Quick answer
If a landing page is fast without parameters but slow from a campaign link, compare the exact query strings. WP Rocket treats documented tracking parameters differently from arbitrary parameters. A mixed URL containing an unrecognized key can bypass its normal cache behavior. Do not remove all tracking values or enable caching for every query just to improve a speed score.
Build three controlled links
Choose one public landing page with no account-specific content. Create a clean URL, a URL containing only a documented UTM parameter, and a third URL adding the suspicious parameter. Keep the browser, consent choice, region, and logged-out state unchanged. Use fictional values and a staging analytics destination where possible.
For example, compare /offer/, /offer/?utm_source=diagnostic, and /offer/?utm_source=diagnostic&ref=sample. The final ref key is only an example. Your actual extra key might come from an email system, affiliate plugin, personalization tool, or redirect service. Record every redirect because the final URL may differ from the original link.
Record the response, not just the stopwatch
Open browser developer tools and inspect the document request. Record the final URL, status, response timing, cache headers provided by your host or CDN, and whether the response came from browser storage. Repeat each URL to separate an initial request from a warmed request. Do not assume a response header from the CDN describes the WordPress page cache.
Compare a stable public text marker in the returned HTML. A fast response that contains the wrong campaign offer is a failure. Equally, an unchanged HTML document can still allow browser-side attribution to read the current address. Sharing a public HTML cache does not by itself prove that attribution was lost.
Classify the additional parameter
Ask the integration owner what the key changes. A parameter used only to label a visit is different from one selecting currency, language, pricing, a customer record, or a signed access grant. Cache policy must follow response behavior, not the parameter name alone.
WP Rocket documents standard-cache exceptions and dedicated query caching. Adding a parameter to dedicated caching can create separate variants for its values. Before considering that option, estimate the number of values your campaigns create. A unique value per recipient can create an impractical cache footprint and may expose personal content if the design is wrong.
Use a small evidence worksheet
Record one row per URL and repeat. The expected result should be written before changing settings. That makes it possible to reject an apparently faster configuration when it breaks the campaign or displays another visitor's content.
variant | final URL | document cache | public text | attribution
clean | record | record | baseline | none expected
utm | record | record | compare | diagnostic
mixed | record | record | compare | diagnostic
Change only the layer responsible
If WP Rocket eligibility explains the difference, review the specific parameter against the vendor guidance. If the CDN treats the variants differently, inspect its existing rule separately. Do not add a broad Cache Everything rule to a commerce site as a shortcut. Cart, checkout, authenticated, and personalized responses need their existing protections.
After a narrowly reviewed change, clear only the relevant cache layers through their supported controls. Repeat the same three requests. Submit a fictional lead through the intended form and check the resulting test record. A page-speed improvement is not a successful fix when the sales team loses the source of that lead.
Close the test
Document which key affected cache behavior, which component owned the rule, and the safe rollback. Remove diagnostic records from operational reporting according to your test-data process. Keep an unmodified campaign link in the final verification set so the production URL is tested exactly as recipients will receive it.
Illustrative diagnostic example. Use your own redacted evidence.One final acceptance check
Include the original email or ad redirect in the last test, not only a manually typed final URL. The redirect can introduce a parameter or remove one. Check the destination form record after that exact navigation so the attribution comparison covers the complete landing path.
When to bring in help
Use WordPress campaign-page performance support when marketing, the CDN, and the cache plugin disagree about URL handling. Bring redacted URL examples and the corresponding cache evidence, not customer identifiers.
Related troubleshooting
For the adjacent diagnostic path, read WP Rocket preload and scheduler diagnostics. Keep its evidence separate from this test so a change in one component does not hide a failure in another.
Helpful references
WP Rocket query-string caching.