TTFB (Time to First Byte)

Key Definition

Time to First Byte (TTFB) is a performance metric that measures the time between sending an HTTP request and receiving the first byte of data from the server. It captures three components: DNS lookup time (resolving the domain name), connection time (establishing the TCP/TLS connection), and server processing time (the server generating the response). TTFB is the earliest indicator of your server's responsiveness and directly impacts how quickly AI crawlers can access your content. A good TTFB is under 200 milliseconds; anything above 800ms is considered slow.

Why It Matters for AI SEO

TTFB is the gateway metric for AI crawler access. Before an AI search tool can read, process, or cite your content, it must first receive a response from your server. AI crawlers operate under strict time budgets — when retrieving sources for a single user query, they may fetch dozens of pages in parallel, each with a timeout threshold measured in seconds.

If your server's TTFB is slow, AI crawlers may time out before receiving your content and move to a faster competitor. Research indicates that sites with First Contentful Paint under 0.4 seconds (heavily dependent on TTFB) are cited by ChatGPT up to 3x more often than slow sites. TTFB is the first link in this chain: if the server is slow, no subsequent optimization — content quality, structured data, Schema markup — has any chance to take effect.

For comprehensive guidance on server-side performance, see our guide on server response optimization for AI crawlers.

How It Works

TTFB is measured from the moment a client (browser or crawler) sends an HTTP request to the moment it receives the first byte of the response. This interval includes several sub-steps:

DNS resolution (5-50ms typically): The client resolves your domain name to an IP address. Using a fast DNS provider and enabling DNS prefetching reduces this step.

TCP/TLS handshake (10-100ms typically): The client establishes a connection and negotiates encryption. HTTP/2 and HTTP/3 reduce handshake overhead, and TLS 1.3 completes the handshake in fewer round trips than older versions.

Server processing (the variable component): The server receives the request, runs application code, queries databases, generates HTML, and begins sending the response. This is where the largest gains are typically available — and where poor configuration causes the worst bottlenecks.

For AI crawlers specifically, the server processing component is critical because it determines whether the crawler receives content within its timeout window. AI crawlers from OpenAI, Perplexity, and Google typically operate from US-based data centers, so geographic distance to your server adds latency. Using a CDN with edge locations near these data centers minimizes the geographic penalty.

A well-optimized page achieves TTFB under 100ms from CDN edge locations, giving AI crawlers near-instant access to your content. A poorly optimized page may have TTFB of 2-5 seconds — well beyond the timeout threshold of many AI crawlers.

Practical Implications

• Measure TTFB from AI crawler locations. Your TTFB from your own network may be fast, but AI crawlers operate from specific data center regions (primarily US). Use tools like WebPageTest or Lighthouse to measure TTFB from locations that mirror AI crawler origin points.
• Implement a CDN. A Content Delivery Network caches your content at edge locations worldwide, dramatically reducing TTFB for remote clients including AI crawlers. Cloudflare, Fastly, and AWS CloudFront are popular options that can bring TTFB under 50ms.
• Enable server-side caching. Generating HTML on every request is the most common cause of slow TTFB. Implement page caching (Varnish, Redis, Nginx FastCGI cache) so repeated requests for the same page serve from cache rather than regenerating content.
• Optimize database queries. Slow database queries during page generation inflate TTFB. Profile your slowest queries, add appropriate indexes, and consider read replicas for high-traffic pages.
• Upgrade hosting if necessary. Shared hosting environments often have TTFB of 1-3 seconds under load. VPS, dedicated servers, or managed hosting platforms provide consistent sub-200ms TTFB.
• Monitor TTFB continuously. Server performance fluctuates with traffic, deployments, and infrastructure changes. Set up automated monitoring with alerts when TTFB exceeds your threshold, as covered in our Core Web Vitals for AI SEO guide.