AI offers transformative ways to streamline the delivery of SEO projects and increase the overall quality of our output, if it is done correctly. However, there is a precarious, fine line between AI adding genuine value and it being used as a shortcut without clear guardrails. At its absolute worst, a lack of caution with AI adds to the ever-growing mountain of “AI slop” on the web: content that is thin, repetitive, and generated without appropriate human oversight.
The philosophy I’ve learnt is that AI should be a lever for efficiency, not a replacement for expertise. Over the last few years, I have refined a few workflows that leverage AI to enhance everything from internal documentation to technical analysis and high-level content production.
Here is an overview of 10 ways I have found that work when leveraging AI in SEO, focusing on converting internal documentation into professional business artefacts, achieving successful content production, and automating those low-value-add processes that traditionally eat up an SEO’s billable hours.
Business Artefacts & Internal Documentation
One of the most immediate benefits of AI in a professional setting is its ability to handle internal, non-customer-facing documentation. These documents are critical for project success, alignment, and accountability, yet they are often ripe for automation because they follow predictable, narrative structures.
1. Writing Project Scope Documentation
I recently joined the AI SEO Show (you can catch the full episode here) and discussed my affinity for Amazon-style documentation in their 1-pager and 6-pager formats. These narrative-heavy documents are shared in advance of meetings to ensure stakeholder discussions remain structured and effective. However, drafting documents from scratch is time-consuming.
AI makes a great partner for this process. It can take a project plan in bullet point form and synthesise it into a professional, cohesive business case. While it is automated, it remains critical to avoid basic errors that could lead to reputational damage, much like the high-profile errors by consulting firms like Deloitte recently. To use AI for your project scoping, try these steps:
- Write your project outline in bullet points or note form, ensuring you cover all key technical and business points.
- Provide your LLM of choice with a template for your 1-pager or 6-pager so it understands the required structure.
- Feed in simple spreadsheets or tables with key data points (such as traffic projections or keyword difficulty) for the AI to reference.
- Prompt your LLM to generate the document. While a standard instance works well, I often recommend using a custom GPT or a Gemini Gem to build in your specific style guide or reference previously completed documents for consistency.
- Thoroughly proofread and edit the output to ensure the narrative aligns with your actual goals.
2. Drafting Standard Operating Procedures (SOPs)
Standard Operating Procedures are the backbone of any scalable SEO operation. They ensure tasks are completed successfully and consistently. As an SEO consultant, I frequently provide these to my clients, particularly those building a new internal SEO function, so they can execute recommendations and build their own team’s skillset.
I also write down my own SOPs for internal use. This ensures that when I return to a task I haven’t performed in months, I don’t have to rely on memory. It is the same reason why pilots use checklists for every phase of their flight, even if they have performed that exact maneuver hundreds of times.
You can use AI to turn a quick recording or a list of steps into a polished manual by following these steps:
- Bullet point the key technical steps involved in the task.
- Give your LLM a generic SOP document template for reference.
- Prompt the LLM to transform your raw bullet points into the structured, templated format.
- Proofread and edit the final document for technical accuracy.
3. Streamlining Technical SEO Reporting
Technical SEO reports are another area where AI can save hours of manual formatting. However, a major caveat is necessary here: this should only happen after the website has been fully analysed by a human with technical SEO expertise. You must never use an LLM to provide the primary analysis or rely on it to determine priorities or impact assessments; it simply doesn’t have the context of the business’s specific technical stack or commercial goals, and can often overstate issues.
My typical workflow involves using Screaming Frog to crawl and analyse a site. I export those reports and conduct a manual review of the data. I then present findings in a table that includes impact assessments (High/Medium/Low) and effort requirements (High/Medium/Low). While this spreadsheet is the “source of truth,” many of my clients prefer a more digestible narrative report that summarises the key issues. This is where the LLM can drastically speed up delivery.
- Complete your technical SEO audit and compile all recommendations into your main spreadsheet.
- Provide the AI with a templated technical SEO report that corresponds to the key elements you want pulled out, such as descriptions of issues, impact assessments, and a summary of key initiatives.
- Prompt your LLM to generate the narrative report based on your specific data and template.
- Finalise with a manual proofread and edit to ensure the tone is professional and the technical details are correct.
High-Quality Content Production
Content writing is perhaps the most common use case for AI, but it is also where the most “slop” is produced. To stand out, thoughtful human input is essential to ensure uniqueness and quality. One of the routes to a high-quality page is providing the AI with unique source material. Without it, the LLM is just guessing based on existing information on the web, resulting in an average page that probably doesn’t deserve to rank in Google.
4. Strategic Content and Briefing
To create content that actually adds value, I use a variety of “proprietary” source materials in my prompts. For example, I’ve recently started interviewing subject matter experts (SMEs) and transcribing their unique insights using AI. This is an approach championed by Mark Williams-Cook and his agency, Candour (you can learn more about this process in this podcast episode, starting at the 30-minute mark). For the clients I’ve followed this approach with, it’s made it much quicker for them to provide input, and I believe it results in better content.
I also include product information documents so the LLM can bridge the gap between technical features and user benefits. For a recent health drink project, I set up a custom Gem in Gemini for that client that included specific product ingredients and benefits that they provided. This allowed the AI to generate a series of pages explaining the health benefits of the product category without hallucinating ingredients or benefits that didn’t apply.
To produce high-tier content, I follow these steps:
- Prepare your source material: this should include SME interview transcripts, previous content for brand voice, ICP information, and “People Also Asked” data from tools like AlsoAsked.com to ensure you are answering real-world queries.
- Provide a brand style guide and specific constraints, such as minimising the use of em dashes or specific sentence structures, to tune out common LLM writing traits that users are becoming attuned to
- Craft a prompt that contains details of the topic and key points, referencing all the resources you’ve gathered.
- Finish the page with rigorous human editing for style and, most importantly, fact-checking.
5. Documenting Ideal Customer Personas (ICPs)
An extension of content production is using AI to document your Ideal Customer Personas. This ensures your keyword research and clustering efforts are relevant to what your customers are actually searching for, preventing wasted effort on pages your audience will never read. Detailed ICPs also help you speak directly to customer pain points and can be referenced in project proposals to build a stronger business case for your SEO initiatives.
- Use your LLM to help you draft interview questions designed to probe for customer needs and experiences.
- Run an interview session with customer advocates, sales, and support teams, ensuring you record and transcribe the call.
- Create an ICP information document template.
- Prompt your LLM to review the interview transcript and use it to complete the template.
- Review and fact-check the resulting persona with the stakeholders who took part in the interview.
6. Generate Article Title Ideas
Coming up with a compelling title for an article can be difficult. A very simple LLM use case is to generate title ideas to take inspiration from.
I typically do this after I’ve produced the content, and already have confidence that the page captures the right search intent and provides helpful guidance. I’ll then provide the document to the LLM and ask it to identify three title ideas that capture the aim of the content.
Sometimes I’ll use one of these titles, but more often than not, the three suggestions help to spark some new ideas, whether by combining a few different ideas from my LLM or leading me on to something different.
Outside of the world of SEO, I use this approach each week to write match report headlines for the field hockey team I manage. Whether I’m trying to optimise for a high-volume search term or just trying to encourage match report views in our club group, the logic remains the same: use the AI as a creative partner to explore angles you might have missed while being too close to the project.
7. Identifying AI Platform Visibility Opportunities
Once you have your ICP information finalised, you can use it to identify how your brand might appear in AI-driven search platforms like ChatGPT, Perplexity, or Gemini. LLMs are particularly good at this because their training data includes vast amounts of user-generated content from places like Reddit, where real users ask candid questions.
To explore this, I often use a simple but effective prompt: “If I was [persona] trying to find [keywords], what might I ask?”. I repeat this for all my ICPs to generate a map of long-tail, conversational queries that we should optimise for. For a deeper dive into this concept, I highly recommend watching Mark Williams-Cook’s “RAG to Riches” talk, specifically from the 7-minute mark.
Automating Repeatable SEO Tasks
There are several “bread and butter” SEO tasks that are incredibly repetitive but necessary for on-page performance. These are perfect candidates for AI automation.
8. Generating Meta Data
Meta data is incredibly easy to automate once the page content is finalised. For meta descriptions, I provide the LLM with the finished page content and ask it to summarise it into a description of up to 155 characters. For meta titles, I usually prompt for several different versions and provide the primary keyword I want included to give me a range of inspiration.
9. Building FAQs for Visibility
Even though Google reduced the prevalence of FAQ snippets in 2023, FAQs remain a powerful tool. They provide easily scannable answers for users, which is vital since most people skim-read on the web. Furthermore, FAQs are a structured format that LLMs understand with ease, and when supported by FAQ schema, they help present information clearly to all search engines. You can read more about the technical benefits in Anthony Barone’s article on FAQ Schema.
Get started with generating FAQs using these steps:
- Provide the LLM with your page content so it draws answers directly from your text.
- Specify the FAQ questions you want answered, or ask the AI to generate some ideas based on your ICP.
- Prompt the LLM to provide the answers from your content and specify a maximum character length.
- Review and fact-check every response before publishing.
10. Writing Python Scripts for SEO Automation
One of my most rewarding uses for AI has been using it as a “junior developer” to write Python code. While I had dabbled in coding for years, I was never able to advance beyond the basics due to the time commitment required. With an LLM, I can now build functional tools quickly and join the “vibe coding” trend.
I recommend copying any AI-generated code over to a Jupyter Notebook or a Google Colab document rather than running it within the LLM interface. This allows you to share the code with others and ensures you aren’t tied to an LLM subscription just to run your script. Some of the scripts I have built or adapted include:
- LLM Prompt Builders: Scripts that automatically compile source materials for article briefs, saving me significant time in the research phase.
- Data Wrangling: Simple scripts for tasks like combining multiple CSV files (this was actually the very first thing I built with Python)
- Internal Linking: Sourcing internal linking opportunities by analysing content similarity (inspired by scripts from Jonathan Boshoff and Daniel Heredia).
- International SEO: Translating metadata and headings to identify relevant hreflang pairings across different languages or mapping redirects for large site migrations.
If you want to turn an LLM into an effective coding partner, I suggest reading Addy Osmani’s “Prompt Engineering Playbook for Programmers”.
Wrapping Up: The Human Element
LLMs are incredibly useful and can significantly accelerate your work, but it is crucial that you remain curious and vigilant about what they generate. Human oversight is critical; many high-profile AI failures are simply the result of an operator failing to think critically about the output.
Do not believe the hype that LLMs can create viable, high-quality output in a single click. You will always need to edit, refine, and above all, fact-check everything. This list is by no means exhaustive, and the possibilities are endless if you approach the technology with a problem-solving mindset. What other uses would you add?