Your Landing Page Scored 94/100 for AI Overviews. The Price Is Still Invisible.

Your AI-Readiness Report: 94/100 (up from 37)

Subject: Your landing page is now optimized for AI Overviews. No action required.

Dear valued customer, our software processed your landing page on Tuesday night and applied 147 AI-readiness improvements in 42 seconds. Your previous score of 37 is now 94. The page copy was not checked against your offer. Nobody looked for your price in the HTML a bot receives on its first request. Nothing was checked in Google or in an actual AI answer.

None of that was required for the score to increase. The change log follows.

Change 1: We added 14 FAQs below the footer

Before optimization, your page answered one question badly. After optimization, it answers 15 questions you were never asked, in accordion format, below the footer. Examples include: What is a landing page? What are AI Overviews? Why are landing pages important in today’s digital world? Each answer runs 38 to 44 words, starts with “Great question,” and was inserted without reading your product, your reviews or your sales calls.

Your original headline said, “We fix slow billing for dental clinics.” Our system flagged this as narrow. It then added 14 FAQ pairs marked up with FAQ schema, broadening the page’s apparent subject from a specific service to the internet in general. Your price, service area and first step remain absent. Those fields were left blank to preserve brand voice.

The new questions are available to be quoted somewhere, by someone, in theory. They do not tell a dental clinic whether you serve it or what it would buy. That distinction falls outside the FAQ module’s remit. The module counts questions; it does not ask whether any of them belong to a buyer.

Status: FAQ volume increased. Buyer questions unchanged.

Change 2: “Comprehensive” now appears 31 times

The system detected thin topical coverage. Your page used the word “comprehensive” zero times. It now uses it 31 times in 612 words: once per sentence in paragraph two, three times in a single bullet, and repeatedly in the footer phrase “comprehensive comprehensive solutions for comprehensive needs.” That phrase scored highly on our internal comprehensiveness check.

Before: “Flat-fee billing cleanup for dental clinics. $450 per location. First cleanup in 14 days.”

After: “Our comprehensive platform delivers a comprehensive solution with comprehensive insights for businesses seeking a comprehensive approach to comprehensive growth.”

The price, timeline and customer are gone. Readability declined 22%. Semantic density increased 4,100%. This trade-off was accepted automatically.

The replacement also gives a reader less to repeat. “Flat-fee billing cleanup” identifies the work. “Dental clinics” identifies the buyer. “$450 per location” and “14 days” give that buyer a basis for a decision. The new sentence offers several ways to describe itself and no way to describe the offer. We classify this as expanded topical coverage.

Status: Keyword target met. Original answer removed.

Change 3: A Key Takeaways box summarises itself

To improve citability, we inserted a shaded box titled Key Takeaways at the top of the page. It contains four bullets:

• The page contains comprehensive solutions.
• The FAQs below answer common questions.
• Read the Key Takeaways for a summary.
• The preceding three bullets are the key takeaways.

The box added no price, steps or definition tied to your service. It did, however, give our report a new content-structure item to mark complete. I used to tell clients that a box like this helped Google understand a page. I was wrong about what it reliably helped. It helps a vendor invoice for structure.

A controlled test of content structure found that rewriting structure without changing meaning lifted citation rate by 17.3% across 377 queries on six engines, with document architecture doing most of the work. Our box did none of that work. It reorganised a missing answer into a more prominent missing answer. It is architecture in the sense that a hat is architecture.

Status: Summary visible. Substance pending.

Change 4: Schema now describes your office plant

We injected 412 lines of JSON-LD covering Organization, Product, FAQPage, HowTo, Article, BreadcrumbList, LocalBusiness, Review snippet and one potted fern. The fern sits in reception and is now typed as an Employee with the job title Greenery Specialist. Our validator reports zero errors, the metric this change was designed to improve.

The markup does not supply a price that the page fails to show. Nor does it turn the new FAQs into answers about your offer. According to Google’s guidance on AI Overviews, there are no additional requirements or special optimizations for appearing in AI Overviews or AI Mode; a page indexed and eligible for a normal Search snippet can be eligible for an Overview. Your page was already indexed. The schema did not make it newly eligible.

It made the report longer. Longer reports renew.

Status: Validator passed. Fern employed.

Items flagged as “outside optimization scope”

Our scanner found three issues the score did not reward us for fixing. They were classified as infrastructure or brand voice, which are different departments and therefore different invoices. They are reproduced below with their resolutions. Please note that an outside-scope label is not the same thing as an answer.

robots.txt disallows GPTBot (status: noted)

Your robots.txt contains:

User-agent: GPTBot
Disallow: /

Your server blocks OpenAI’s training crawler across the site. Our tool logged this as informational and moved on. For Google AI Overviews, that part is not the blocker: Overviews are a Search feature, and a GPTBot disallow does not keep Googlebot out. Likewise, blocking Google-Extended does not prevent inclusion in Search. Treating every crawler rule as an AI Overviews emergency would be as useful as treating every crawler as the same bot.

The distinction still belongs in an audit. Training crawlers such as GPTBot, CCBot and ClaudeBot serve a different purpose from live-answer fetchers such as OAI-SearchBot and PerplexityBot. If you do not want training crawlers using your text, you may choose to block them. If you want a live-answer bot to retrieve a page, check whether that bot can fetch it. A training-bot disallow should not be sold to you as the reason Google has not cited your landing page.

GPTBot appears in 5.52% of Disallow rules, blocked about 2.33 times as often as it is allowed. That statistic made an attractive slide. It did not change which bot needs which page.

Resolution: No change to crawler rules; distinction recorded in report.

Pricing loads after the HTML arrives (status: noted)

View-source on your page shows an empty <div id="pricing-root">. The price, plan names and $450-per-location offer load through JavaScript after render. Our optimizer awarded a green check because the text appears in the rendered DOM after 3.2 seconds. A first fetch of the raw HTML still contains none of it.

That difference matters. Google can render JavaScript, but the HTML delivered on the first request does not give every fetcher the same answer. In a December 2024 traffic sample, GPTBot fetched JavaScript in 11.5% of requests and ClaudeBot in 23.84%, yet neither ran the code. Pages that need JavaScript can take nine times longer to crawl than static HTML. A bot reading the raw response sees an empty pricing container where a buyer needs a number.

Our score measures the page after our preferred rendering step. It does not measure whether the price is available before that step. The difference has been filed as an implementation detail, since “the answer exists after 3.2 seconds” fits more neatly on a dashboard than “the answer is absent from the first response.”

Resolution: Green check retained. Price still absent from raw HTML.

The page does not answer a buyer’s question (status: brand voice, untouched)

Your headline now reads: “Unlock comprehensive growth with our comprehensive AI-powered solution.” Say you run dental clinics in Ohio and spend $20k a month. The page does not tell you whether the service is for you, what you would pay or what happens in the first 14 days. Those sentences existed in the earlier copy. They were deleted as redundant during the semantic density boost.

The FAQs cannot repair that omission because they answer general questions about landing pages. The Key Takeaways box cannot repair it because it summarises those questions. The schema cannot repair it because markup describes what is on a page; it does not restore the offer we removed. Each completed change now points to another completed change, and none points to the information the buyer came for.

What the deck calls synergy, I call deleting the answer and selling the summary of the deletion.

Resolution: Brand voice preserved. Buyer answer not restored.

A message from your Customer Success Partner

Hi, this is Jess from Customer Success. Just wanted to reach out personally. This email was triggered automatically when your score crossed 90.

Congratulations on the 94. I reviewed your account and see strong momentum on FAQ volume, semantic coverage and takeaway clarity. Your competitors averaged only 61 on our internal scale last quarter, which we also invented. I have attached an AI-Visible badge for your footer. It links to us, not to your citation, because there is no citation to link to.

Your plan includes monthly re-optimization. We will re-add “comprehensive” if you delete it and re-validate the fern. If you are asking about tools intended to get pages featured in AI Overviews rather than just monitor them, that is outside my scope. I can offer our Visibility Dashboard add-on for $199 per month. It checks whether your score stayed at 94. It does not check Google.

No one has churned from the dashboard, because no one has learned anything from it that would make them churn. Reply YES to renew.

Remediation request: four changes the score cannot award

For customers who insist on checking the page rather than its medal, our support team accepts the following requests. They will not increase your AI-readiness score. They may make the offer easier for a bot, and a person, to retrieve and understand.

1. Check access for the bot that needs the answer. Keep the distinction between training crawlers and live-answer fetchers. A GPTBot block does not explain Google AI Overviews. To inspect what a live-answer bot receives, test with curl -A OAI-SearchBot and read the raw HTML rather than only the rendered DOM. The technical SEO fixes guide walks through that split. If the answer is missing from the first response, a higher score will not put it there.

2. Put the money facts in the HTML. Price, service area, timeline and the one-sentence description of who the service is for belong in text available on the first request. Your rendered pricing module may look correct in a browser while the raw response still shows an empty container. Ask to see the response, not a screenshot taken after the page finishes loading.

3. Answer the buyer in the first 80 words. The deleted sentence was more useful than the replacement: “Flat-fee billing cleanup for dental clinics in Ohio. $450 per location. First cleanup in 14 days.” It says what the service does, who it serves, what it costs and when work starts. “Comprehensive solutions for evolving needs” says none of those things, however many times the report counts it.

4. Keep a real number on the page and keep it current. A price, turnaround time or coverage map gives a reader something concrete to carry away. I used to tell SaaS clients to hide pricing to lift demo bookings. I was wrong about that approach for AI search: removing the number also removes a useful answer. The correction is not another numerical score. It is the number the buyer came to find.

If you are shopping for landing-page optimization software, request one screen before the demo ends: the raw HTML your page sends on the first request, with pricing visible. If the vendor shows you a score instead, you have your answer. I start with a curl and CDN check that reads the site as an AI-search fetcher would before touching copy. Then I read the first 80 words aloud and check whether a buyer could quote back one useful number. None of those checks requires a dashboard add-on.

Delete the 94 and keep the answer.