrases people actually use, then writing something worth landing on.

Two audiences, one text

An SEO text has to satisfy two readers. A person, who decides within seconds whether the page is worth their time, and a machine, which decides whether the page is worth showing at all.

For a long period these pulled in opposite directions, which is why so much SEO content from that era is unreadable: the same phrase repeated to hit a target density, headings written for crawlers, three hundred words of introduction before the actual answer. Search engines have moved on. They now approximate topical coverage and usefulness well enough that writing badly for the algorithm mostly costs you the reader without buying anything in return.

Research decides more than writing

The most common reason content fails in search is not the quality of the prose. It is that the page targets a phrase it was never going to win, or a phrase nobody searches, or the same phrase as another page on the same site.

Before writing we work out what people actually type, at what volume, and what they want when they type it. Someone searching a general term is reading; someone searching a specific product with a qualifier is buying. Those two need different pages, and giving them the same page serves neither.

Then we look at what already ranks. If the first page is occupied by large publishers, marketplaces and sites with far more authority than yours, and the phrase does not lead towards revenue, we will say the target is not worth pursuing. Recommending a page that will settle in the forties is easy money and bad advice.

Structure is what the machine reads

Once the text is written to be useful, structure makes it legible to software.

That means headings that describe the content beneath them, an answer placed near the question that introduces it, sections that hold together when read on their own, and meta titles and descriptions written to earn the click rather than to repeat the H1. It means internal links that connect supporting articles to the pages you actually sell from, so authority accumulates somewhere useful. Where relevant, it means structured data, so a machine can tell what type of thing the page describes without inferring it.

None of that damages the reading experience. Most of it improves it.

Search is no longer only search

An increasing share of queries never produces a click. The user asks an assistant or reads a generated summary, and the answer is assembled from sources the model considers reliable. Being one of those sources is now part of the job.

The good news is that it demands very little that good writing did not already demand. Language models quote content that states things plainly, defines terms, answers directly and organises itself into sections that survive being extracted from context. Padding, throat-clearing and vague claims are as unhelpful to a model as they are to a person. The practical adjustments are small: put the direct answer immediately under its heading, keep definitions unambiguous, and make sure each section reads sensibly on its own.

Limits, and how we fit into your setup

It cannot fix a slow site, a broken information architecture or a domain with no authority. Content is one input among several, and on a technically damaged site it underperforms regardless of quality.

It also cannot deliver a guaranteed position. Anyone offering one is either counting on phrases so obscure that ranking is trivial, or hoping you will not check. What content can do is make a page genuinely deserve the position, which is the only durable part of the process, and give you a clear read on which targets are realistic before you spend anything on them.

Because content is only one input, it usually sits alongside other work. If you already have an SEO agency or an internal specialist, we fit into that. Usually they own technical work, links and strategy, and we produce content against their keyword map, or write the briefs if that helps. What we want to avoid is two teams targeting the same phrases from different documents, which happens more often than anyone would like.

What you get

Keyword and intent research

The phrases people use, grouped by what they are trying to do: learn something, compare options, or buy. Intent decides the format as much as the topic does.

Content briefs

Per-page briefs covering the target phrase, the questions to answer, the structure and the internal links, so writing is not guesswork.

Optimised page copy

Headings, opening paragraphs, body and meta fields written together, using the search language naturally rather than inserting it afterwards.

Content structured for AI answers

Clear definitions, direct answers placed near their questions, and self-contained sections. That is the format language models quote from most reliably.

Internal linking plan

A deliberate structure connecting supporting content to the pages that make money, rather than links added ad hoc as articles appear.

Rewrites of underperforming pages

Diagnosis and repair of pages that get impressions but no clicks, or rank just below the point where traffic starts.

How we work

  1. 01

    Audit what you already have

    We look at which pages get impressions, which get clicks and where you rank just outside the useful range. Often the fastest gains are in content that already exists.

  2. 02

    Research phrases and intent

    We map the phrases worth targeting, check what already ranks for them and judge honestly whether the page is winnable. Some are not, and we say so.

  3. 03

    Plan the pages and avoid overlap

    Each target phrase gets one page. Two pages chasing the same query compete with each other and both underperform, which is a common and avoidable problem.

  4. 04

    Write for the reader, structure for the crawler

    The text is written to be useful, then structured (headings, answer placement, internal links, meta fields) so machines can parse what it covers.

  5. 05

    Measure and revise

    After publication we track positions, impressions and click-through in Search Console, and revise the pages that are close but not converting attention into visits.

Tools and technology

Where a solid open-source tool exists, we choose it over a closed one. No lock-in to a single vendor, and costs you can actually predict.

  • WordPress
  • Ghost
  • Next.js
  • Astro
  • Schema.org
  • LanguageTool
  • Matomo
  • Sanity
  • Google Search Console
  • Google Analytics 4
  • Google Keyword Planner
  • Ahrefs

Frequently asked questions

More services in this category

Read testimonials from companies that trusted us

They're always a few steps ahead.

Mikołaj

CEO & Founder, GBS®

View on Clutch
GBS® logo

Delivered well ahead of the deadline.

Yasniel

CEO, IMEGA Sp z o.o.

View on Clutch

ZanReal's individual approach is impressive.

Adam

Executive, w-studio.pl

View on Clutch

Knowledge and business intuition make them a valuable partner.

Magda

Designer, DIGITALUNI

View on Clutch

Quick solutions that reduced costs by 99%.

Andrei Kapytau

Team Lead, busel.uk

View on Clutch

+20% deliverability for our email campaigns.

Joan Calabria

Sales Director, 36NORTH

View on Clutch
36NORTH logo

Latest

Technical guides, security insights, and what we've learned building with AI.

Is your content ranking for the phrases customers use?

Message us

Point us at the pages that matter most. We will check how they perform today and suggest where new or rewritten copy will make the biggest difference.

Zanek

Can't keep up with changes in AI world?

Let us do the heavy lifting. Every week we distill the most important AI developments into a focused 5-minute briefing — so you stay ahead without the noise.

Find out more
Weekly AIonline