The SEO playbook for content that ranks
Choose the page job with a post type, assemble it from rule-bound elements, and run it through an ordered process that decides what ships and proves what worked.
Business context changes priorities. Foundations explain the rules. Results close the loop.
Why briefs fail and systems work
A brief can describe a desired page, but description is interpretable. As volume grows, unwritten rules become inconsistent choices, reviews become subjective, and each new contributor inherits less context than the last.
- ✓One instruction, many interpretations — ‘include a useful comparison’ does not define rows, evidence, decision criteria, exclusions, or what makes the comparison complete.
- ✓Drift compounds — by article 40, teams review against recent examples instead of the original standard, so accidents quietly become precedent.
- ✓AI fills every blank — an agent does not stop at ambiguity; it chooses a plausible structure, tone, claim, or CTA and presents the choice confidently.
- ✓Typed rules make review observable — required fields, ranges, order, source rules, and done-when conditions turn taste-based feedback into a pass, fail, or documented exception.
What goes wrong without a content system
A brief is prose. Prose is a weak storage format for repeatable decisions, because every reader has to translate it back into structure.
“Write the definitive guide, add a comparison, cover objections, make it authoritative.” Every word sounds sensible. None of them specifies the opening answer, the comparison criteria, the source threshold, section order, length, or the release test.
So one writer produces a narrative, another a list, a third copies the top-ranking page. All three can honestly say they followed the brief.
Editors then review against personal taste. “Make it punchier” fixes one draft without defining the rule the next draft should follow. By article 40, nobody can say whether the newest page matches the standard or merely resembles the last one accepted.
AI agents amplify this: fluent output hides that the specification never authorized the choice.
The remedy is not a longer brief. It is a system that separates stable rules, which belong to post types, elements, and process, from the page-specific inputs every assignment still carries.
The three layers: post types, elements, process
Each layer answers a different class of question. Keeping them separate makes the rules reusable; connecting them makes the output coherent.
Process
Decides Work order
- What gets built, in what order, and by whom?
- How does the team prove it worked?
Measure → learn → update the spec
Fixing the post type before drafting is what makes the next layer cheap. It settles the section order and the reader's job once, so elements only has to decide what fills each already-fixed slot — never whether the slot should exist.
The fields and length bands an element specification requires become exactly what process can gate on later: a missing verified price or a table with no source is a stop condition, not a style note, because the field was declared upstream.
Process is where evidence closes the loop. A page that performs attaches a reusable proof point to its post type; a page that fails sends the exception to that same spec — not to one writer's memory.
See where your content already stands
Run a free AI visibility check before deciding what the playbook should fix first.
Six pillars of the playbook
The listing is a route into the method, not the method itself. Start with the decision blocking your team today.
54 specifications. Choose the document shape matching the reader's task, from ultimate guides and comparisons to product pages, use-case pages, and case studies.
For: strategists, editors, and anyone deciding what page should exist.
74 specifications. Assemble pages from typed blocks with rules for purpose, position, fields, evidence, schema, accessibility, and QA.
For: writers, designers, developers, and AI-agent operators.
18 operating models. Prioritize formats and topics for ecommerce, SaaS, marketplaces, local services, publishers, B2B services, healthcare, finance, agencies, and nine more.
For: owners and growth leads translating a generic playbook into commercial priorities.
18 phases plus 13 checklists. Move from goals and data access through technical readiness, research, competitive gaps, information architecture, and pre-publish QA.
For: agencies, content leads, and teams needing ownership and repeatability.
5 guides. Why systems outperform isolated briefs, how consistency scales, how structure signals trust, and how humans and retrieval systems share a page.
For: leaders setting standards and reviewers deciding when an exception is justified.
1 measurement hub. Separate visibility, selection, engagement, and business outcomes, then connect each movement to a time window and next decision.
For: stakeholders asking whether the work changed anything that matters.
A system is what lets the fifth person to touch a page make the same decision as the first.
Six pillars, three layers, one specification. Nothing in the playbook exists merely to make writing faster. It exists to make the decision repeatable.
Where to start
You do not need to read the playbook in order. Choose the route matching your role, then follow the joins to the other layers only when useful.
Planning the portfolio? Start with Post Types and Business Types. Repairing one underperforming page? Identify its post type, audit its elements, then use Results to define recovery. Building an agent workflow? Start with Elements for output constraints and Process for tool access and stop conditions.
What a specification gives you
The goal is not to make every page identical. It is to make intentional differences visible and accidental differences less likely.
Every rule starts with the reason it exists. A direct answer belongs near the opening because a reader, or a retrieval system, needs a self-contained response before the supporting detail.
The reason is what lets an editor judge the unusual case. Without it, a rule becomes ritual.
A typed block names the information it must contain. A comparison needs decision criteria, comparable options, evidence, and a conclusion. “Add a table” names only a visual format.
Fields let a writer gather missing inputs before drafting, and let an agent stop instead of inventing a plausible value.
“Concise” means different things to different people. A word band, a maximum number of takeaways, or a defined table scope gives the reviewer an observable boundary.
When a page has to break a range, the owner records why, and that exception becomes evidence for changing the specification.
A useful acceptance condition can be observed: every material claim maps to a source; every internal route resolves; the direct answer stands alone; visible FAQ answers match the structured data.
Expert review then goes where it belongs: truth, strategy, and nuance.
The system still leaves room for voice, original research, argument, and design. It places those choices inside a declared page job. Consistency then means the same standard of reasoning, not the same sentences on every URL.
Turn your own backlog into a specification
Bring a page that needs an owner. We will show how a post type, its elements, and one process phase turn ambiguity into something a reviewer can pass or fail.
A playbook AI agents can execute
A content system was always useful for people. It became necessary the moment AI agents started writing at volume, because an agent will fill any gap a brief leaves: confidently, plausibly, and invisibly.
Give an agent a topic and a keyword and it has to invent the page: which sections exist, how deep each goes, what evidence is required, where the answer belongs.
A human writer knows what they do not know and asks. An agent does not stop.
The post type names the job and fixes the section order. The element library defines every block it may use, with required fields, length bands, and exact position.
Nothing is left to interpretation, so nothing has to be invented.
Typed fields make missing inputs detectable. If a comparison needs a verified price and no source exists, the agent halts and flags the gap rather than inventing a number nobody checked.
That is the difference between an agent that scales your standards and one that scales your errors.
The same specification governs article 1 and article 400, a glossary term and a category page. Because each type declares its own structure, an agent producing forty page types still produces one standard of reasoning.
Every page in the playbook is precise enough that a copywriter and an AI agent working from the same document produce the same structure, and precise enough to hand straight to SEO agents or to your own tooling through the AmICited MCP server.
The playbook by the numbers
Library counts were verified from the content tree on August 27, 2026. Outcome figures come from published case studies and are not guarantees.
These outcomes combine strategy, execution, market conditions, tools, and time; the specification alone does not cause them. The narrower claim is the useful one: a governed content system makes sustained production possible and gives measurement a stable object to evaluate.
How the playbook works with AmICited
A method without observations becomes opinion. A dashboard without a method becomes motion without a decision. Each is most useful when it keeps its own job.
You can run the method with another toolset, because its core units are plain operational decisions. AmICited shortens the distance between those decisions and the evidence: which prompts are missing your brand, which competitors are cited, and how visibility changes after you publish.
SEO Playbook FAQ
Yes. The playbook is written as an operational specification: post types define the job and required structure, elements define exact block-level rules, and process pages define inputs, owners, order, and acceptance checks. Product behavior and page counts can change, so each guide should still be checked against the current site before production.
Yes. The playbook is public and usable without an AmICited account. You can adopt one post type, element, or checklist at a time, copy the decision logic into your own workflow, and measure results with your existing tools. AmICited adds instrumentation and automation; it is not a condition of using the method.
They can. The specifications give an AI agent explicit fields, order, constraints, evidence requirements, and pass conditions instead of asking it to infer quality from a prose brief. A human still owns strategy, claims, exceptions, and release approval, while deterministic checks should be automated wherever possible.
A content brief describes one assignment in prose. This playbook defines reusable content types, typed elements, joins between them, and a governed production process. A brief can still carry query research and page-specific evidence, but it inherits stable rules instead of rewriting or improvising them for every page.
Turn the next content idea into a specification
Free check · 7-day trial · card required