How to reduce page weight without guessing
Start with the categories and individual resources that actually dominate the download. Change one cause at a time, then measure under comparable conditions.
Use the receipt to choose where to investigate
Print a receipt for the page you want to improve. Review its HTML, CSS, JavaScript, images and fonts, then open the analysis panel to inspect the largest completed resources and contacted domains. A large category is a starting point for investigation, not proof that every byte in it is unnecessary.
Set a page-weight budget that reflects the purpose of that page and your users’ constraints. A budget is a chosen limit you can monitor; WebReceipt does not claim there is one correct size for every website. Include the scan profile in the baseline so later numbers remain interpretable.
When images dominate
Check whether source dimensions are much larger than the displayed image. Serve appropriately sized variants and compare modern encodings against the original for both appearance and file size. Decorative assets and repeated images deserve particular attention.
Lazy loading below-the-fold images can reduce the initial download, but it may move bytes into a later part of the visit. Check the complete experience separately. Keep the first meaningful image available promptly, and do not call an initial-scan reduction proof that the entire journey transfers less data.
When JavaScript or fonts dominate
Review which scripts are required for the page’s main task. Remove unused dependencies where possible and defer features that are not needed immediately. Third-party widgets can add both requests and JavaScript; evaluate their usefulness against the measured cost.
For fonts, review the number of families, weights and language subsets you deliver. Avoid fetching variants the page never uses. Test the text experience as well as byte savings: reducing downloads should not make content unreadable or break an interaction.
Measure the change fairly
Repeat the same URL under the same profile and record the dates. WebReceipt normally reuses results under six hours old; an unchanged receipt is not evidence that a deployment failed. Follow the refresh rules or measure a deliberately separate public test URL when appropriate.
Compare resource categories and completed transfers, not just the total. Check failed or unfinished requests and measurement warnings before celebrating a reduction. A smaller receipt caused by missing content is different from a successful optimization. Keep Core Web Vitals and real-user experience as separate checks.