SEO & GEO/AEO

Verified against ChatGPT · 2026-08-10

Write a schema markup brief a developer can implement without guessing which fields are required

Turns a page's content into a structured JSON-LD schema brief with the correct schema type, required versus optional properties clearly separated, and a plain-language note on what could go wrong if a field is filled in inaccurately.

ChatGPT (GPT-5.1)2 fillable variables

The prompt

Ready to copy — highlighted parts are example details you can swap.

You are writing a schema.org markup brief for a developer to implement — the developer knows how to write JSON-LD but doesn't know this specific page's content or which schema type actually fits it, so your job is to make both of those decisions explicit and defensible, not just hand over a code block.

PAGE TYPE AND CONTENT
A recipe page for a sourdough bread recipe, includes prep time, cook time, ingredient list, and 40 user-submitted star ratings averaging 4.6.

EXISTING SCHEMA (IF ANY)
No schema currently implemented on this page.

STEP 1 — SCHEMA TYPE DECISION
Name the specific schema.org type (or types, if nesting applies) that fits this page, and explain in one or two sentences why this type fits better than the next most plausible alternative — do not just declare a type without ruling out the obvious runner-up, since picking between adjacent types (e.g. Product vs. Service, Article vs. BlogPosting, FAQPage vs. QAPage) is often the actual point of confusion.

STEP 2 — PROPERTY BREAKDOWN
List the properties this schema type needs, split into two groups: required for the markup to validate and be eligible for rich results, and optional-but-recommended because they meaningfully improve how the result can display. For each property, state what page content it should be filled from — do not invent placeholder values; if the page content given doesn't actually contain a value for a property, say so and mark it as needing input rather than guessing a number, date, or rating.

STEP 3 — ACCURACY WARNING
For any property where an inaccurate value creates real risk (aggregateRating, price, availability, review count) — not just a validation error but a value that could misrepresent the page to a user or trigger a manual action — add a one-line warning about what happens if that field doesn't match what's actually and visibly true on the rendered page.

OUTPUT FORMAT
1. Schema type decision with reasoning.
2. Property table: Property | Required/Recommended | Source of value | Accuracy warning if applicable.
3. A JSON-LD skeleton with the correct structure and placeholder markers (like <<FROM_CMS_FIELD>>) instead of invented values, ready for a developer to wire up to real data.

Customize

Optional — swap in your own details for the highlighted parts above.

Why this works

The step that requires ruling out the next most plausible schema type exists because the actual difficulty in schema selection is rarely picking a type in isolation — it's distinguishing between two adjacent types that both technically apply, and a model asked simply to "pick the schema type" will name one confidently without surfacing that the choice was close, leaving a developer no way to sanity-check the decision against their own judgment of the page. Splitting properties into required-for-validation versus optional-but-recommended matters because these two categories carry different implementation priority and a flat list forces a developer to separately go look up which is which in Google's documentation — providing the split in the brief itself is what makes this handoff-ready rather than just informative. The instruction to mark a property as needing input rather than guessing a value directly targets a known and costly failure mode: GPT-5.1 will readily invent a plausible aggregateRating, price, or review count if asked to produce complete JSON-LD from thin content, and shipping a fabricated numeric value in structured data is worse than shipping no value at all, since search engines can take manual action against pages whose structured data doesn't match visibly rendered content. The accuracy-warning step is scoped specifically to properties with real consequences rather than applied blanket to every field, because flagging every property equally would train the developer to ignore the warnings — reserving them for rating, price, and availability keeps the signal meaningful exactly where a mismatch is a policy risk rather than a cosmetic one.

What you get back

Schema type: Recipe (not HowTo) — the content is a specific dish with ingredients and yield, not a general multi-step process without a food outcome, which is the distinguishing line between the two types. Property table: name (required, from page title) | prayTime/cookTime (required, from listed prep/cook times) | aggregateRating (recommended, from the 40 ratings averaging 4.6 — WARNING: must match the visibly displayed rating exactly, mismatched aggregateRating is a common trigger for manual structured data actions).

Verified against

ChatGPT GPT-5.1 · 2026-08-10

Changelog

  • 2026-08-10 — Initial publish, verified against ChatGPT GPT-5.1.

Building this for real?

This is a free starting point. If you'd rather have SEO built and running for your business, that's Scult's day job.

EXPLORE SEO
All SEO & GEO/AEO prompts

Check your AI visibility

One URL in, a 0–100 score and the exact fixes out.

RUN THE CHECK

Browse all the tools

15 tools across six categories
13 of them never send your data anywhere

Free · No signup · No trial clock

SEE THE DIRECTORY