A client sends me a tool that promises to get their paid landing page into Google AI Overviews. I would rather build a different page. AI Overviews quote passages that answer questions; a landing page is built to turn a visitor into a customer. Forcing one URL to do both jobs usually makes each job harder. The trouble starts well before anyone installs the tool.
Myth 1: “Adding schema markup will get us into AI Overviews”
I understand the appeal. Structured data gives a vague ambition a tidy technical task: add FAQPage or SoftwareApplication markup, check that the code works, and wait for citations. It is much easier to put a JSON-LD script in a header than to write a useful answer to a difficult question. It also gives an agency something concrete to put on an invoice.
But schema helps a machine interpret information; it does not make the information worth quoting. Marking a string of digits as a phone number is useful. It does not tell a retrieval system why your paragraph answers a searcher better than another paragraph. In an experiment across 29 client domains and 36 service locations documented by Evergrow Marketing, removing schema for five weeks and restoring advanced LocalBusiness markup for another five produced no statistically significant movement in AI Overview visibility or Google Search rankings.
Google Search Central’s guidance says sites do not need specialized schema to appear in AI Overviews. Keep markup where it helps describe the page. Do not buy a schema package as a substitute for writing the answer the page is supposed to give.
Myth 2: “An llms.txt file turns citations on”
This one became popular because an engineer can ship an llms.txt file in twenty minutes. Put a short Markdown summary in the root directory, call it AI readiness, and marketing gets to close a ticket without changing a sentence on the site. I have seen less convenient work postponed for less satisfying reasons.
The problem is that publishing a file is not the same as getting search systems to use it. In Limy.ai’s analysis of more than 500 million AI crawler requests over 90 days, only 408 requests targeted an llms.txt file directly. The draft’s underlying issue remains the page itself: if its facts are thin or hidden behind heavy scripts, a neat summary elsewhere does not repair them. Google Search uses its standard crawling and indexing infrastructure, not that experimental file, to evaluate pages.
An llms.txt file may be quick to make. So are racing decals. Fix the engine first.
Myth 3: “We have to rank #1 organically before we can be cited”
Years of SEO reports train people to treat the top organic position as the only seat worth having. It is a useful instinct for some search goals, but it makes teams dismiss answer pages that could help with AI visibility before those pages get a chance.
An organic ranking and an AI citation are not the same selection. In an Ahrefs study covering 863,000 keyword results and 4 million cited URLs, 37.9% of cited URLs ranked in the organic top 10. Another 31.2% ranked between positions 11 and 100, while 31.0% were outside the top 100. Being outside the top 10 plainly does not rule a page out.
That is not permission to ignore SEO or publish a paragraph on a broken site. It means I would not wait to win position one before making a specific answer easy to find and quote. Work on the passage in front of you rather than treating a domain-wide rankings campaign as the entry fee.
Myth 4: “A tool that monitors AI Overviews will get us into them”
A dashboard can make a bad month look impressively managed. For $499 a month, a vendor might track whether Gemini mentions your brand across hundreds of prompts, draw animated trend lines, and send a weekly PDF. Someone can forward the PDF to the board. That feels like progress because everybody can see it.
The tracker records a result; it does not change the page that produced it. As Ansvisor’s explanation of AI Overview trackers puts it, these tools observe search results, record cited domains, and chart changes. None of that puts a clearer answer in your HTML. It is like buying a Google Ads reporting tool and expecting it to prune search terms or adjust bids. It will graph your conversions falling without lifting a finger.
I am not against knowing where you stand. I am against confusing the scoreboard with the work. If a competitor appears for a question you should answer, inspect the query, identify what your page fails to explain, put a concise answer where a crawler can read it, and address rendering problems. Our look at tools for getting featured in Google AI Overviews makes the same distinction: monitoring can point to a problem; execution has to fix it.
Myth 5: “More pages mean more citations”
Automated drafting makes this one hard to resist. If ten location or industry pages give you ten chances to appear, surely five hundred near-identical pages give you five hundred. That is the citation lottery theory: print enough tickets, change the city name on each, and wait.
The flaw is in what those pages contribute. Changing a noun is not adding an answer. Forty versions of a sales page, each with a different location in the heading, still offer little to extract if the underlying copy says the same thing. They can also leave a site full of repetitive, low-factual-density pages. Volume does not solve the problem that made the first page a poor citation candidate.
I would put the drafting effort into fewer pages that address distinct questions with direct, useful explanations. A page about one question should earn its existence by answering that question, not by wearing a different city name. If the plan starts with a page count rather than the questions readers ask, stop before generating page 500.
Myth 6: “Technical SEO is a one-time fix”
A team audits its site in January, fixes three canonical tags, and declares the technical work done. Then someone updates the landing-page builder. An ecommerce plugin changes. The design team replaces a template on Thursday evening. PPC operators know the Friday-morning version of this story: a page changes without warning, and the ad account gets the blame for what happened downstream.
AI visibility can fail more quietly. If the answer is not available to a crawler when it reads the page, polished copy elsewhere on the site will not rescue that passage. Alneeko’s guide to JavaScript rendering and AI crawlers describes the risk of putting core content behind client-side JavaScript. Googlebot can queue a page for rendering through its Web Rendering Service, but that extra step takes time; other evaluation passes may encounter the initial HTML before the content appears. A factual answer, comparison, or pricing detail trapped in a client-rendered component is harder to retrieve than plain text delivered with the page.
That does not mean every script is a crisis. It means a clean audit from four months ago tells you little about what a crawler can read today. After a redesign or a sudden loss of visibility, check the page again. Our technical SEO diagnostics for AI visibility are a better next stop than ordering another round of copy before you know whether the existing copy is visible.
Myth 7: “Our landing page should be the page that gets cited”
This is the one I find hardest to kill, because the reasoning starts in the right place: revenue. If you spend $20,000 a month on Google Ads, you have probably spent more time polishing the page those clicks reach than any article on your site. Of course you want Google to quote its offer and send searchers straight to the sign-up form. The page matters. It is just built for a different job.
A paid landing page usually removes detours. It leads with the offer, clears objections, and keeps the call to action in view. An AI Overview needs a passage that answers the searcher’s question. As the Search Engine Land analysis of AI Overview queries illustrates, these results lean toward informational searches. A short sales pitch, pricing table, and form are not much of an explanation. Adding a long definition above the offer to feed a crawler may give a reader the wrong page at the wrong moment.
The placement matters, too. In CXL’s analysis of 100 AI Overview citations, 55% of cited passages appeared within the first 30% of the source page. That is a useful argument for putting a direct answer early on an informational page. It is not a reason to bury the call to action on a paid landing page under a wall of definitions. One page does not have to win every kind of visit.
The better division is simple: let the landing page convert paid traffic, and give the question its own answer page. Put the explanation up front, then link naturally to the relevant commercial offer for readers ready to take the next step. Our passage-first guide to getting cited in AI Overviews goes deeper on that page structure.
If an agency is currently selling you AI Overview optimization for your commercial landing page, I would change that scope on Monday. Keep the page focused on the offer that is converting. Then give the team three jobs:
- Split the jobs across two URLs. Keep the paid landing page tuned for conversion. Build a dedicated answer page for the informational question, with a clear response near the top and a contextual link to the offer. The link gives an interested reader a route into the funnel without making the funnel page pretend to be an encyclopedia.
- Check what crawlers can actually read. Use Google Search Console’s URL Inspection tool to examine the page, and check the initial HTML as well as rendered content. If a key definition or comparison appears only after client-side JavaScript runs, consider moving that content into HTML delivered with the page. Do this before paying someone to rewrite an answer the crawler may not be seeing.
- Pay for work, not just observation. A tracker can show where competitors appear; it cannot write the missing answer or fix the page. If you need ongoing execution across paid search and organic visibility, that is why we built groas: specialized AI models manage campaign bids, monitor intent shifts, and optimize search visibility continuously, while a named human strategist sets direction and guardrails. The point is action tied to outcomes, not another chart to discuss.
I keep hearing that the fastest route into AI Overviews is to tweak the landing page harder. It is an appealing answer because the page already exists and somebody has a tool to sell you. But the sales page should sell; the answer page should answer. Give each one its own job, and link them together.