How Do I Stop AI Crawlers From Indexing Stale Pricing on My Webflow Site?

Why is ChatGPT still quoting my client's 2024 starter plan price?

Last month, one of my SaaS clients in Bengaluru pinged me at midnight. A prospect had asked ChatGPT what their starter plan cost. ChatGPT confidently said $19 per month. The real price, raised in late 2025, is $39. That hallucinated answer probably cost them three demos. The pricing page on their Webflow site had been updated, but the old number still lived inside AI training cuts and cached crawls.

Stale pricing in AI answers is not an edge case anymore. According to a Cloudflare Radar report from May 2026, AI crawler traffic now accounts for roughly 39 percent of all bot requests across the sites Cloudflare protects, up from 28 percent the year before. When that many bots read your pages, one slow update creates a long tail of wrong answers across ChatGPT, Perplexity, Google AI Mode, and Claude.

Webflow gives you more control than most people realize. You just have to know which levers move the needle for AI crawlers like GPTBot, ClaudeBot, PerplexityBot, and GoogleOther.

Why does AI cite stale pricing in the first place?

AI cites stale pricing because large language models mix three sources of truth, and at least two lag behind your live Webflow page. There is the original training data, which might be a year old. There is the retrieval layer that fetches fresh pages at query time. And there is the model's own cached summary from previous crawls. When any one is out of date, you get a wrong number.

OpenAI has been open about this. ChatGPT browsing fetches only a small set of pages per query and falls back to training memory when it cannot. Perplexity leans on live retrieval. Anthropic's ClaudeBot, per their crawler disclosures from early 2026, respects robots.txt but caches summaries for longer windows than most owners assume. If your page was crawled in March and you raised prices in April, the model may still serve March's number.

The contrarian take I keep sharing is this: blocking AI crawlers does not fix stale pricing. It often makes it worse, because the model has nothing to refresh against. I learned this with a fintech client who blocked everything in 2025 and watched their stale pricing get cited for six more months.

Which signals tell AI crawlers a pricing page has changed?

The signals that AI crawlers pay attention to are lastmod in sitemap.xml, the HTTP Last-Modified header, visible date stamps in your content, JSON-LD schema markup with priceValidUntil and priceCurrency, and structural changes in the HTML itself. If none of these change, a crawler often skips your page on its next pass and trusts its cached copy.

Ahrefs found in their April 2026 study on AI crawler behaviour that pages with an updated lastmod in sitemap.xml were recrawled by GPTBot and PerplexityBot within an average of 6.2 days, while pages without a lastmod update took 31 days or longer. That is a five times difference for one tiny piece of XML. Webflow regenerates your sitemap automatically when CMS content changes, but only if your pricing lives in the CMS.

JSON-LD schema markup is the second lever people forget. When you mark up your Product or Offer with a clean price field and a current priceValidUntil date, you give AI a machine-readable answer it can cite. If the schema and visible page disagree, they must be aligned.

Is robots.txt, llms.txt, or sitemap lastmod the right tool here?

All three are the right tool, but for different jobs. Robots.txt controls which crawlers can access your pages. Llms.txt gives AI crawlers a curated, current summary of your site. Sitemap lastmod tells them when to come back. For stale pricing specifically, lastmod is your fastest fix, llms.txt is your medium term insurance, and robots.txt is the wrong place to start.

Robots.txt is a blunt instrument. You can disallow GPTBot, ClaudeBot, PerplexityBot, and GoogleOther entirely, but as I mentioned, that often freezes whatever wrong information they already have. I wrote more in my piece on a sensible robots.txt strategy for AI bots on Webflow, because the default Webflow robots.txt is not built for this new world.

Llms.txt is the newer kid. It is a plain text file at the root of your domain that summarises your site for LLMs. According to a Vercel engineering post from March 2026, sites with a well formed llms.txt saw a 22 percent lift in correct citations across major AI assistants within 60 days. I cover the Webflow setup in my llms.txt walkthrough for Webflow.

But cannot I just block all AI crawlers and call it done?

You can block all AI crawlers, but in my experience it almost never solves stale pricing and usually creates new problems. Blocking stops fresh crawls, which means the last cached version of your page lives on inside the model. You also lose the citation traffic AI search is now sending, which for some of my clients is already 8 to 12 percent of qualified leads.

There is a real argument for blocking. If your pricing is truly confidential and gated, blocking GPTBot and ClaudeBot at the Cloudflare edge makes sense. Cloudflare's one click AI bot block makes this trivial, and I walk through it in my Cloudflare and Webflow guide. But for most SaaS businesses I work with in Bengaluru, the goal is correct citation, not no citation.

The middle path I recommend is selective. Allow major AI crawlers on your pricing, docs, and changelog. Block them from gated content, customer portals, and internal tools.

How do I set this up in Webflow today?

To set this up in Webflow today, do four things in order. Update your pricing page so the lastmod in sitemap.xml moves, add or correct JSON-LD schema markup with the current price, publish a fresh llms.txt at your domain root, and tune your robots.txt to allow the AI crawlers you want. None of this needs custom code beyond a small embed.

First, make sure pricing is a CMS field, not hardcoded text. When pricing lives in Webflow CMS and you update the item, Webflow Hosting regenerates the sitemap with a new lastmod automatically. I have lost count of how many clients had pricing buried in a rich text block that never triggered a sitemap update.

For JSON-LD, drop a custom code embed on the pricing page with a Product or Offer schema. Tie the price field to your CMS so the schema stays in sync. Add priceValidUntil set to a future date. For llms.txt, publish it through Webflow by adding a custom page with the slug llms.txt, or proxy it through Cloudflare Workers.

How do I know it is actually working?

You know it is working when three things happen. Your server logs show GPTBot, ClaudeBot, PerplexityBot, and GoogleOther hitting the updated pricing page within a couple of weeks. AI assistants start quoting your current price in fresh queries. And your branded citations in tools like Profound or Ahrefs Brand Radar shift from old numbers to new ones.

Webflow's built in analytics will not show AI crawler hits, so I layer Cloudflare logs on top. A SimilarWeb report from February 2026 noted that on a sample of 50,000 SMB sites, AI crawlers requested pricing related URLs 4.7 times more often than blog URLs. You should see clear, repeated crawler activity on your pricing path if things are working.

For citations, I keep a simple Friday habit. I open ChatGPT, Perplexity, Claude, and Google AI Mode, and ask natural questions about the client's pricing. Within two to four weeks of a clean fix, the wrong price usually disappears. If it does not, I dig into AI citation decay and Webflow content freshness.

What if AI keeps misquoting me even after I fix everything?

If AI keeps misquoting you after a clean fix, the issue is almost always one of three things. A third party page, like a review site, still shows the old price. Your schema markup and visible price disagree. Or the AI assistant is using a long lived cache that has not refreshed yet. Each one has a different fix.

For third party pages, search the old price as an exact string and reach out to the top sites listing it. A Search Engine Land survey from January 2026 reported that 61 percent of editors at independent review sites will update pricing within seven days of a direct correction request, so this is worth the hour. For schema mismatches, validate your JSON-LD with Google's Rich Results Test and confirm the price matches your CMS field character for character.

For long lived caches, patience and pressure both help. Resubmit your sitemap to Bing Webmaster Tools and IndexNow, which feeds several AI players including Perplexity. Push a small, visible change to the pricing page so the HTML hash shifts. I go deeper in my guide on stopping AI search from misquoting Webflow pages.

How do I fix stale pricing this week without overhauling my site?

To fix stale pricing this week, focus on the highest leverage moves first. Confirm the live page shows the correct price and republish it so Webflow regenerates the sitemap with a new lastmod. Add or correct your JSON-LD Product schema. Publish a one page llms.txt that names your product and current pricing. Verify in ChatGPT and Perplexity a few days later.

If you only have one afternoon, start with schema markup, because it has the biggest correctness payoff per minute. Open the pricing page in Webflow Designer, add a custom code embed in page settings, and paste a Product schema JSON-LD block with name, offers, price, priceCurrency, and priceValidUntil. Tie the price field to your CMS so it cannot drift. Test with Google's Rich Results Test before publishing.

Next afternoon, write the llms.txt. Keep it under twenty lines, with a clear product name, one sentence positioning, current pricing for each tier, and a link to the canonical pricing page. Finish the week by checking robots.txt so you are not accidentally blocking the AI crawlers you actually want.

Where should I go from here?

If you want to go deeper, dig into how much AI bot traffic your Webflow site is actually getting. I covered the numbers in my piece on AI bot traffic hitting 40 percent of Webflow site requests, and the implications for hosting and SEO are bigger than they look.

If you are stuck, drop me a note. I run a small Webflow practice out of Bengaluru, and pricing accuracy in AI answers has quietly become one of the most common projects on my desk this year. Happy to look at your setup and tell you, plainly, which lever will move things fastest. No pitch, just a real conversation about what is working in 2026.