Somebody taps your ad on a phone, on mobile data, on a Tuesday, halfway through doing something else. That is the real test conditions. Not your desktop, on fibre, with the site already cached.
Under those conditions, a three-second load isn’t a three-second load. It’s a decision.
Speed is a brand signal before it’s a metric
People don’t think “this site has a poor Largest Contentful Paint”. They think this feels cheap, and they leave without ever forming the sentence. The judgment happens below the level where they could explain it to you.
Which means every millisecond you claw back is doing two jobs: it’s keeping the session alive, and it’s telling the visitor that the business behind the site is competent. Those are the same job.
The four things that are actually slowing you down
After enough audits, the culprits stop varying much.
1. Images nobody compressed. A 2.4 MB hero photo is not a design decision, it’s an oversight. Modern formats, correct dimensions, lazy loading below the fold. This alone fixes most sites.
2. Fonts loading from someone else’s server. Every third-party font request is a DNS lookup, a handshake and a render block before a single word appears. Self-host them. Preload the two you actually use.
3. A page builder rendering forty nested divs to draw a button. Convenience at build time, tax on every visit forever.
4. Tag manager sprawl. Nine tracking scripts, six of which fire for a tool nobody in the business has logged into since last year. Audit them. Delete ruthlessly.
What “fast enough” means
Forget the vanity score for a second. Three numbers matter:
- Largest Contentful Paint under 2.5 seconds — the main thing on screen, visible.
- Interaction to Next Paint under 200ms — taps respond immediately.
- Cumulative Layout Shift under 0.1 — nothing jumps under the thumb mid-tap.
Test on a throttled 4G connection with a mid-range device profile, because that’s your actual audience. Testing on your own machine is how sites ship broken.
The part everyone skips
Speed gets you the visitor’s attention. It doesn’t tell them what to do with it.
I’ve seen sites hit every Core Web Vitals threshold and still convert at under one percent, because the fastest-loading page in the world was a hero image of a handshake and the headline “Solutions for a changing world.”
Fast and vague is still vague. It just fails more efficiently.
So fix the load time — then use the attention you bought. Say who it’s for. Say what changes. Put the next step within reach of the thumb.
The site is only the first promise. Everything after it is the one that counts.
