Prerender SEO: How Prerendering Works and When You Need It

There are two ways to do prerender SEO. You can build the HTML ahead of time, at deploy, so every visitor gets it. Or you can put a prerender service in front of your site that loads each page in a headless browser, caches the result and serves it to bots.

It matters more now because of AI search. Published tests, summed up in our guide to dynamic rendering and prerendering for AI crawlers, found that OpenAI's, Anthropic's and Perplexity's crawlers read raw HTML without running JavaScript. This post covers how each kind of prerender SEO works, the main services with prices as of 1 October 2026, a worksheet for estimating your render bill, the stale-cache risk and how to test the result.

How Prerendering Works: Build Time vs On Demand

Both kinds of prerender SEO end the same way: a crawler gets full HTML. They differ in when that HTML is made and who receives it.

Build-time prerendering

Build-time prerendering is the low-maintenance form of prerender SEO. web.dev, Google's web developer site, defines prerendering as "running a client-side application at build time to capture its initial state as static HTML." The output is one HTML file per route, served to people and bots alike. Most current frameworks do this with a setting: React Router's prerender list, SvelteKit's export const prerender = true, Nuxt's routeRules, Angular's RenderMode.Prerender. Astro prerenders every page by default. Our hub guide's framework table lists the switch for nine frameworks.

The limit is data that changes between builds. A page prerendered on Monday shows Monday's prices until the next build, so tie a rebuild to the events that change the page.

On-demand prerender services

A prerender service adds a layer in front of your site. Prerender.io's docs describe the flow, and the other services work much the same way:

1. Your CDN or server checks the request's user agent and decides whether it's a crawler.
2. People go straight to your normal site. Crawler requests go to the service.
3. If the page is in the cache, the service returns the stored HTML.
4. If it isn't, the service renders the page in a headless browser and returns it. Prerender.io calls this a "cache miss."
5. The service refreshes stored pages on a timer, from your sitemap, or when you call its recache API.

Because this serves bots something people don't get, it's what Google calls dynamic rendering. Google's page says it "was a workaround and not a long-term solution," and treats it as cloaking only if bots get different content. Our dynamic rendering SEO guide covers that line in detail.

Prerender Services Compared (Prices as of 1 October 2026)

These are the services you'll meet most often in prerender SEO, with prices as listed on each vendor's page on 1 October 2026. Each counts usage in its own unit, so convert your traffic before you compare.

A few notes that don't fit in the table:

• Prerender.io moved Starter to $59 with 30,000 renders on 15 September 2026, up from 25,000 renders, and ended its free plan on 15 October 2025. Its docs warn that "GPTBot, ClaudeBot, PerplexityBot, and others may not appear in older integration versions," so check your middleware's list.
• SEO4Ajax describes itself as "the solution that implements dynamic rendering." For other refresh rates, its pricing page says to contact it.
• Ostr.io bills re-renders at the same per-render rate when its opt-in "Re-cache Upon Expiration" setting is on.
• Headless-Render-API's Node middleware sends all traffic through the service by default. Its README advises against bot-only mode, citing "potential google cloaking penalties."
• Netlify made its Prerender extension generally available on 16 December 2025 and retired its older prerendering feature in stages from January to March 2026.
• Cloudflare has a tutorial on pre-rendering pages for crawlers with Browser Run's /content endpoint. You write the bot rules and the cache. Workers Paid has a $5 monthly minimum.
• Rendertron, the open-source renderer Google once suggested, was archived on 6 October 2022. Its README calls it deprecated.
• Services for AI-built apps have appeared too. As of 1 October 2026, Encited (formerly LovableHTML) lists a Basic plan at $19 a month, Hado SEO a Starter plan at $19 a month and Crawllify a Starter plan at $9 a month. Check your builder first, though: Lovable now says its apps get server rendering or static snapshots, "free on all tiers."

The Render Budget Worksheet

The bill for prerender SEO grows with three numbers: how many URLs you cache, how often each one refreshes, and how many versions of each page get rendered. We call this the Render Budget Worksheet:

Monthly renders = URLs × (30 ÷ refresh interval in days) × versions per page

Prerender.io's pricing FAQ says each page is rendered once for desktop and once for mobile, so use 2 for versions there.

Here's the worksheet for Tallyfold, a fictional invoicing app for agencies. Its marketing site is a client-rendered app with 1,200 public URLs: 80 marketing and pricing pages, 220 blog posts and 900 help articles. The plan prices are Prerender.io's, as listed on 1 October 2026.

Option B buys freshness the expensive way: most help articles don't change daily, so it pays to re-render pages that haven't changed. Option C keeps the weekly schedule and asks the service to re-render only the 80 marketing pages when prices change. Prerender.io says its recache API is available on every plan. The cost stays at $59, and the window in which bots could see an old price shrinks from up to 7 days to however long the recache queue takes.

The worksheet assumes a 30-day month and three price changes a month. Swap in your own counts. If the result is far above your plan, that's a sign build-time prerendering or server rendering would cost less.

Cache Freshness and the Stale-Page Risk

A prerender cache is a copy, and copies age. It's the part of prerender SEO that teams forget after launch. Prerender.io defines cache freshness as "how often your pages are re-rendered to stay up-to-date for visitors and search engines." On its Starter plan you can set that between 24 hours and 7 days. Between refreshes, bots get the last snapshot.

That gap causes three kinds of trouble:

• Old facts. A price, a stock level or a date changes on the live page and not in the bot copy. Search results and AI answers can quote the old one.
• Different content. Google treats dynamic rendering as acceptable when bots get "similar content." A snapshot from weeks ago may not be.
• Error pages. If a page breaks during a render, the cached copy can be the error. Prerender.io says it caches only 200 responses and drops pages that return errors on recache, which limits this. Other services document caching differently, so check yours.

Three habits keep the gap small. Trigger a recache from the same event that changes the page, such as a price update or a publish. Set shorter refresh times only on pages whose facts move. And show a visible "updated" date on pages like pricing, so you can spot a stale snapshot at a glance.

How to Test Your Prerender SEO Setup

Test a prerender SEO setup the way crawlers meet it: by user agent, without JavaScript.

1. Compare a browser request with bot requests. On a working setup, bot requests get far more words than a browser request to the same client-rendered page. The script below prints the status and word count for six user agents.
2. Check every bot you care about. Include Googlebot and the AI search bots. A bot missing from the list gets the same shell as the browser line.
3. Check a fact that changed recently. Search the bot copy for today's price or newest headline. If it's missing, your refresh schedule is too slow for that page.
4. Check status codes. Request a URL that doesn't exist. Bots should get a 404, not a 200 with a rendered error page.
5. Run Rankbox's free AI search readiness check. It fetches without JavaScript and with its own user agent, so it shows what a crawler outside your bot list receives.
6. Watch real traffic. Your prerender dashboard and server logs show which bots got cached pages. Our AI crawler log analyzer counts AI bot hits from a log file.

#!/usr/bin/env bash
# prerender-ua-check.sh: does each bot on your list get the prerendered copy?
# Usage: ./prerender-ua-check.sh https://example.com/pricing
url="$1"
agents=(
  "Browser|Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36"
  "Googlebot|Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)"
  "OAI-SearchBot|Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/131.0.0.0 Safari/537.36; compatible; OAI-SearchBot/1.4; +https://openai.com/searchbot"
  "ChatGPT-User|Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; ChatGPT-User/1.0; +https://openai.com/bot"
  "PerplexityBot|Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; PerplexityBot/1.0; +https://perplexity.ai/perplexitybot)"
  "Claude-User|Claude-User" # Anthropic names this agent but publishes no full string
)
 
for entry in "${agents[@]}"; do
  name=${entry%%|*}
  ua=${entry#*|}
  status=$(curl -sL --max-time 20 -A "$ua" -o page.html -w '%{http_code}' "$url")
  words=$(perl -0777 -pe 's/<!--.*?-->//gs; s/<(script|style|noscript)\b.*?<\/\1>//gis; s/<[^>]+>/ /g' page.html | wc -w | tr -d ' ')
  printf '%-14s HTTP %s  %6s words\n' "$name" "$status" "$words"
done

The user agents come from the vendors' docs: Google's crawler list, OpenAI's crawler page and Perplexity's crawler page. The script was run on 1 October 2026 against two pages without a prerender service, to confirm it works: a server-rendered page returned 1,627 words for all six agents, and a client-rendered page returned 6 words for all six. Behind a working prerender service, the bot lines should show the full page while the browser line stays small. Real bots also come from the vendors' IP ranges, so a firewall may treat these spoofed requests differently.

When You Need Prerender SEO, and When You Don't

Use this order of questions before you pay for any prerender SEO setup:

1. Do your important pages fail the raw HTML test? If the words are already in the first response, you don't need prerendering. Many WordPress, Shopify and server-rendered sites are in this group.
2. Can you change the build? If yes, prerender at build time or switch on server rendering. It reaches every crawler with no bot list, and Google recommends it.
3. Does the content change faster than you deploy? Prices, stock and listings suit server rendering, or a build-time setup with event-driven rebuilds.
4. Is the build out of reach for now? A legacy app, a vendor platform or a busy team is a fair reason for a prerender service. Treat it as a bridge, with a parity check and a date to revisit.
5. Are the pages behind a login? Then skip it. Pages that crawlers never reach don't need a crawler copy.

Rankbox doesn't run prerendering or a CDN, so the setup is your developer's call. It helps with what goes on the pages: it researches the questions buyers ask AI assistants and writes source-backed articles with the Citation-Ready Writer, which reach your site through Rankbox's API. Rankbox doesn't track AI citations today. The Business plan is $49.50 a month; see pricing.