This is Part 2 of a 3-part series on website performance for business owners. Part 1 covers the business case for speed. Part 3 covers the fixes that usually matter most.
Part 1 established why performance matters. The next useful question is where the delay is entering the pages that matter most, and whether the bottleneck is media, scripts, backend response, or infrastructure.
The Pages That Matter Most
An abstract sitewide score usually says less than a page-level read of the URLs tied to leads, sales, or support load. The pages that usually deserve first attention are:
- Your homepage
- Primary service or product pages
- Landing pages tied to paid campaigns
- Contact, quote, booking, or checkout flows
If those pages are slow, the business impact is immediate. If they are healthy, lower-priority pages can usually wait.
Field Data and Lab Data Tell Different Stories
Performance tools usually show two kinds of information.
Field data reflects what real visitors experienced over time. It is useful because it captures actual devices, networks, and browsing conditions.
Lab data comes from a controlled test run. It is useful because it gives you diagnostic detail right away, even if your site does not have enough traffic to produce strong field data.
PageSpeed Insights combines both when data is available. Search Console is useful for spotting patterns across groups of pages. Chrome DevTools is useful when you need developer-level detail.
What Core Web Vitals Are Telling You
Core Web Vitals focus on three different parts of the user experience.
Largest Contentful Paint
Largest Contentful Paint, or LCP, is about when the main visible content appears ready. If LCP is weak, common causes include slow server response, oversized hero images, render-blocking CSS or JavaScript, and assets loading in the wrong order.
Interaction to Next Paint
Interaction to Next Paint, or INP, is about responsiveness after the page is visible. If INP is weak, the problem is often too much JavaScript running on the main thread, heavy third-party tools, or interface code that tries to do too much work before letting the browser respond.
Cumulative Layout Shift
Cumulative Layout Shift, or CLS, is about visual stability. If content jumps while the page loads, the issue is often missing image dimensions, late-loading banners, unstable embeds, or font loading that changes the layout after the page is already visible.
Read the Audit as a Bottleneck Report
A PageSpeed report is most useful when you read it as a map of where delay is entering the page load.
If the server is slow before anything renders, the bottleneck is probably hosting, caching, or backend work.
If the page waits on large images, the bottleneck is media handling.
If the browser is busy parsing scripts before it can paint or respond, the bottleneck is JavaScript, third-party tags, or front-end architecture.
If the page looks mostly loaded but still feels erratic, the problem may be layout stability rather than raw load time.
The score matters less than the pattern. Different causes can produce similar overall scores while requiring completely different fixes.
Match the Symptom to the Likely Cause
Some common pairings:
- Slow initial render often points to server response, caching, or large above-the-fold assets
- Sluggish clicks and taps often point to heavy JavaScript or third-party scripts
- Shifting content often points to missing reserved space for images, embeds, or banners
- A fast desktop result with a poor mobile result often points to overly heavy pages that desktop hardware can hide
This is why generic checklists only help so much. A site with weak hosting and a site with overloaded third-party scripts can both feel slow for entirely different reasons.
Prioritize by Impact, Not by How Easy a Tool Makes It Sound
Diagnostic tools list many opportunities. Not all of them deserve the same attention. The highest-value priorities are usually the items most likely to affect the pages that matter most:
- Fix the largest asset or response bottleneck first
- Remove or defer code that blocks rendering or interaction
- Address layout shifts on pages with forms, buttons, or calls to action
- Re-test after each meaningful change instead of batching everything together
Performance work goes better when you isolate cause and effect. Change something material, measure again, and confirm whether the field data trend is improving.
Part 3 covers the fixes that usually move the needle first, and where the line is between reasonable in-house cleanup and work that needs deeper technical access.
Previous: Part 1: Why Speed Is a Business Problem Next: Part 3: Fixes That Move the Needle
Related reading: Website Performance for Business Owners, Part 1: Why Speed Is a Business Problem, Website Performance for Business Owners, Part 3: Fixes That Move the Needle, and What a Website Audit Actually Finds and Why It Matters.