This guide explains how to diagnose a slow website without becoming a developer. You will learn which metrics matter, how to find the real bottleneck, and which fixes return the most value first.
why speed hits revenue before SEO rank moves
Google has used page experience signals for years, but the immediate pain is conversion. Research commonly cited in retail and CRO circles: each extra second of load time can reduce conversion by up to 7%. On mobile, more than half of visitors leave when load exceeds 3 seconds on commerce journeys.
That means a homepage moving from 2 seconds to 5 seconds on mobile can lose a double-digit share of orders during peak campaigns, even if rankings stay stable for weeks.
hat widgets, tag managers, and page builders often cause high TBT even when the page looks loaded.
start with PageSpeed Insights on real URLs
Run Google PageSpeed Insights on your homepage, top category page, and product or lead page. Lab data shows controlled tests. Field data (CrUX) shows what real Chrome users experienced in the last 28 days. Trust field data when available.
Record four numbers: Performance score (context only), Largest Contentful Paint (LCP), Interaction to Next Paint (INP) or Total Blocking Time (TBT) in lab tests, and Cumulative Layout Shift (CLS). One bad metric on checkout matters more than a perfect blog post.
what LCP, FCP, and TBT tell you
LCP (Largest Contentful Paint): how long until the main content appears. High LCP usually means slow server response, huge hero images, or render-blocking CSS and fonts.
FCP (First Contentful Paint): when anything first paints. A good FCP with bad LCP suggests the hero asset or slider is the culprit.
TBT / INP: how much JavaScript blocks interaction. Chat widgets, tag managers, and page builders often cause high TBT even when the page looks loaded.
CLS: layout shift score. Banners, late-loading fonts, and embeds pushing content cause mis-taps on mobile checkout buttons.
Use this matrix to decide where to invest first.
| Symptom | Likely layer | First fix |
|---|---|---|
| Slow first byte | Hosting, database, cache | Object cache, query audit, PHP workers |
| Huge hero image | Front-end assets | Compress, responsive srcset, modern formats |
| Page interactive late | JavaScript | Defer non-critical scripts, audit tag manager |
| Layout jumps on mobile | CSS and embeds | Reserve space for ads, fix web font loading |
| Only checkout slow | Plugins and gateway | Profile checkout AJAX and payment scripts |
Speed hurting conversions?
Talk to us →find the bottleneck in three tests
Speed hurting conversions?
Talk to us →Test 1: Load a plain HTML file on the same host if possible. If that is also slow, the problem is infrastructure.
Test 2: Load the site with a default theme and plugins disabled on staging. If speed jumps, plugin or theme code is the limiter.
Test 3: Compare logged-in vs logged-out and mobile vs desktop. Admin toolbars and personalization scripts often explain gaps between lab scores and complaints.
prioritize fixes by business impact
Fix templates by revenue share, not by vanity scores. If 60% of orders start on three product URLs, optimize those before the about page.
- Compress and lazy-load images on top templates.
- Remove or delay third-party scripts not tied to revenue.
- Enable server-level and page caching with a clear exclusion list for cart and checkout.
- Audit database queries on slow WooCommerce or archive pages.
- Only then consider theme rebuilds or platform migration.
One retail client cut mobile LCP from 4.8s to 2.2s by fixing hero images and deferring a chat widget. Conversion on paid traffic rose 11% without changing ads.
when DIY optimization is not enough
If TTFB stays above 600ms on a warmed cache, field LCP fails on money pages after two optimization sprints, or checkout INP fails during campaigns, you need a structured audit across hosting, theme, and plugin stack.
A performance audit maps bottlenecks to cost and timeline. Ongoing maintenance prevents regressions when plugins update weekly.
final takeaway
Measure on the pages that earn money, fix the worst metric first, and retest after every change. Speed work is iterative, not a one-time plugin install.
Slow sites rarely have one magic fix. They have a short list of bottlenecks ranked by impact. Start there.
