A filter reacts late even after the page has loaded. INP (Interaction to Next Paint) describes the delay between an input and the next visible update. Inspect the work the browser performs around the click.
Choose a concrete action, such as changing a variant or opening a filter, and record it in performance tools. Inspect long tasks against the timing of the input. Third-party scripts are possible contributors, not a proven cause without evidence.
Build a controlled comparison
On a protected test environment, create a version without the suspected provider. Change as little else as possible. Repeat the same action under comparable conditions and record both responsiveness and visible results.
Fictional investigation record:
Action: open colour filter
Suspect: external review script
Comparison: same page without that script
Inspect: main-thread tasks and visible response
Check functionality too. A faster page is not a valid correction if essential controls stop working. Where evidence points to a script, assess later initialisation, a smaller integration or removal of unnecessary features.
Separate individual laboratory recordings from field measurements. One recorded test click does not represent INP across all real visits. Document the interaction and remaining limitations so the optimisation can be checked reproducibly later.