Something on your site is wrong and no error message names a cause. The standard advice is to deactivate your plugins one by one until the problem stops. On a site with thirty plugins that is up to thirty deactivations, thirty page loads, thirty reactivations, and a live site behaving unpredictably for the whole of it.
There is a faster method that most published troubleshooting never mentions, and there is a trap in the standard method that costs people their real cause. Both are below.
Before you start: three preconditions
A tested backup, or an accepted risk. Deactivating a plugin is usually harmless and occasionally is not. Membership, e-commerce and form plugins can behave badly when switched off and on with live traffic passing through. Know which of yours fall into that group.
A written list. Before you touch anything, write down which plugins are currently active. If a test goes sideways, that list is how you get back to where you started. The Plugins screen gives you this in one look and it takes a minute.
A quiet hour, or a copy of the site. The right place to do this is a staging copy, where nothing you do reaches a customer. Our guide to testing on a copy of your site first covers setting one up. If you have no staging site, pick your quietest hour and accept that the site will be strange for ten minutes.
The trap in the one-at-a-time method
The classic method has a flaw that is rarely stated. You deactivate plugin 1, the fault persists, so you reactivate it and move to plugin 2. That sequence assumes the fault is caused by exactly one plugin acting alone.
Conflicts are, by definition, not that. If plugins A and D are fighting, turning off A alone fixes it, and turning off D alone also fixes it, and turning off B, C, E and everything else changes nothing. You get the right answer for the wrong reason and you never learn that the other plugin is involved.
Worse is the case where the conflict is between a plugin and your theme, which the plugin-only sweep will never find at all.
So the test is not "which plugin is broken". It is "which combination reproduces the fault", and the method below is built for that.
The method: halve, do not step
Deactivate half your plugins at once, not one.
Round 1. Thirty plugins. Deactivate fifteen. Check the fault.
Round 2. If the fault is gone, the cause is in the fifteen you turned off. Reactivate about seven or eight of them. Check again.
Round 3 onward. Keep halving whichever group still contains the fault.
Thirty plugins resolves in about five checks. A hundred plugins resolves in about seven. The arithmetic is the whole trick, and it is the reason this test takes ten minutes rather than an afternoon.
If the fault is still there after round 1, the cause is in the half you left running. Same procedure, other half.
If the fault survives with every plugin deactivated, plugins are not the cause. Switch to a default WordPress theme next. If it clears then, the theme is your answer. If it survives both, the cause is core, the server, or your content, and no amount of plugin work will find it.
Confirming, not assuming
You have one plugin left and the fault follows it. That is a strong result and it is worth one more minute to make it a certain one.
Turn everything back on except the suspect. If the fault stays away, the suspect is confirmed alone. Then turn the suspect back on by itself, with everything else off. If the fault appears, it is broken in its own right. If it does not, you have found a conflict rather than a broken plugin, and you now want to know what it is fighting with.
Finding the second party is the same halving procedure, run with the suspect permanently on.
What to do with the answer
A confirmed culprit gives you four options, roughly in order of preference.
Update it. Check whether a newer version exists. A large share of conflicts are already fixed upstream by the time anyone notices them.
Check what it says about compatibility. A plugin's "Tested up to" field is developer-declared and nobody verifies it, which is worth understanding before you rely on it; our article on checking plugin compatibility before you update covers how to read those signals honestly.
Report it. The plugin's support forum is a legitimate route and a well-described conflict, naming both plugins and both versions, tends to get answered. You now have exactly that description.
Replace it or live without it. Sometimes the honest answer is that a plugin has been abandoned and the feature is not worth the breakage.
What you should not do is delete the plugin as a reflex. Deleting can remove its data as well as its code, and that is a different and much less reversible action than deactivating.
Two habits that make the next one faster
Write down the result. Date, symptom, plugins involved, what fixed it. In a year, when it happens again, that note is worth more than the whole test.
Reduce the surface. Every plugin you keep is another party to a possible conflict and another thing that can break at the next core release. A plugin you have not consciously used in a year is a maintenance cost with no return.
When to stop testing
If the fault involves the database, if the site is fully down and you cannot reach the admin, or if the test itself is causing visible problems for real visitors, stop. Restore the list you wrote down at the start, put the site back to its known state, and get help with a clear description rather than a half-finished experiment.
A person who arrives to help you with "the fault is reproducible with plugins A and D active, and not with either alone" is starting from a good place. A person who arrives to find fifteen plugins off and no record of which fifteen is starting from a worse one than you began with.
Version context. Written against WordPress 7.1, released 2026-08-19, the current major release on our record. Bulk deactivation and reactivation from the Plugins screen is long-standing core behavior; the screen itself was restyled in the 7.x admin but the actions are unchanged. Release dates come from our own dated record of the WordPress.org release page, last read 2026-08-19.
