The problem
SEO content is one of the highest-demand services marketing agencies offer — and one of the most difficult to deliver profitably. Clients want a consistent output of well-researched, properly optimised articles that rank for relevant keywords and actually drive organic traffic. Delivering this at scale requires keyword research, competitor analysis, outline development, writing, on-page optimisation, and editorial review. For a single 1,500-word article, the total process can take four to six hours from brief to publication-ready draft.
Multiply that across a client base where five or ten clients each want four articles per month, and SEO content production quickly becomes the department that demands the most resource and generates the most bottlenecks. The junior writers you can afford to hire do not have the strategic SEO knowledge to produce good work without extensive briefing and editing. The senior strategists who have that knowledge are too expensive to spend their time writing.
The AI opportunity in SEO content is substantial but requires care. AI can write fluently and quickly, but it can also produce content that is generic, factually uncertain, or structurally weak from an SEO perspective if not properly directed. The solution is not to simply ask AI to write an article and publish it unedited. It is to use AI at every stage of the process — research, outlining, drafting, and optimisation — while keeping human judgement in the loop at the decisions that actually determine whether content ranks and resonates.
The system
Step 1: Keyword and topic research (Perplexity)
Start with Perplexity to understand the topic landscape before touching any keyword tool. Perplexity gives you a synthesised view of what is currently being discussed and ranked around a topic, including the angles, subtopics, and questions that are generating engagement.
Run queries like:
"What are the most important questions people have about [topic] in [year]?" "What are the top-ranking articles about [topic] covering, and what are they missing?" "What are the common misconceptions or underserved angles in content about [topic]?"
Use the outputs to identify content opportunities that are not already saturated — angles where you can produce something genuinely better than what is currently ranking.
Combine this with your preferred keyword research tool (Ahrefs, Semrush, or similar) to validate search volumes and competition levels. Document your target keyword, secondary keywords, and the search intent (informational, navigational, transactional) for each article.
Step 2: Create a detailed article brief (Claude)
Before drafting, produce a structured article brief that specifies everything the AI needs to produce a well-structured, properly optimised piece:
"You are an SEO strategist creating a content brief for a marketing agency. The target keyword is [keyword]. The client is [describe client and their expertise]. The target audience is [describe]. The search intent is [informational/transactional/navigational]. The top three competing articles for this keyword are about: [briefly describe what ranks]. Write a detailed article brief including: recommended title (include the keyword naturally), recommended meta description (under 160 characters), target word count, proposed H2 and H3 structure, key points to cover under each heading, 3 to 5 secondary keywords to include naturally, a note on what angle would differentiate this from existing content, and any specific expertise or data points the client can contribute. UK English."
This brief becomes the instruction set for the drafting step.
Step 3: Generate the first draft (Claude)
With the brief in hand, generate the first draft:
"Write a [word count]-word SEO article based on the following brief: [paste brief]. Requirements: Write in a clear, authoritative style suitable for [target audience]. The primary keyword should appear in the title, first paragraph, and at least two subheadings. Use a natural density of secondary keywords throughout. Include: an engaging introduction that hooks the reader and signals the value of the article, clear subheadings using H2 and H3 as specified in the brief, practical examples or actionable insights under each section, and a conclusion with a clear takeaway or call to action. Do not use generic filler phrases. Write as if you have genuine expertise in this subject. UK English."
The output will be a strong structural draft. It will occasionally contain generalisations or lack the specific depth that a true subject-matter expert would bring. This is what the editing stage addresses.
Step 4: Add expertise and differentiation (human review)
This is the critical human step. Review the draft for: factual accuracy (verify any statistics or claims), depth of insight (add specific examples, data, or client expertise that the AI could not know), and differentiation (ensure the article has at least one angle, example, or insight that does not appear in the competing articles you reviewed in Step 1).
Typically this takes 30 to 45 minutes per article. It is where the human skill adds real value — not in generating words, but in elevating the content from "fine" to "genuinely useful and worth linking to."
Step 5: Optimise and prepare for publication (Claude + Notion AI)
Run a final optimisation pass:
"Review the following article for on-page SEO. Suggest: a stronger title if the current one could be improved, any subheadings that should include the target keyword more naturally, internal linking opportunities (the client's site covers [describe main topic areas]), and a more compelling meta description if the current one is weak. Target keyword: [keyword]. Here is the article: [paste]."
Use Notion AI to write the social media promotion copy for the article — a LinkedIn post, a Twitter summary, and an email newsletter teaser — so the content gets maximum reach after publication.
The results
Before this workflow, a 1,500-word SEO article took four to six hours from keyword research to publication-ready draft. With this system, the same output takes 90 minutes to two hours: 20 minutes of research, 20 minutes of brief creation, 10 minutes of AI generation, 45 minutes of human review and enrichment, and 15 minutes of optimisation and publication prep.
An agency doing 40 articles per month across ten clients would previously have needed two to three full-time content writers. With this workflow, one experienced content strategist can manage the same output — reviewing, enriching, and directing AI generation rather than writing every word from scratch. Content quality typically improves alongside productivity, because the human time that was previously consumed by drafting is now invested in the insight and differentiation that makes content rank and retain readers.