Contents
- Fast answer
- Quick facts
- What this covers
- Why it matters
- Where to go next
- Outcomes
- Step by step
- The methodology: what the planner produces from your four inputs
- The depth standard: what a page that competes actually contains
- Four traps that cost real work, and how to avoid them
- Planning pages for a search world that answers before it clicks
- What still ranks, and on what timeline
- Planning page portfolios without producing thin content
- What changed in 2025-2026 for SEO planning
- The limits: a brief is not a keyword strategy
- When not to use this planner, and what to do instead
- At a glance
- Mistakes to avoid
- FAQ
Fast answer
An AI agent SEO page planner should force a page to be useful before it exists, requiring clear search intent, an original workflow angle, internal links, a free asset, a next-step tool, and a course path that fits the reader. Google guidance is blunt: SEO works best when it helps search engines understand people-first content, so every page should solve a real workflow problem and then route readers to a calculator, template, or course. Choose the intent first, labeling the page as learn, compare, build, use, or buy rather than mixing jobs into one. Write the angle by explaining what the visitor should know, decide, calculate, or build afterward. Attach internal links to one tool, one template, one use case, and one course when relevant. Add free value such as a checklist, worksheet, calculator, or worked example before any course pitch. Only route to a course once the visitor understands why the workflow is worth learning, and reject pages that cannot provide useful differentiation.
Facts last verified 2026-07
Quick facts
- Click-through with AI summary
- 8% of visits vs 15% without, Pew Research, 2025
- CTR impact on top result
- 58% lower average CTR when an AI Overview is present, Ahrefs
- New pages reaching top 10
- 1.74% within a year; average #1 page is ~5 years old, Ahrefs
- Zero-click searches
- 58.5% of US Google searches ended without a click in 2024, Semrush
- LLM referral growth
- 9.9x since Nov 2024; ChatGPT holds 92.4% of it, Previsible, July 2026
- Google's AI content stance
- Rewards quality however produced; automation to manipulate rankings is spam
On this page
What this page covers
A tool visitor should leave with a decision, not just a number: build now, prepare first, choose another workflow, or follow a course path.
- 01Keyword
- 02Intent
- 03Page angle
- 04Internal links
- 05Free asset
- 06Course CTA
Why does this matter now?
Google's own guidance is blunt: SEO works best when it helps search engines understand people-first content. For A8gent, that means every page should solve a real AI workflow problem and then route readers to a calculator, template, tutorial, or course.
Internal path
Where to go next from this page
These links are part of the A8gent learning and conversion path. Use them to move from concept, to diagnosis, to workflow build, to course.
What you should be able to do after this
- Choose page intent
- Build internal links
- Plan a free asset
- Route readers to the right course
How do you do it, step by step?
1. Choose the intent
Label the page as learn, compare, build, use, or buy. Do not mix five jobs into one page.
2. Write the useful angle
Explain what the visitor should know, decide, calculate, download, or build after reading the page.
3. Attach internal links
Connect the page to one tool, one template, one use case, and one course when relevant.
4. Add free value
Give away a checklist, worksheet, calculator, template, or worked example before any course pitch.
5. Define the course route
Only route to a course when the visitor has enough context to understand why the workflow is worth learning.
The methodology: what the planner produces from your four inputs
The planner takes a target keyword, a search intent, and a choice of free asset, and returns a page brief. The keyword is slugified by lowercasing it, replacing every run of non-alphanumeric characters with a hyphen and trimming the ends. The intent chooses the folder: Build routes to the agents section, Compare to the comparison section, Use to the tools section, and Learn to the learning section. Those two combine into the proposed URL.
The title is your keyword with the first letter of every word capitalised, which is a starting point rather than a finished title, and the angle is one of four stored sentences selected by intent: help the reader choose, show the workflow and review rules, give something applicable now, or teach the concept with examples and risks. The suggested internal links are a fixed set of three, identical for every brief. The free asset you choose is echoed back as the next step and does not affect anything else.
So the planner is a discipline device, not a research tool. It has no keyword volume, no difficulty score, no competitor analysis and no view of what you have already published. What it does is refuse to let you start writing before you have named an intent, an angle, a link plan and a free asset - which is the checklist that actually decides whether a page earns anything, and the one people skip.
The angle field is the important one and the one to take literally. If you cannot say in a single sentence what this page adds beyond the current results, the page is not ready to be written. That test alone prevents most of what search guidance treats as unhelpful content and most of what answer engines decline to cite, and it costs thirty seconds.
The depth standard: what a page that competes actually contains
A brief is not a specification until it has a depth target attached. The standard we hold entity and tool pages to, derived from measuring what the pages currently winning these queries contain, is 2,000 or more words of substantive section prose, a fixed set of seven sections in a fixed order, at least two tables per page, 14 or more distinct internal links rendered, and 8 to 9 FAQs of 60 to 110 words each. Those numbers are not aspirational: they are roughly what the incumbent results carry, and a 900-word page does not compete with them regardless of how well written it is.
Fixed section order matters more than it sounds. Using the same seven sections in the same order across an entire page type makes the set comparable to a reader, predictable to a crawler and, in practice, far easier to keep honest, because a missing section is visible rather than merely absent. For an entity page the order that works is use cases, real cost, the pricing gotcha, production considerations, integrations, alternatives, and migration. Two of those carry most of the weight.
The use cases section has to name concrete workflows with the trigger, the systems touched and the realistic outcome, and it has to include at least one thing the subject is genuinely bad at. Automate customer service is not a use case. The alternatives section targets the alternatives query, which is top-three demand for almost every tool or entity, and it needs a table naming four or five real siblings and a plain recommendation of a competitor where the competitor is the better fit. A directory that never recommends anyone else is not trusted and does not earn links, and that is a ranking observation rather than a moral one.
Each section should also carry one liftable sentence that stands alone if quoted verbatim, because that is the surface an answer engine extracts. Write it as if it will appear with no surrounding context, since increasingly it does.
Four traps that cost real work, and how to avoid them
Measure the rendered title, not the string you typed. Most sites append a brand suffix to every title through a template, so a title that reads as 52 characters in your content file can render at 70 in the results and get truncated. Budget for the suffix explicitly - if the suffix costs nine characters, your raw title has to come in around 51 - and decode HTML entities before counting, because an ampersand written as an entity is one character on screen and five in the string. Counting entities as five characters once made dozens of over-long titles look compliant.
Do not publish a page just because a keyword exists. Programmatic page portfolios fail at the margins, one skipped field at a time, and the field that gets skipped is always the angle. If two planned pages would answer the same query, they are one page, and splitting them makes both weaker. Check your existing titles before adding a brief to the queue.
Verify every route before you link to it. Invented internal links are the most embarrassing defect in a programmatic build because they are invisible in a draft and obvious to a crawler, and we have had to repair links to routes that never existed. Check that the target slug is real, and give every page at least three contextual internal links plus one funnel link so nothing is orphaned.
And set a timeline that matches reality. Published research on ranking age consistently finds that only a small percentage of new pages reach the top ten within a year and that most top-ten pages are several years old. A plan that promises traffic next quarter from a competitive keyword is a plan for disappointment. The workable model is a portfolio where this quarter's pages compound into next year, with the free asset on each page earning links and engagement in the meantime.
Planning pages for a search world that answers before it clicks
The planner asks for intent before angle because the click math changed. Pew Research Center's 2025 study of real user behavior found that when a Google AI summary appears, users click a traditional result on only 8 percent of visits versus 15 percent without one, and links inside the AI summaries themselves were clicked in just 1 percent of visits. Ahrefs' follow-up research found the presence of an AI Overview correlates with a 58 percent lower average click-through rate for the top-ranking page. Pages planned on 2023 assumptions about position-one traffic are planned for a market that no longer exists.
The volatility cuts both ways, which is why intent matters more than volume. Semrush's AI Overviews study tracked AIO presence swinging from 6.49 percent of studied keywords in January 2025 to a peak near 24.6 percent in July before settling around 15.69 percent by November 2025 - and, counterintuitively, found zero-click rates on keywords slightly fell after an AI Overview appeared. Informational head terms lost the most clicks; specific, tool-shaped, and transactional intents held value. The planner's intent field is effectively asking: will the searcher's need survive an AI summary, or is the summary the answer?
Meanwhile a new referral channel grew underneath the old one. The Previsible AI traffic report from July 2026 measured LLM referral sessions growing 9.9x since November 2024, with ChatGPT holding 92.4 percent of trackable LLM referrals and Claude growing 64x over the period. The absolute numbers remain small next to Google, but they compound, and the pages that earn those citations are exactly the kind this planner is built to produce: specific, factual, and structured so an answer engine can quote them.
What still ranks, and on what timeline
Patience remains the entry fee. Ahrefs' ranking-age research found only 1.74 percent of newly published pages reach Google's top 10 within a year - down from 5.7 percent in its 2017 study - and the average page holding a number-one position is around five years old, with 72.9 percent of top-10 pages older than three years. A page plan that promises traffic next quarter from a competitive keyword is a plan for disappointment; the realistic model is a portfolio where this quarter's pages feed next year's compounding.
Google's official position on AI-assisted content is more permissive than the folklore: its Search Central guidance states it rewards high-quality content however it is produced, and that appropriate use of AI or automation is not against guidelines - while using automation to manipulate rankings is spam. The operational reading is that the bar is originality and usefulness, not authorship method. Pages that merely restate what the top ten already say fail that bar however they were written; pages with your own data, worked examples, and verified sources pass it.
That bar is also what earns AI citations. Answer engines quote pages that state facts cleanly, attribute them, and structure them for extraction - quick-fact boxes, tables, dated claims, named sources. The planner's free-asset and internal-link fields push in the same direction: a page with a genuinely useful tool and a coherent cluster around it earns links, engagement, and citations that a thin keyword page cannot, and those are the signals that survived every algorithm and interface shift of the last two years.
Planning page portfolios without producing thin content
The planner routes every page through the same checklist - keyword, intent, angle, internal links, free asset, course CTA - because portfolio SEO fails at the margins, one skipped field at a time. The angle field is the differentiation check: if you cannot state in one sentence what this page adds beyond the current results, the page is not ready to build. That single discipline prevents most of what Google's helpful content guidance penalizes and most of what AI engines decline to cite.
Internal links are the portfolio's compounding mechanism. A cluster where the tool page, the how-to guide, the comparison, and the course landing page all reference each other concentrates authority and mirrors how users actually move - from question to tool to deeper commitment. Orphan pages, by contrast, depend entirely on search for discovery, which the click-through data above says is a shrinking bet to make with every page.
Cadence beats bursts. Orbit Media's 2025 survey of 808 bloggers found those publishing multiple times per week report strong results at 37 percent versus a 21 percent baseline, while roughly half of marketers publish only a few times a month - and the Content Marketing Institute's research consistently shows documented strategy correlates with success, with 62 percent of top B2B performers having one against 37 percent overall. A planner used weekly is a documented strategy; a planner used once is a bookmark.
What changed in 2025-2026 for SEO planning
The measurable events: AI Overviews expanded and stabilized as a permanent SERP feature, Pew documented the click suppression, Ahrefs quantified it, and Semrush documented that 58.5 percent of US Google searches already ended without a click in 2024 before the AI layer thickened. Google simultaneously published guidance for its AI features in Search, and the industry coined generative engine optimization for the practice of earning citations in AI answers. None of this ended SEO; it repriced which pages are worth building.
The winners repriced accordingly. Tool pages, calculators, original data, and genuinely differentiated comparisons gained relative value because they attract clicks an AI summary cannot satisfy, earn links that summaries cite, and hold engagement that thin informational pages lost. That is the strategic thesis behind pairing every planned page with a free asset, and it is now measurable in the LLM referral data: answer engines disproportionately cite pages with concrete, quotable substance.
Measurement had to expand too. A 2026 page plan tracks classic rankings and organic sessions, plus AI Overview presence for target keywords and referral traffic from ChatGPT, Claude, Perplexity, and Gemini - all trackable in standard analytics as referral sources. Expect the LLM slice to be small and growing; treat its growth rate, not its size, as the signal, and re-run this planner's assumptions quarterly because the interface layer of search is still moving.
The limits: a brief is not a keyword strategy
The planner has no data. It does not know search volume, difficulty, current ranking pages, seasonality or whether anyone searches your keyword at all. It will produce a confident-looking brief for a phrase with zero demand and for a phrase you cannot win in three years, and it cannot tell them apart. Bring the demand question from elsewhere and use the planner for what happens after you have answered it.
It does not know what you have already published. Duplicate target queries across two pages are the most common self-inflicted wound in a content portfolio, and the tool cannot see the collision. Keep your own list of published titles and target queries and check every brief against it before writing.
The three suggested internal links are fixed and generic. Real link planning means naming the cluster this page belongs to: the sibling pages it should reference, the hub above it, and the deeper commitment below it. That work is specific to your site and takes about ten minutes per page, and it is the mechanism by which a portfolio compounds rather than sitting as a pile of orphans.
And the folder mapping is one opinion about site structure. Build to an agents section, Compare to a comparison section, Use to a tools section and Learn to a learning section is a sensible convention and not a law. If your information architecture differs, keep yours and use the intent answer for what it is really telling you, which is what the page has to do for the reader rather than where it should live.
When not to use this planner, and what to do instead
Do not use it as your keyword research step. It has no data and takes your word for the keyword. Do the demand and difficulty work first, decide the page is worth building, and then use the planner to force intent, angle, links and asset onto it.
Do not use it for a page whose job is conversion rather than discovery. A service page, a pricing page and a contact page are not planned this way, and applying an SEO brief structure to them tends to bury the offer under context nobody asked for.
Do not use it to plan a hundred pages at once. Programmatic scale is where thin content is manufactured, and the constraint is not the planner but whether each page has something specific to say. If you cannot write the angle sentence for page 40 of 100, pages 40 to 100 should not exist. Plan a cluster of six to ten, publish them properly, measure, then extend.
And if the real bottleneck is production rather than planning, that is an automation problem. Research, first drafts, repurposing and internal link suggestions all sit comfortably in a workflow with a human owning the claims and the final voice, which is a build we would scope in a free automation audit. The SEO checklist template covers the per-page mechanics if you would rather work through it by hand.
At a glance
AI search impact on organic clicks (published studies, 2025-2026)
| Finding | Value | Study |
|---|---|---|
| Click rate when AI summary present | 8% vs 15% without | Pew Research Center, 2025 (68,879 searches) |
| Clicks on links inside AI summaries | 1% of visits | Pew Research Center, 2025 |
| CTR drop for #1 result under AIO | -58% average | Ahrefs, 300,000 keywords |
| AIO keyword prevalence | 6.49% (Jan 2025) to 15.69% (Nov 2025), 24.6% peak | Semrush AI Overviews study |
| Sessions ending on AIO result pages | 26% vs 16% without | Pew Research Center, 2025 |
| LLM referral sessions | 9.9x growth, Nov 2024 to May 2026 | Previsible AI Traffic Report, July 2026 |
What mistakes should you avoid?
- Making pages only because a keyword exists.
- Publishing thin pages that repeat generic AI definitions.
- Forgetting a free next step.
- Linking every page to the same course instead of matching intent.
FAQ
- What does the AI agent SEO page planner actually generate?
- A page brief from four inputs. Your keyword is slugified and combined with a folder chosen by intent - Build to agents, Compare to comparisons, Use to tools, Learn to learning - to propose a URL. The title is your keyword in title case, the angle is one of four stored sentences selected by intent, the internal links are a fixed set of three, and the free asset is echoed back as the next step. There is no keyword data behind it. Its value is refusing to let you write before intent, angle, links and asset exist.
- How long should an SEO page be?
- Long enough to be the most complete answer available, which for competitive entity and tool queries currently means 2,000 or more words of substantive section prose, at least two tables, 14 or more distinct internal links, and 8 to 9 FAQs of roughly 60 to 110 words. Those figures come from measuring what the pages currently winning those queries contain, not from a preference for length. Padding to hit a count is worse than a short page. If you cannot fill the depth with substance, the honest conclusion is that you do not yet have enough to say.
- Why does a fixed section order help?
- Because it makes a page type comparable, predictable and honest. When every page in a set carries the same seven sections in the same order, a reader can navigate any of them after learning one, and a missing section is visible rather than quietly absent. It also forces the uncomfortable sections to exist: the real cost, the pricing gotcha, the alternatives. Those are the sections a writer skips when the structure is free-form, and they are the ones readers came for.
- Should a page recommend a competitor?
- Yes, where the competitor is genuinely the better fit for a named situation. The alternatives query is top-three demand for almost every tool or entity, so the section has to exist anyway, and a comparison that never concedes anything is not read as a comparison. Naming four or five real alternatives with a table and a plain recommendation earns links and citations that a self-serving page does not. This is a ranking observation as much as an ethical one: pages that route readers honestly get referenced by other people.
- What is the most common technical mistake in page planning?
- Measuring the wrong title. Most sites append a brand suffix to every title through a template, so a title that looks correctly sized in your content file renders longer and gets truncated in results. Budget for the suffix explicitly, and decode HTML entities before counting characters, because an ampersand written as an entity occupies one character on screen and five in the raw string. Miscounting entities has made whole batches of over-long titles look compliant when they were not.
- How long before a new page ranks?
- Longer than most plans assume. Published research on ranking age consistently finds that only a small percentage of newly published pages reach the top ten within a year, and that pages holding top positions are typically several years old. Plan a portfolio rather than a campaign: this quarter's pages compound into next year, and the free asset on each page is what earns links, engagement and citations in the meantime. If a page has to produce traffic next quarter, choose a much longer-tail query than the one you were considering.
- Can I use this planner for programmatic page generation at scale?
- For a cluster, yes. For a hundred pages, only if each one has something specific to say. Programmatic portfolios fail at the margins, one skipped field at a time, and the field that gets skipped is always the angle: what this page adds beyond the current results. If you cannot write that sentence for page forty, pages forty onward should not be published. Plan six to ten, build them to the depth standard, measure what happens, then extend. Check every brief against your existing target queries so two pages never chase one query.
Sources & further reading
Was this page helpful?
