Skip to content
NewProduct Q&A: shoppers ask on your product page and get your answers.See how it works
IdukkiIdukki
AI search

Why Your Highest-Converting Page Is Invisible to AI Engines (And the 3-Line Fix)

Your best-converting page is usually client-rendered, so AI engines read it as blank. The fix is a three-line move to SSR you can ship in an afternoon.

Your best-converting page had a strong month. Ask ChatGPT a question it answers, and another store gets cited. Fetch the page with curl and you see why: the raw HTML is almost empty. The audit takes ten minutes and the fix is three lines.

Open your funnel report. Find the bottom-of-funnel page that converts hardest. It is usually a comparison page, an "as featured in" page, or a hand-curated collection. Now fetch that same URL with curl and read the raw HTML that comes back.

In this article

Is the content missing from that raw HTML until JavaScript runs? Then you have found your biggest AEO gap in one command. The page doing the most to convert shoppers is the page AI engines read as blank. It is one of the most common faults on JavaScript-rendered storefronts.

Why does this happen?

Modern storefronts split content two ways: server-rendered HTML in the initial payload, and client-side hydration that JavaScript fetches after the page loads. High-converting pages tend to sit on the hydrated side, because they got built late, by product teams, on component frameworks that default to client rendering. Nobody decided this. It is just where those pages land.

Googlebot copes: it runs the JavaScript and waits for hydration before it indexes. GPTBot does not wait. Nor does ClaudeBot. Nor PerplexityBot. They take the initial HTML payload and leave. Content that is not in that payload simply does not exist to the agent. The crawler behaviour behind this split is documented in GPTBot vs Googlebot divergence.

How do I detect a blind page?

The fastest check is one command:

bash
curl -A "GPTBot" https://yourdomain.com/your-highest-converting-page | grep "expected-content-string"

Empty grep means the content is not in the raw HTML. Run it across your top 20 bottom-of-funnel pages. Every page that comes back empty is an AEO blind spot, and now you have the list.

Rendering modeGooglebotGPTBot / ClaudeBot / PerplexityBot
Server-side rendered (SSR)Reads fullyReads fully
Static / ISR (pre-rendered)Reads fullyReads fully
Client-rendered (hydrated via JS)Reads after running JSReads as blank
Content injected via useEffectReads after hydrationNever sees it
Rendering mode vs what each crawler can read.

What an AI crawler reads

Client-rendered page

  • Googlebot reads it after running JS
  • GPTBot, ClaudeBot and PerplexityBot read it as blank
  • Content injected via useEffect is never seen

SSR or static page

  • Googlebot reads it fully
  • GPTBot, ClaudeBot and PerplexityBot read it fully
  • The content is in the raw HTML curl returns

What is the three-line fix?

On Next.js, Astro, Remix, SvelteKit or Nuxt, the fix is to move the page from client-rendered to server-rendered or static. In Next.js that is, almost literally, three lines:

typescript
// Add to the top of your page file:
export const dynamic = 'force-dynamic'
export const fetchCache = 'force-no-store'
// Then move your data fetching from useEffect into the Server Component.

On Shopify with a Liquid theme, render the section in Liquid (server-side) instead of loading it through JavaScript. Either route, the change is small and well-bounded, the kind of thing that fits in one PR.

What to convert first

A long list of client-rendered pages is normal. Work it in this order.

  1. 1Pages with the highest conversion rate. These move revenue per AI-engine visitor the most.
  2. 2Pages cited heavily by your competitors but not you. Closing the citation gap on a comparison page is high-leverage.
  3. 3Product detail pages with the highest sessions. Volume + conversion.
  4. 4Resource and FAQ pages. These get quoted the most by agents in shortlists.

Common objections

"We need client rendering for personalisation"

Render the structured content server-side and personalise the visual layer client-side. The agent reads the static HTML, the visitor gets the personalised hydration, and you stop having to choose.

"Our pages are too dynamic for SSR"

Most pages are far less dynamic than the engineering team believes. Variant data, reviews, FAQ copy: all of it changes on the order of hours or days. Almost none of it changes per request. ISR on an hourly revalidate covers 95% of those cases without anyone noticing the cache.

"It is too much work"

Budget 2-8 hours of engineering per page. The lift in AI citation rate is reliable enough that the work tends to pay for itself inside the first 14 days after launch.

How do I measure the lift?

Take a baseline before you touch anything: curl the page as GPTBot once a week, count distinct AI-engine referrals per week, and count how often the URL gets cited in a fixed set of test queries. The same SSR move helps Core Web Vitals when you reach for static generation, which we cover in Core Web Vitals for UGC widgets.

What to watch for after the change (illustrative ranges, not measured data):

  • GPTBot fetches double or triple within two weeks as the URL becomes useful to the crawler.
  • AI-engine citations to the page rise 3-8x within 30 days.
  • Referrer sessions from chatgpt.com, claude.ai, perplexity.ai rise 4-12x within 60 days.

Closing

This is the cheapest AEO win on the board right now. Run the curl check this afternoon, and the pages that come back empty are your three-line-fix backlog. A half-day audit usually turns up five to fifteen of them, and the engineering fits comfortably inside a sprint. Once the page is readable, the next lever is the structured data on it, which the 42-signal AEO scorecard lays out in full.

  • 41%

    of 18-34s use GenAI to search products

    Bazaarvoice 2025 SEI

  • $2.6T

    Global social commerce 2025

    eMarketer

  • +108%

    TikTok Shop US growth YoY 2025

    eMarketer

AEO + AI shopping benchmarks.

Sources & notes

  1. 1Bazaarvoice, 2025 Shopper Experience Index · +144% conversion / +162% RPV among UGC-engagers; +354% conversion on PDPs with reviews vs without.
  2. 2Model Context Protocol (Anthropic) · Open standard for exposing tools that AI agents can invoke. Transport layer of agentic commerce.
  3. 3Schema.org · Shared vocabulary for structured-data markup. Defines Product, Review, AggregateRating, FAQPage, DefinedTerm types.
  4. 4Google web.dev, Core Web Vitals + conversion · Brands meeting all three CWV thresholds see 24% lower bounce rates than brands missing any of them.
#ssr#js-rendering#aeo#technical

Continue reading

4 pieces in this cluster

These long-form pieces on the Idukki blog link back to this article, go deeper on the cluster.

More from Rohin Aggarwal