Measure multiple page types
Check the homepage, a major category, a product with options, search and cart. Compare first and repeat requests, and guest and signed-in customers. Record the same test conditions: device, network, time, and cache state. One quick revisit doesn't describe the new visitor experience.
Continue with: E-commerce caching: keep customer data private.
Find the expensive component
If the server is taking a long time to respond, investigate slow SQL queries, external APIs, modules, and PHP queues. If the HTML is coming fast, look at the size of the main image, blocking resources, and long JavaScript tasks. Disable the suspicious extension only on the copy, comparing the same scenario before and after the change.
Continue with: Core Web Vitals: understand LCP, INP and CLS.
Verify functionality after optimization
Catalog cache can help, but cart, personal data, and dynamic pricing need separate rules. After optimization, check two different buyers, stock changes and coupons. The report should contain the measured effect and limitations, not just the score of one service. Make a decision about the server after analyzing the load and available resources.
Do you need perfect PageSpeed scores?
The score helps you find problems, but is not a substitute for real-world buying scenarios, field metrics, and testing store stability under load.