HTTP caching for business sites: what to cache, what never to share
Understand HTTP caching controls and distinguish reusable public responses from private content that must not be shared.
Caching makes sites faster and cheaper to run, and set up wrongly it can show one customer another customer's page. MDN's guide to HTTP caching (https://developer.mozilla.org/en-US/docs/Web/HTTP/Guides/Caching) explains how it works. This article picks the parts a site owner should understand.
Why cache
MDN says a cache stores a response and reuses it for later requests. The closer the cache is to the client, the faster the response. When a response is reused, the origin server does not need to process the request, such as parsing it, restoring a session from a cookie, querying a database or rendering a template, which reduces server load.
Private vs shared caches
- A private cache belongs to one client, typically a browser, so it can store personalized responses for that user.
- A shared cache sits between clients and the server and can store responses for many users.
MDN's warning: if personalized content is stored somewhere other than a private cache, other users may be able to get it, which can leak information. To keep a personalized response in the private cache only, send Cache-Control: private. MDN adds that personalization is usually controlled by cookies but the presence of a cookie does not make a response private.
Fresh and stale
A stored response is fresh while its age is within its allowed lifetime and stale after. Age is the time since the response was generated, much like a time-to-live in other caches.
no-cache and no-store are different
MDN says no-cache does not prevent storing a response; it prevents reusing it without revalidation. If you do not want a response stored in any cache, use no-store. Also verify CDN and service-worker configuration: managed caches can use their own storage rules and may not follow that header automatically. MDN also notes that in practice "do not cache" usually covers three situations: privacy, always-fresh content and uncertainty about old implementations, and that no-store is not always the best choice for all of them.
Defaults and cache busting
MDN says a response without Cache-Control is not simply "not cached": browsers may apply heuristic caching. To avoid that, it prefers giving every response an explicit default Cache-Control. A common default to always get the latest version is no-cache, and for sites with logins or personalized content, add private: Cache-Control: no-cache, private.
For files that change, such as scripts and stylesheets, MDN describes cache busting: change the URL whenever the content changes (a version or content hash in the file name or query), so each URL can be cached for a long time safely.
A checklist for your developer
| Content | Suggested handling (our reading of the guide) |
|---|---|
| Pages with account or cart details | private, and consider no-cache or no-store |
| Public HTML pages | Explicit Cache-Control; decide freshness deliberately |
| Scripts and styles with version/hash in file name | Long cache lifetime |
| Anything without explicit headers | Check what heuristic caching does |
Limits
We make no performance claims. We did not read MDN's sections on validation, Vary, request collapsing or deletion in detail. Test caching rules in a staging environment with two different accounts before going live.
By Omnistu.