This is a deep dive from our roundup of 25 AI prompts and workflows that speed up SEO turnaround time. Here we focus on the research and planning stage — the part that decides whether everything downstream is aimed at the right target.
The fastest way to waste a week of content production is to point it at the wrong pages. Most teams reach for AI to write, but the practitioners who cut their turnaround the most use it one step earlier — to diagnose the search landscape and design the site’s structure before a single draft exists. Five of them shared the exact workflows below.
Diagnose search intent before you write anything
Heath Squier’s rule is deliberately counterintuitive: don’t ask the model to write the page first, ask it to diagnose the search pattern. He combines ChatGPT with search data, Search Console exports and the live SERP, then prompts:
“Act as an SEO analyst. Cluster these queries by intent, identify the page type Google is rewarding, note which questions AI Overviews or answer engines are likely to summarize, then give me the content gaps, schema opportunities and internal-link targets.”
The output turns a messy keyword list into a working brief — primary intent, secondary questions, comparison angles, objections, entity terms, and what needs proof — so the blank-page phase disappears and the team goes straight to editorial and technical decisions.
Build a master entity table before the first page
Adam Yong builds a single entity table before writing anything. He feeds in the business name, services and address and asks for a table of entities, services and related topic classes, with columns for the entity, its importance, and its page location. That one table becomes the template for the entire site, so nothing important is missed.
His prompt structure never changes regardless of the tool: role, business details, required output format — in that order. The step he credits most is turning that approved table into a set of questions for each page; because the structure was locked in first, the output stays consistent across 30–60 pages instead of drifting.
Convert strategy into a content taxonomy
Ivy Bernabe uses a custom GPT to convert a marketing strategy into a full content taxonomy — pillars, clusters, articles — without writing a word of the articles themselves. She provides business objectives, audiences, products/services and priority keywords, then instructs the model to build the hierarchy Strategy > Content Pillars > Topic Clusters > Articles with no duplicated topics. What used to take hours in spreadsheets now takes 20–30 minutes and gives her team a proper structure to map keywords and internal links against.
Reverse-engineer what AI search actually rewards
Oscar Scolding researches the answer engines themselves. He takes a buying question, runs it in 8–10 variations through Perplexity, and pastes every source it cites into a spreadsheet. After about fifteen minutes the patterns are obvious — the same pages keep getting pulled, and their shared format (comparison tables, numbered lists) tells you what the model prioritizes. He also adds “list your sources with URL” and “justify why you picked each source” to every prompt, which makes the model explain its own reasoning. Because these preferences shift weekly, he re-runs the analysis monthly.
Make the model your junior analyst — not your strategist
James Weiss frames the whole approach: don’t ask AI for SEO recommendations, give it the project and let it analyze. He feeds real data — page exports, keywords, search intent, titles, headers, links, conversion targets — and asks the model to identify topics to cover, page duplication, competing pages, and whether a merge or rewrite is needed. Audits that took hours now reach first-draft in half the time, and everyone walks into the meeting with the same information. His mental model: treat it like a junior analyst who knows nothing about the client.
The takeaway
Every workflow here shares one move: give the model your real data and a fixed structure, and ask it to diagnose rather than generate. Intent mapping, entity tables, and taxonomies are all ways of deciding what should exist before spending any time on what to say. Get that right and the writing stage — covered in the content-brief guide — becomes far faster and far less error-prone.
Read the full collection: 25 AI Prompts and Workflows That Speed Up SEO Turnaround Time »
Related deep dives: Client context & AI content briefs · Answer Engine Optimization
Need this built into a repeatable system for your site? Talk to Salam Experts »
