xSpeed Cache
Get Plugin
Features Pricing Reviews FAQ
Documentation Blog Support Community Free vs Pro Compare
Get Plugin

Adaptive Cache Intelligence

4 min read

Adaptive Cache is an AI advisor for your cache expiry time. It reads the last 24 hours of cache traffic, works out a recommended TTL, and lets you apply that number into the Cache module with one click.

What this does

Adaptive Cache never touches your cache contents. It only recommends a number: how many hours cached pages should live before they are regenerated.

The recommendation is calculated, not guessed by the AI. Adaptive Cache reads the 24-hour hit and miss counters plus the hourly traffic buckets, gives your traffic a one-word character, and computes the TTL from that. The AI’s job is limited to writing the short paragraph that explains the recommendation.

The recommended TTL is always kept inside a safe band of 1 to 168 hours (7 days). The AI cannot move the number outside that band, or move it at all.

Traffic characters

CharacterWhat it means
low_trafficToo few requests in 24 hours for a reliable signal.
cold_cacheThe cache hit ratio is low.
spikyHourly traffic varies a lot, with peaks far above the average.
stable_warmSteady traffic with a high hit ratio.
normalTraffic and hit ratio are within normal bands.

A stable, warm cache leads to a longer TTL, because pages are being reused and a longer life means more hits and less server work. A cold cache or very low traffic leads to a shorter, more conservative TTL. Spiky traffic gets a moderate TTL that keeps peak-hour pages reusable without holding stale content through quiet hours.

Why use it

Picking a cache expiry by hand is a trade-off between freshness and hit rate, and the right answer depends on how your site is actually being visited. Adaptive Cache does that reading for you and explains the trade-off in plain language, while the safe band keeps the suggestion from going somewhere risky.

Requirements

  • xSpeed Cache Pro must be active.
  • An AI provider must be configured in the AI Provider panel.
  • The Cache module is where the recommended TTL is applied.

Note: Adaptive Cache sends aggregated counters, not visitor data, so its prompt is allowed to pass the AI Privacy gate. The AI provider connection still has to be configured for the AI-written reason to appear.

Note: If the AI provider is not configured, or the AI call fails, or there was no traffic in the last 24 hours, you still get a recommendation. The reason text simply comes from the built-in explanation instead of the AI.

How to use it

  1. Open the Adaptive Cache panel in the xSpeed Cache dashboard.
  2. Review the traffic profile: the character, the 24-hour request total and hit ratio, the average and peak per hour, and your current TTL.
  3. Run Analyze to refresh the profile and get a new recommendation.
  4. Read the recommended TTL, how it differs from your current TTL, and the reason.
  5. Click Apply to write the recommended TTL into the Cache module.
  6. (Optional) Use the flush action to clear the stored recommendation.

What Apply changes

Apply writes one value: the cache expiry, in hours, in the Cache module. Every other Cache setting, such as excluded URLs and cookies, is left exactly as it was.

Important: A longer TTL means visitors may see slightly older content between regenerations. A shorter TTL means fresher content but more work for your server. The recommendation is a starting point, not a rule.

Best practices

  • Let your site collect a full day of traffic before trusting the recommendation. With very low traffic, Adaptive Cache deliberately stays conservative.
  • Re-run Analyze after a traffic change, such as a campaign or a seasonal peak.
  • If you publish frequently, be careful about applying a long TTL. Consider how quickly you need edits to show up.
  • Check the Cache module after applying to confirm the new expiry value.

Troubleshooting

The character says low_traffic and the TTL barely moves. That is by design. With fewer than about 50 requests in 24 hours, the data is too thin to recommend confidently, so the TTL stays conservative.

The reason is not AI-written. Either the AI provider is not configured, the AI call failed, or there was no traffic in the last 24 hours to analyze. The recommendation itself is still valid because it comes from the calculation, not the AI.

Apply returns “No recommendation available.” Nothing has been analyzed yet. Run Analyze first, then Apply.

The traffic profile shows zeros. Cache traffic counters come from xSpeed Cache Free. If they are not recording, there is nothing for Adaptive Cache to read.

FAQ

Does the AI choose my cache expiry? No. The number is calculated from your traffic and clamped to the 1 to 168 hour band. The AI only writes the explanation.

Does this clear or rebuild my cache? No. It only changes the expiry setting when you click Apply.

How far back does it look? The last 24 hours of cache hits and misses, broken into hourly buckets.

Can I still set the TTL myself? Yes. The cache expiry remains a normal setting in the Cache module. Adaptive Cache just offers a suggested value.

  • Connecting an AI Provider
  • Page Cache
  • AI Privacy and GDPR Mode
  • Smart Image AI