Should every section open with a direct answer

No. Not every section, and not every paragraph. Open with a direct answer on the H2s that own a buyer question on pages you actually want cited. Leave narrative, legal, and case-study sections to their jobs. Flattening every block into a dictionary lead is how AEO advice turns a site into commodity copy that models have no reason to prefer.

This is the section-level rule under the Answer Engine Optimization playbook. The page-level opener — first breath, then depth — is a different object; do not treat this spoke as a second copy of that job. I have been SEO certified since 2021. The citation version of that work is still a lead you can lift from one heading, not a style you stamp onto every paragraph.

The short answer

• Yes: citation-intended sections (definition, how-to, comparison, pricing, eligibility, objections) open with the take in sentence one.
• No: not every blog paragraph, not manifesto scenes, not legal clauses, not case-study narrative.
• A section lead is 1–3 sentences that still work if a model lifts them with no surrounding page.
• Google does not require this format. It is an editorial default for pages meant to be quoted, not a ranking hack.
• Rewrite money-page H2s first. Leave the brand film a brand film.
• Later paragraphs under an answered H2 prove the take. They do not repeat a second capsule.

Should every section open with a direct answer?

No. The correct default is narrower: every citation-intended section on a page you want quoted should open with the answer to the question that heading owns. Other sections should open with whatever their job requires — a scene, a rule, an outcome, a constraint — not a fake FAQ sentence.

Google’s own generative-AI guidance tells you to organize pages with paragraphs, sections, and headings people can follow, and says you do not need to rewrite copy just for AI systems (Google: optimizing for generative AI features). Treat section-first as that organization with the conclusion moved to the top of the H2s that will be compressed. It is not a secret markup file.

Section job	Open with a direct answer?	Open with instead
Definition / how-to / comparison / pricing / eligibility	Yes	—
Objections / “when not to” / measurement	Yes	—
Brand narrative / manifesto / filmic about	No	Who, for whom, then the scene
Legal / policy / terms	Partial	The rule, then the exception
Case study / field story	Partial	Outcome and constraint, then the scene
Method paragraphs under an already-answered H2	No	The next fact, not a second capsule

If the heading is a question a buyer would type, the first sentence is the answer. If the heading is a chapter in a story, the first sentence is the story. Mixing those jobs is how you get FAQ sludge on a case study and a buried offer on a service page.

Worked leads — example copy for a service URL, not a published Spurlock SLA. Put your real range in the right-hand shape. If you do not have a real range, write the variable that moves it and stop. Fake precision in a section lead is worse than a buried lead. Models will repeat the fake number with your URL attached.

H2	Buried / wrong job	Citation-intended lead (shape)
How long does onboarding take?	“Timing depends on a lot of factors, so let’s start with how we think about discovery.”	“Onboarding takes [N days/weeks] after [named input]. The variable is [locales / SKUs / properties], not the kickoff call.”
What’s included?	“Our process is collaborative and tailored.”	“Included: [three named deliverables]. Not included: [the thing sales keeps implying]. There is no citation guarantee.”
What happened in week three?	“A case study is a structured narrative of a client engagement.”	Outcome first if you have a sourced result; otherwise the scene. Never “a case study is…”

Should every paragraph open with a direct answer?

No. That is how you get FAQ sludge. A paragraph under an already-answered H2 should open with the next fact, the next step, or the next constraint — a topic sentence, not a second capsule.

NN/g’s one-idea-per-paragraph rule is still right: scanners skip extra ideas if the first few words do not catch them (NN/g: How Users Read on the Web). That is not permission to define the same noun fourteen times. Once the H2 is answered, later paragraphs earn depth. Google’s generative-AI guide says there is no ideal page length and no requirement to chunk content into tiny pieces for AI to understand it (Google: optimizing for generative AI features). Paragraph-level capsules are chunking with better grammar.

Block	First-line job	Repeat the H2 answer?
Page opener	Answer the URL’s primary question	Once
Citation-intended H2 lead	Answer that heading	Once
Proof paragraph under that H2	Next fact, step, or constraint	No
Table / procedure	Bounded object the lead promised	No
Closing line of the section	Point, not a restated definition	No

Checklist — paragraph discipline:

• The H2 already has a standalone answer in sentences 1–3
• Paragraph two does not begin “In other words, X is…”
• You can delete paragraph three and the lead still cites
• Adjacent paragraphs do not start with the same two words
• Proof still sounds like the studio, not like a glossary

If a later paragraph answers a new question, it is in the wrong section. Promote it to its own H2 and give that heading a lead. Do not hide a second question in sentence eight of the first one.

How is a section lead different from a page lead or a topic sentence?

Three different objects. Teams collapse them, then either rewrite the whole archive or refuse to touch anything.

The page lead is the first 2–4 sentences of the URL. That job is covered on the answer-first sibling. A section lead is the first 1–3 sentences after a specific H2. A topic sentence is the first line of a paragraph so a scanner can skip or stay. Nielsen Norman Group’s inverted-pyramid advice is recursive — headline, first paragraph, first sentence of each section, first words of each sentence (NN/g: Inverted Pyramid) — but “most important” is not the same as “direct answer to a search query.”

Object	Where it lives	What it must do	Failure if you confuse it
Page lead	First screen of the URL	Answer the primary question of the page	You polish H2s while the hero still withholds the offer
Section lead	Sentence one after an H2	Answer the heading’s question, standalone	You write a topic sentence that cannot be cited
Topic sentence	First line of a body paragraph	Signal the paragraph’s one idea	You stamp “X is Y” on every graf and kill voice

Checklist — name the object before you rewrite:

• This URL has a page-level answer above the first H2
• This H2 owns one question, not three
• The first sentence after the H2 would still make sense in a chat answer
• Later paragraphs under that H2 are proof, not a second capsule
• You are not converting a scene, a statute, or a result into a glossary entry

Topic sentences organize. Answer capsules get quoted. Use the capsule where a model will be asked the heading out loud.

Which sections must open with a direct answer?

The ones a query fan-out can issue as their own search. Google documents that AI Overviews and AI Mode may fan out into related sub-queries and attach supporting links to the assembled response (Google Search Central: AI features and your website). Eligibility to appear as a supporting link is blunt: indexed, and allowed to show in Search with a snippet. There is no extra AI heading format.

That is why section leads matter more than a prettier hero. The model is not obligated to quote your page opener. It can quote the H2 that matches the sub-query.

Heading shape	Fan-out shaped ask	Required lead
What is X / Does X include Y	Definition or scope	Name X and bound it in sentence one
How do I X	Procedure	State the sequence, then number the steps
Should I X or Y	Decision	Pick a default and name the exception
How much / how long	Constraint	Give the range and the variable that moves it
What breaks if I X	Failure	Name the break and the cost
How do I know it worked	Measurement	Name the metric and the window

Decision list:

• If a buyer would ask that heading into ChatGPT Search or Perplexity, write a section lead.
• If the heading exists to move the reader through a story, do not.
• If you cannot say which question the H2 owns, delete or split the heading before you write a lead.

A clever H2 that hides the point until sentence four is still a buried lead. The heading is not the citation. The first honest sentence under it is.

What should a narrative, legal, or case-study section open with instead?

The exception is the point of this post. Forcing a clinical definition onto those sections is how AEO contractors make sites feel dead.

Narrative and manifesto pages still need a who / what / for whom in the first screen. They do not need every H2 to open like a support article. Legal and policy writing already has a better rule than “answer capsule”: state the rule, then the exception. The U.S. National Archives’ clear-writing principles, based on the Federal Plain Language Guidelines, tell drafters to avoid leading with exceptions and to put the rule or category first (National Archives: Principles of Clear Writing). Case studies should put the outcome and the constraint in the first breath of the page, then tell the story in the sections. They should not open a “what we did in week three” heading with a glossary sentence about what a sprint is.

Page or section type	First sentence job	Do not open with
Narrative / about / manifesto	Identity and stake, then the scene	“X is a digital experience studio that…” as every H2
Legal / policy / terms	The rule that applies, then the exception	A marketing definition of the product
Case study	Outcome + constraint (numbers only if sourced)	“A case study is a narrative that…”
Citation-intended service / FAQ / comparison	Direct answer to the heading	Atmosphere, then the answer in paragraph four

Checklist — keep the exception honest:

• Narrative H2s start in scene or conflict, not in a textbook definition
• Legal H2s state the rule before the carve-out
• Case-study H2s start with what changed, not with what a methodology is called
• You did not paste the same “X is Y. Here’s why it matters.” template into all three
• The money URL next door still has real section leads — exceptions are not an excuse to bury the offer

Before / after on the three exceptions (still example copy):

Exception	Bad first sentence	Better first sentence
Narrative	“Spurlock Studios is a visibility and automation studio that helps brands get cited.”	“The homepage had a beautiful reel and no sentence a model could lift. The offer lived in a caption on frame four.”
Legal	“This section explains what a refund is in plain language.”	“Refunds must be requested within 14 days of delivery. Digital work already accepted in writing is not refundable.”
Case study	“Here is why process matters in a redesign.”	“The live site still used the old NAP for six weeks after launch. Models kept citing the zombie Webflow URL until the 301 shipped.”

The legal row follows Archives guidance: rule first, then the carve-out. Do not put “except digital work” before the reader knows the rule exists. The case-study row starts with what broke, not with a methodology name. If you do not have a sourced incident, do not invent one — pick a different section or keep the scene you actually have.

A dictionary lead on a film page is still a buried story.

What belongs in the first two sentences after an H2?

A lift-ready answer plus one constraint. Not a slogan. Not a throat-clear. Not “as we said above.”

NN/g’s F-pattern work is the human half of this: first lines get more gazes than later lines, and the first few words on the left of each line get more fixations than the rest (NN/g: F-Shaped Pattern of Reading). Their antidote list starts with putting the important points in the first two paragraphs, using visible headings, and starting those headings with information-carrying words. A section lead is that antidote applied under the H2, not a new file type.

Ingredient	Include	Cut
Subject	Named in plain language	Orphan “this” / “it”
Answer	The take, not a teaser	“It depends” with no default
Constraint	One tradeoff, range, or exception	A second unrelated claim
Entity	The product, page, or rule the heading is about	Brand mythology
Proof pointer	“Table below” only after the take is already stated	“See above” as the whole lead

Section-lead quote test (different from a page-level 60-word paste):

1. Cover the rest of the section.
2. Read only the H2 and the next 1–3 sentences.
3. Ask whether a model could cite that block as the answer to the heading.
4. If it fails, rewrite the block. Do not add 400 words of context hoping the model “gets it.”

Isolated-paste failure	What you see	Fix
Orphan pronoun	“This is why teams struggle”	Name the subject
Context leak	“As noted above” / “in the next table”	Restate the fact
Slogan	Category claim, no mechanism	Add how, or the constraint
Second question	Two answers in one lead	Split the H2
Dictionary on a story	Definition where a scene belonged	Restore the exception

If you would not stand behind that isolated block next to your URL in a buyer’s chat, it is not a section lead yet.

Why do scanners and models both stop at the first sentence?

Because both are trying to spend as few fixations as possible. People scan. Models compress. Neither owes you paragraph four.

NN/g’s 1997 web-writing study found most test users scanned rather than reading word by word, and that one idea per paragraph matters because people skip extra ideas if the first few words do not catch them (NN/g: How Users Read on the Web). The later F-pattern work adds other scan modes, including a layer-cake pattern where eyes hit headings and skip the prose underneath, and a bypassing pattern where people skip the first words of lines that all start the same way (NN/g: F-Shaped Pattern of Reading).

That last one is the AEO own-goal. If every section starts with “The answer is…” or “X is a…”, scanners skip the template and models get a pile of interchangeable sentences.

Scan / extract mode	What they actually use	What a buried lead gives them
F-pattern	First lines + left-edge words	Atmosphere on the left, answer on the right
Layer-cake	Headings only	A question H2 with no take underneath
Bypassing	They skip repeated openers	Twelve sections that all start “X is”
Query fan-out	The passage that matches the sub-query	A page that is “about” the topic
ChatGPT Search / Perplexity	A sentence that survives paraphrase	A brand paragraph that cannot be attributed

ChatGPT Search is built to return answers with links to web sources (OpenAI Help: ChatGPT Search). Compression prefers a passage that already is the answer. If sentence one is a throat-clear, the system invents your point or cites a competitor who stated one.

Google’s people-first guide is the same constraint in ranking language: unique point of view, not a restatement of what anyone could generate (Google: creating helpful, reliable, people-first content). A section lead that only restates the category — “A visibility audit is a review of your AI presence” — is commodity. A section lead that states the constraint — what you will not score, and that a cite is not a promise — is the thing a model can prefer.

Headings are the layer cake. The section lead is the frosting you actually want eaten.

What fails when every paragraph gets an answer capsule?

Voice dies, and citations do not magically appear.

The failure mode: an AEO contractor rewrites every blog paragraph into the same shape — definition, three bullets, “here’s why it matters.” Conversion drops. The prompt panel still quotes a competitor because eligibility, entity agreement, and corroboration were never fixed. Google’s generative-AI guide calls out commodity pages that restate common knowledge, and it lists “chunking” content into tiny pieces as a tactic you can ignore for Google Search (Google: optimizing for generative AI features; Google: creating helpful, reliable, people-first content). Stamping an answer capsule on every paragraph is chunking as a style guide.

Cost: a quarter of trust with buyers who can smell factory content, plus a site that no longer has a point of view a model can prefer.

Keep	Reserve	Kill
Direct answers on citation-intended H2s	Page lead on URLs you want quoted	Identical first sentences on every paragraph
Voice in proof, process, and narrative	FAQ answers that a human can read	Hidden FAQ JSON-LD the page does not show
One table that only this heading needs	Pillar link when the system is the next step	Boilerplate sections that would fit any post
Legal rule-then-exception	Outcome-first case study	“What is a case study?” as H2 one

If every paragraph is a definition, none of them is. Models already have the commodity version. They need the constraint only you can state.

How do I audit a page for buried section leads?

Do not start in the blog archive. Start on the URLs a model already tries to quote: homepage, one money service or product page, About, a FAQ or proof page. Open the HTML, not the design file.

Procedure:

1. List every H2 on the URL.
2. Write the question that heading actually owns — or delete it if it owns none.
3. Highlight sentence one under each H2.
4. Mark each as answer, topic sentence, scene, rule, or mush.
5. Rewrite only the mush and the topic sentence on citation-intended headings.
6. Leave scene and rule alone unless they are pretending to be definitions.
7. Confirm the answer is in visible text, not a poster frame or a JSON-LD-only blob (Google: AI features and your website requires important content in textual form).

Audit mark	What sentence one does	Action
Answer	Resolves the heading, standalone	Keep; add a table or steps below if missing
Topic sentence	Signals the paragraph, does not answer	Rewrite into an answer or split the H2
Scene	Starts a story	Keep on narrative / case-study sections
Rule	States the obligation or prohibition	Keep on legal / policy; put exceptions second
Mush	Throat-clear, slogan, or “as above”	Rewrite or cut

Pre-publish checklist for the URL, not the whole site:

• Citation-intended H2s pass the isolated H2 + 1–3 sentence test
• Narrative / legal / case-study H2s were not dictionary-rewritten
• No three adjacent sections start with the same two words
• Each citation section has one table, procedure, or checklist
• Facts in the lead match the rest of the page and any schema you ship

Score the URL, not the writer’s feelings:

Score	What you counted	Ship?
0	No H2 owns a buyer question	Rename headings before writing leads
1	Questions exist; sentence one is mush on most of them	Rewrite leads only
2	Citation H2s pass the isolated paste; exceptions still dictionary-sludge	Restore scene / rule / outcome on those H2s
3	Citation H2s pass; exceptions kept their jobs; one structured object each	Index and log
4	Panel quotes the new lead on at least one frozen prompt	Keep. Do not roll the template sitewide

A 1 is the usual money-page state. Teams skip to 4 by publishing and screenshotting a lucky chat. That is not a score.

If the page is not indexed or not snippet-eligible, stop rewriting sentences. Fix eligibility. Then come back to the leads.

How do citation gaps reveal a buried lead?

A citation gap is the distance between the prompts that make you money and the URLs answer products actually quote. The method lives in citation gaps in competitive AI answers: freeze a prompt panel, log who gets named and who gets cited, treat the missing URLs as the brief.

Section-first changes what you look for in that log. If the model cites a competitor for “how long does X take” and you rank for the parent query, you often already have a page. You buried the duration in paragraph six of an H2 called “Our process.” The competitor opened that section with the range.

Log row	What it usually means	Rewrite target
Competitor cited for a sub-question you cover mid-page	Buried section lead	Promote the fact to sentence one under a question H2
You mentioned, competitor cited	Extractable competitor passage; your lead is slogan or mush	See mentions vs citations — two columns, not one score
Nobody cited; three directories named	You do not have a page that owns the question	New URL or a new H2 on an existing money page
You cited on one engine, absent on another	Non-deterministic; not a reason to rewrite the archive	Keep the panel frozen; fix the worst engine last

Do not invent a share-of-voice percentage from one screenshot. One run is an anecdote. The gap list is the rewrite queue.

Operator rule: if the competitor’s cited sentence could have been your H2 lead, that is an editorial ticket, not a PR ticket.

How do I measure whether section-first is working?

You measure whether the new lead is what gets quoted, not whether you shipped a rewrite. Citations are non-deterministic. Google is explicit that meeting requirements does not guarantee indexing or serving (Google: optimizing for generative AI features).

Run a frozen 15–25 prompt panel weekly. For each prompt, log mention, citation, and which passage was used if you can see it. In Search Console, confirm the URLs are indexed and snippet-eligible. If your property has the generative-AI performance reports Google started rolling out on June 3, 2026, use that view for Overview / AI Mode impressions — it was launched to a subset of sites first, so treat a missing report as “not in the cohort yet,” not as a failed rewrite (Google Search Central: generative AI performance reports).

Column	What you write	Pass / fail
Date + engine	Same panel, same wording	Moving the panel hides the result
Mention	Brand named in the answer	Awareness only
Citation	Your URL credited	The actual AEO scoreboard
Passage	Section lead vs mid-page vs wrong URL	Lead rewrite worked only if this flips
Accuracy	Offer / city / constraint correct	A cite of a wrong fact is a defect
GSC	Indexed, snippet-eligible, gen-AI impressions if available	Eligibility floor

Do this instead of a vendor AEO grade:

• Panel frozen (queries do not rotate weekly)
• Competitor set frozen
• You score mention and citation as separate columns
• You note whether the quoted sentence is the new H2 lead
• You re-measure at two weeks and eight weeks, not tomorrow

A mention without a URL is a different ticket than a cited competitor page. Keep them apart or you will “fix writing” when the real hole is corroboration.

What “working” looks like on the log, hedged:

After the rewrite	Treat as	Do not treat as
Same prompt now quotes your new H2 lead on one engine	Directional win for that engine	A sitewide AEO program that is “done”
Mention up, citation flat	Extractability still losing to a competitor passage	Proof that section-first failed
Citation up, facts wrong	Defect. You taught a bad number	A win because the URL appeared
GSC gen-AI impressions appear	Eligibility + selection somewhere in Search AI features	Proof ChatGPT or Perplexity moved
No change at two weeks	Too early, or you rewrote the wrong H2	A reason to restyle legal and case studies

As of June 2026, Google’s dedicated generative-AI reports were rolling out to a subset of sites (Google Search Central: generative AI performance reports). If you do not see the report, you are not automatically failing. You still have URL Inspection, ordinary Search performance, and the prompt panel.

What should I skip if I only have a week?

Skip the blog calendar. Skip rewriting manifesto copy. Skip legal. Skip case studies. Skip sitewide “AEO voice” templates. Spend the week on section leads for the money URLs.

Google’s mythbusting list is the skip list in official language: you do not need llms.txt, special AI markup, or chunked AI versions of pages for Google Search (Google: optimizing for generative AI features). A week spent on those files is a week you did not promote “how long / how much / what’s included” to sentence one.

Day	Do	Do not
1	Pick five citation-intended URLs. List their H2s.	Open the 2019 blog archive
2	Rewrite mush leads on the money service or product URL	Dictionary-rewrite the case study
3	Homepage + FAQ/proof H2s	New llms.txt, AI-only schema, chunked “AI pages”
4	About facts that must match the money page	Legal/policy restyle as FAQ capsules
5	Isolated quote test + index check	Change the prompt panel mid-week
6–7	Freeze the panel. Log mention vs citation once. Stop.	Ship a 40-post “section-first” rewrite

If you only have a week, you are not doing literature. You are promoting the facts a fan-out will ask onto the first line of the H2 that owns them.

Week-one kill list:

• No new blog posts “for AEO coverage”
• No manifesto H2s converted to definitions
• No terms-of-service FAQPage
• No case-study glossary intros
• No rotating the prompt panel to chase a prettier screenshot

A week that only ships four honest H2 leads on one money URL beat a week that restyled forty posts.

When is this not worth doing yet?

When the page cannot be a source, or when the URL’s job is not to be cited.

Blocker	Why section leads will not save you	What to do first
Sitewide noindex or login wall	Not snippet-eligible	Remove the accident; read AI features
Answer lives in a canvas, image, or data-nosnippet wrapper	Not textual	Put the take in HTML
Legal name in flux	You will teach two entities	Freeze the name, then write
No offer you will take	Money-page H2s are fiction	Pause the rewrite
You are about to restyle terms of service as a listicle	Wrong job	Keep rule-then-exception
Duplicate domains (old Webflow + new Next)	Models cite the zombie	301 the zombie, then rewrite leads on the survivor

Checklist — proceed only if:

• The URL is on the canonical domain and can show a snippet
• You can name the question each target H2 owns
• You can edit HTML, not only a tag-manager banner
• You will not apply the rule to narrative, legal, or case-study sections this sprint
• One person owns the sentence that must stay true across homepage, service, and About

If those boxes are empty, a visibility audit is the right next step, not a style guide. The lane page is /visibility. The booking path is a visibility audit.

Section-first is a constraint for pages meant to be cited. It is not a religion for every URL you have ever published.

What is the rewrite order on a money URL?

One URL. Then stop. Do not “roll out the voice” across the marketing site.

1. Write or repair the page lead so the primary question is answered before the first H2.
2. Rename H2s toward the questions people actually ask. Clever labels lose layer-cake scanners.
3. Rewrite sentence one under each citation-intended H2 until the isolated paste works.
4. Add one structured object per those H2s — table, numbered steps, decision list, or checklist.
5. Leave narrative, legal, and case-study blocks on this URL alone unless they are currently fake definitions.
6. Diff-check facts against About, footer, and schema so the new lead does not invent a second offer.
7. Request indexing. Log the URL on the frozen panel. Do not declare victory from the pull request.

Step	Done looks like	Not done
Page lead	First 2–4 sentences answer the URL’s question	Hero still teases
H2 names	A buyer could type them	“Our approach” / “The journey”
Section leads	Isolated H2 + 1–3 sentences cite cleanly	Topic sentences and mush
Structure	One bounded object per citation H2	Walls of undifferentiated prose
Exceptions	Story, statute, outcome still sound like themselves	Dictionary sludge
Measurement	Panel row exists for this URL	“We published, so it worked”

Worked order on one money URL (service page, not a blog post):

1. Page lead: who you are for, what they buy, the constraint, the next step — in HTML, above the first H2.
2. Rename “Our approach” to the questions already in your sales calls: what’s included, how long, what you will not do, how you measure.
3. Write four section leads. Each must survive the isolated paste.
4. Under each, add the object: inclusion table, timeline table, out-of-scope list, measurement checklist.
5. Skip the “culture” H2 and the “client story” H2 on this pass. Outcome at the top of the story URL is a different ticket.
6. Diff the four leads against About and the footer. If city or legal name disagrees, you just taught two entities.
7. Log the four prompts that match those H2s. That is the only scoreboard for this URL.

Prompt that should now hit this URL	If the model still cites a competitor	Likely leftover
What’s included in [offer]?	Their inclusion sentence is cleaner	Your lead is still “collaborative”
How long does [offer] take?	They stated a range	Yours is still “it depends” with no variable
Who is [brand] in [city]?	About / GBP / homepage disagree	Entity drift, not heading craft
Should I hire [category] or a generalist?	They picked a default	You wrote a history of the category

Hundreds of production sites later, the pattern that holds: the money page can afford a crisp H2 lead. The art page cannot afford a committee rewrite. Put the answer where you want to be quoted. Leave the rest of the site its job.

FAQ

Should every section open with a direct answer?

No. Open with a direct answer on citation-intended sections — definition, how-to, comparison, pricing, eligibility, objections — on pages you want quoted. Do not stamp that shape on every blog paragraph. Narrative, legal, and case-study sections keep scene, rule-then-exception, and outcome-then-story.

How do I measure whether should every section open with a direct answer is working?

Run a frozen prompt panel and score mention, citation, and whether the quoted passage is the new H2 lead. In Search Console, confirm the URLs are indexed and snippet-eligible, and use generative-AI impression reports if your property has them. A rewrite with no panel is a hope, not a measurement.

What usually fails first when teams try this?

They rewrite every paragraph into a definition, including case studies and legal pages, or they polish blog H2s while the money URL still opens with atmosphere. The next failure is identical openers that scanners skip. Fix citation-intended leads on five money URLs before you touch the archive.

How long does this take to show results?

Eligibility can be checked in days with URL Inspection once the HTML is live. Accurate answers on a frozen prompt panel often take weeks, and citations are not guaranteed — Google’s docs say indexing and serving are never a promise. Re-measure at two weeks and eight weeks on the same queries; do not invent a day-count SLA.

What should I skip if I only have a week?

Skip the blog calendar, manifesto restyles, legal, case studies, llms.txt theater, and AI-only schema. Spend the week promoting how-long, how-much, and what’s-included facts to sentence one on the money URLs. Log the panel once. Then stop.

When is this not worth doing yet?

When the site is not snippet-eligible, the answer is not in HTML text, the legal name is mid-rebrand, or you do not have an offer you will take. Fix eligibility and the offer first. Section-first copy cannot cite a product that does not exist, and it should not be applied to pages whose job is not to be quoted.

CTA

If models quote a competitor’s H2 while your answer is still in paragraph six, promote the take — then book the audit if the prompt log still lies.

Lane: /visibility · Book a visibility audit.