Does page speed affect AI search visibility? Probably, but not in the simplistic way that many SEO conversations suggest.
There is currently no published evidence from OpenAI that a faster Largest Contentful Paint, a perfect PageSpeed Insights score or passing every Core Web Vital directly makes a business rank higher in ChatGPT Search.
What we can say with much more confidence is this:
A crawler cannot use content it fails to retrieve.
Slow server responses, timeouts, repeated 5xx errors, 429 rate limits, long redirect chains and bot-protection problems can all interfere with crawling. Google explicitly adjusts crawling based partly on server response times and site health. OpenAI also tells site owners to investigate unsuccessful crawler access, including 403 and 429 responses, firewalls, CDNs and bot mitigation.
So the useful distinction is:
Page speed as a ranking factor is one question. Page speed and server performance as a crawlability issue is another.
For AI search, the second question is currently much easier to defend.
Table of contents
- What do we actually know about page speed and AI search?
- Is page speed a GEO ranking factor?
- Do Core Web Vitals matter to ChatGPT?
- Why server response time matters for crawlers
- What does TTFB have to do with AI crawling?
- Do AI crawlers have a crawl budget?
- Slow server response vs slow page rendering
- How to check whether speed is affecting crawler access
- What should you fix first?
- How fast is fast enough?
- Frequently asked questions
What do we actually know about page speed and AI search?
Google’s crawl-budget documentation explains that stable response times, including latency and Time to First Byte, can increase crawl capacity, while slower responses, 5xx errors and 429 rate limiting can reduce crawling.
There are three different issues that often get mixed together.
1. Human page experience
A faster site is generally easier and less frustrating for people to use.
Google’s Core Web Vitals measure real-world loading performance, responsiveness and visual stability. Google says Core Web Vitals are used by its ranking systems as part of broader page-experience considerations.
2. Crawl efficiency
A crawler has to request a page and receive a usable response before it can do anything useful with the content.
Google explicitly says stable or improving response times, including latency and Time to First Byte, can increase crawl capacity. If a site becomes slower or starts returning server errors or 429 responses, Google can reduce crawling.
3. AI search ranking or selection
This is where evidence becomes much thinner.
OpenAI documents how to allow OAI-SearchBot so pages can be surfaced in ChatGPT Search. It does not publish a Core Web Vitals target, PageSpeed score or TTFB threshold that determines whether one page is recommended above another.
That difference matters.
Is page speed a GEO ranking factor?
There is currently no strong public evidence that allows us to say:
Faster site = higher AI search ranking.
That would overstate what is known.
OpenAI explains that OAI-SearchBot is used to surface websites in ChatGPT Search and recommends allowing the crawler in robots.txt and through the relevant network controls. Its public crawler documentation does not list Core Web Vitals, PageSpeed Insights scores or a specific response-time metric as ChatGPT Search ranking factors.
So if somebody tells you that reducing LCP from 3 seconds to 2 seconds will directly improve your ChatGPT ranking, ask for the evidence.
The more defensible argument is:
Technical performance can influence whether content is successfully fetched, rendered or processed.
That is a crawlability issue before it is a ranking issue.
Do Core Web Vitals matter to ChatGPT?
There is no published evidence that ChatGPT Search uses Google’s Core Web Vitals as direct ranking signals.
Core Web Vitals are Google’s metrics for user experience:
- LCP measures loading performance
- INP measures responsiveness
- CLS measures visual stability
Google recommends an LCP within 2.5 seconds, INP below 200 milliseconds and CLS below 0.1 for a good user experience.
Those targets are worth working towards because users still visit your website and Google Search still matters.
But they should not be presented as ChatGPT optimisation thresholds.
For broader organic performance work, see Lobo Digital’s SEO services:
Why server response time matters for crawlers
Before a crawler can analyse your beautifully designed page, your server has to answer.
That is why server performance deserves attention.
Google’s current crawl-budget documentation says its crawl capacity can increase when a site responds consistently and its latency and TTFB remain stable or improve.
When response times become longer, or a server starts returning 5xx or 429 responses, Google can reduce crawling.
Google also notes that making pages efficient to load can allow it to retrieve more content.
We should not automatically assume OpenAI uses the same crawl-budget algorithm.
But the underlying web mechanics remain relevant.
A crawler still has to:
- connect to the server
- receive a response
- retrieve the HTML and resources it needs
- process that information successfully
If step two repeatedly fails, the rest does not matter.
What does TTFB have to do with AI crawling?
Time to First Byte, or TTFB, measures how long it takes from requesting a resource until the first byte of the response begins to arrive.
It reflects factors such as:
- redirects
- DNS
- connection setup
- TLS negotiation
- backend processing
- server response time
Google’s web.dev guidance gives 0.8 seconds or less as a rough target for most websites. Importantly, it also says TTFB is not itself a Core Web Vital.
That 0.8-second number is useful as a web-performance benchmark.
It is not an official OAI-SearchBot requirement.
Do not turn a general performance recommendation into a fictional GEO rule.
For a broader technical benchmark, web.dev’s TTFB guidance recommends aiming for a TTFB of roughly 0.8 seconds or less for most websites
Do AI crawlers have a crawl budget?
All crawlers operate with finite computing and network resources.
But we need to be precise with terminology.
Google publishes detailed documentation about its crawl-capacity systems and how site health influences crawling.
OpenAI does not currently publish an equivalent formula for OAI-SearchBot showing:
- a fixed crawl budget per website
- a maximum response time
- a timeout threshold
- how frequently a failed URL will be retried
- a Core Web Vitals requirement
So I would not say:
AI crawlers time out faster than Googlebot and rarely come back.
We do not have enough published evidence for that.
What we can say is that OpenAI advises site owners to inspect unsuccessful access, 403 responses, 429 rate limiting, firewalls, CDN logs and bot-mitigation rules when its crawlers cannot access content.
That is strong enough to justify monitoring technical delivery without inventing a crawl-budget rule.
Slow server response vs slow page rendering
These are related problems, but they are not identical.
Slow server response
The crawler waits before the HTML even starts arriving.
Possible causes include:
- slow hosting
- uncached database requests
- heavy server-side processing
- poor CDN configuration
- redirect chains
- overloaded infrastructure
Slow rendering
The HTML arrives, but substantial work is still required before meaningful content appears.
Possible causes include:
- large JavaScript bundles
- third-party scripts
- client-side API requests
- complex widgets
- JavaScript-generated content
Google confirms that both server response time and rendering time can matter to crawling efficiency.
For AI crawlability, however, I would prioritise reliable HTTP delivery and meaningful initial HTML first.
A fast animation is irrelevant if the crawler receives a 503.
And a perfect LCP score will not help OAI-SearchBot if your firewall returns 403.
This connects directly with our guide on JavaScript and AI visibility:
Why your website may be invisible to AI search because of JavaScript
How to check whether speed is affecting crawler access
Do not rely only on PageSpeed Insights.
Check what crawlers actually receive.
Review server or CDN logs
Look for crawler requests and record:
- URL requested
- user-agent
- HTTP response
- response time
- timestamp
- source IP
- redirects
- WAF action
Cloudflare’s AI Crawl Control can show AI crawler requests, successful responses and unsuccessful requests, as well as status-code distributions.
Watch for these responses
What should you fix first?
If the goal is AI crawlability, I would work in this order:
| Priority | What to check |
|---|---|
| 1 | Public pages reliably return HTTP 200 |
| 2 | Server response time and TTFB |
| 3 | 403, 429 and 5xx crawler responses |
| 4 | CDN, firewall and bot-protection rules |
| 5 | Long or unnecessary redirect chains |
| 6 | Important content in initial HTML |
| 7 | Heavy JavaScript and third-party scripts |
| 8 | Images, fonts and non-critical resources |
| 9 | Core Web Vitals and wider page experience |
Notice that chasing a PageSpeed score of 100 is not first.
A crawler needs a reliable response more than it needs a perfect Lighthouse screenshot.
How fast is fast enough?
There is no published universal AI crawler speed threshold.
For general web performance, Google’s web.dev guidance says most sites should aim for a TTFB of roughly 0.8 seconds or less.
For Core Web Vitals, Google’s current “good” targets are:
- LCP: 2.5 seconds or less
- INP: 200 milliseconds or less
- CLS: 0.1 or less
Those are useful performance benchmarks.
They are not known ChatGPT ranking requirements.
For GEO, I would focus first on:
fast enough, stable, accessible and consistently crawlable.
Not:
perfect score at any cost.
Frequently Asked Questions About Page Speed and AI Search
Common questions about page speed, Core Web Vitals, TTFB, AI crawler access and whether website performance directly affects AI search visibility.
Is page speed a GEO ranking factor?
There is currently no published evidence showing that a specific page-speed metric directly improves rankings across AI search platforms.
Performance still matters technically because pages must be successfully accessed and processed before their content can be useful.
Do Core Web Vitals matter to ChatGPT?
OpenAI does not publicly document Google’s Core Web Vitals as ChatGPT Search ranking factors.
They still matter for user experience and Google Search, so improving them remains worthwhile.
How fast should my website be for AI crawlers?
There is no published OAI-SearchBot speed threshold.
As a general web-performance benchmark, web.dev recommends aiming for TTFB of around 0.8 seconds or less for most sites, but this should not be presented as an AI crawler requirement.
Can a slow website stop AI crawlers?
Potentially, particularly if poor performance produces timeouts, server errors, rate limiting or other failed requests.
A page being slightly slower than ideal does not automatically mean an AI crawler cannot access it.
Does a PageSpeed score of 100 improve ChatGPT visibility?
There is no published evidence that ChatGPT rewards a PageSpeed Insights score of 100.
Prioritise reliable access, useful content, clear HTML and actual user experience over chasing a perfect score.
Is TTFB more important than LCP for AI crawlers?
They measure different things.
TTFB reflects how quickly the server begins responding. LCP measures when the largest visible content element appears for the user.
For basic crawler access, the server must respond before rendering can even begin. For users and Google page experience, both wider loading performance and Core Web Vitals remain relevant.
Speed matters, but make the right diagnosis
A slow website is worth fixing.
But “make the site faster so ChatGPT ranks it higher” is too simplistic.
We know that reliable, efficient servers help crawlers retrieve content. We know errors, rate limits and blocked requests can interfere with access. We know Google uses Core Web Vitals within its wider page-experience systems.
What we do not currently have is evidence that ChatGPT uses Google’s speed metrics as direct ranking factors.
That uncertainty should change how you prioritise the work.
Start with:
Can the crawler reach the page?
Then:
Does the server respond reliably and quickly?
Then:
Can the important content be retrieved without unnecessary rendering work?
After that, improve page experience for the people who actually click through.
If you are unsure whether technical issues are limiting how AI platforms find and understand your website, Lobo Digital’s audit looks at the wider picture rather than one speed score.
