When a Sitecore site gets slow, the cause is rarely exotic. In the audits I run, the same dozen issues come up again and again. This checklist puts them in the order that usually gives the biggest gain for the least risk. It applies to Sitecore XP and XM 9.x and 10.x.
Measure before you change anything
Write down a baseline first, or you won't know which change helped. Useful built-in tools:
/sitecore/admin/cache.aspxshows every cache, its current size and its limit. Caches sitting at their limit are a red flag./sitecore/admin/stats.aspxshows rendering statistics, including how often each rendering runs and how often it's served from cache.- The Sitecore debugger (open a page in debug mode from the Experience Editor ribbon, or add
sc_debug=1on a non-production instance) gives a per-rendering timing profile. - Your APM tool (Application Insights, New Relic or similar) for server response time, SQL calls and exceptions.
Record time to first byte (TTFB) for five or six key templates, such as the home page, a listing page, a detail page and search results. That's your baseline.
1. Turn on HTML caching for renderings
This is the single biggest win on most sites. A rendering with HTML caching enabled is executed once and then served from memory until the next publish. In the rendering definition or on the presentation details, tick Cacheable, then choose the narrowest "vary by" option that still renders correctly:
- Vary By Data for components that depend on their datasource.
- Vary By Parm when rendering parameters change the output.
- Vary By Query String for paging and filtering.
- Vary By User only for genuinely personal output. It multiplies the cache size.
Also check that cacheHtml="true" is set on your site definition, and that the HTML cache clearer runs for your site on publish:end and publish:end:remote. If it doesn't, editors will see stale pages after publishing and someone will switch caching off again.
2. Size the data, item and prefetch caches
Default cache sizes are conservative. If cache.aspx shows the web[data] or web[items] caches at their limit, Sitecore keeps evicting and reloading items from SQL. Raise the limits per database, or globally through the Caching.* settings, and keep an eye on memory. On servers with plenty of RAM, some teams set Caching.DisableCacheSizeLimits to true. That works, but watch the app pool's memory for a week afterwards.
3. Prefetch the items every request needs
Prefetch caches load items when the application starts. Add your site root, global settings items and navigation folders to the prefetch config for the web database, so the first visitors after a recycle don't pay the cost.
4. Remove expensive item queries from renderings
Look for GetDescendants(), Axes.GetDescendants() and Sitecore Query patterns like //*[@@templatename='Article'] in rendering code. On a large content tree these walk thousands of items per request. Replace them with a Content Search query against Solr, or cache the result.
5. Switch off xDB tracking if you don't use it
Many XP installations never use personalization, analytics or marketing automation, but still pay for tracking on every request. If that's you, set Xdb.Enabled and Xdb.Tracking.Enabled to false. Confirm with the marketing team first. If you'll never use xDB, consider whether you need XP or could license XM.
6. Keep Solr healthy
- Give Solr enough Java heap, and run it on its own server for production.
- Review the
autoCommitandautoSoftCommitsettings in each core'ssolrconfig.xml. Very frequent hard commits slow both indexing and queries. - Check index update strategies. The defaults are usually fine (synchronous for master, asynchronous on publish end for web), but a heavily edited master index can benefit from an interval-based strategy.
- Exclude fields and templates you never search on, so indexes stay small.
7. Maintain the SQL databases
Sitecore databases need routine index maintenance like any other SQL Server workload. Schedule index rebuild/reorganize jobs and statistics updates. Then check the size of the housekeeping tables:
SELECT COUNT(*) FROM EventQueue;
SELECT COUNT(*) FROM PublishQueue;
SELECT COUNT(*) FROM History;
Millions of rows here usually means the cleanup agents aren't running or their intervals are too long. Large EventQueue tables slow down every server that polls them.
8. Serve media from a CDN
Put a CDN in front of /-/media/, set a sensible MediaResponse.MaxAge, and keep media request protection on so attackers can't make your server resize images endlessly. Serve images at the size they're displayed.
9. Fix the front end, too
A fast server can still deliver a slow page. Check Core Web Vitals in PageSpeed Insights: render-blocking scripts, oversized hero images and images without width and height are the usual offenders. Defer non-critical JavaScript and load third-party tags after the page renders.
10. Turn down logging in production
A log level of DEBUG or a component throwing the same handled exception thousands of times per hour costs real CPU and disk. Set production logging to INFO or WARN and fix the noisiest exceptions.
11. Warm up after deployments and recycles
Enable IIS Application Initialization and point it at your most important pages, so the first real visitor doesn't trigger a cold start. Avoid scheduled app pool recycles during business hours.
12. Remove what you don't use
Open /sitecore/admin/showconfig.aspx and look for disabled modules that still load, duplicate patches and scheduled agents running too often. Every unused pipeline processor runs on every request.
What to do next
Work through the list in order, re-measure after each change, and keep notes. Most sites see the biggest improvement from items 1, 2, 4 and 7. If you'd like a second pair of eyes, my Sitecore health check covers all twelve, with a written report of what to fix first.
