How slow is too slow on a phone? · Shoebill Design Studio

Two and a half seconds. That’s the published threshold for a good Largest Contentful Paint, which is the moment the biggest thing on the screen actually appears. Google uses it as a page experience signal, but the reason it exists is human: that’s roughly where people start to disengage.

Most local business sites we measure are nowhere near it on a phone. Not slightly over. Multiples over.

What the number actually measures

Largest Contentful Paint is the time until the main thing on the page is visible. Usually that’s the hero image or the headline. It’s a better measure than total load time because it tracks the moment the page becomes useful rather than the moment the last tracking pixel finishes.

Two other numbers sit beside it. Interaction to Next Paint tracks how quickly the page responds when somebody taps, and Cumulative Layout Shift tracks how much the content jumps around while it’s loading. All three are published thresholds and all three are measurable from your own phone in a couple of minutes.

The one that costs you customers is the first one, because a page nobody can see yet is a page nobody is reading.

What we found on a real site

On the Stone Care of Arizona site, before the rebuild, mobile pages were loading their largest image somewhere between 15 and 27 seconds depending on the page.

Nobody had done anything obviously wrong. The images were the sizes they came off the camera. The site carried the usual accumulated third-party scripts. On the owner’s laptop on office wifi it felt fine, which is exactly why it had gone unnoticed for years.

For comparison, the site we built for our own studio measures a Largest Contentful Paint of 0.8 seconds on a desktop test with a 100 performance score, and it carries a real photograph in the first screen. Fast isn’t a trade against having images. It’s a trade against having them at the wrong size.

The three usual causes, in order

Images at their original dimensions. A 4,000-pixel-wide photo displayed 400 pixels wide on a phone still downloads all 4,000 pixels unless somebody set up smaller versions. This is the single most common cause and the easiest to fix.
Third-party scripts. Chat widgets, review widgets, tag managers, font services, analytics, each from a different domain, each needing its own connection. They’re rarely individually terrible and they add up fast.
The hosting and the platform. Real, and usually smaller than people expect. Moving hosts won’t rescue a page that’s downloading eight megabytes of photography.

The order matters, because effort spent on the third one before the first two is effort wasted.

What you can check in the next thirty seconds

Two ways, both quick.

The honest one: put your phone on cellular data, turn off wifi, and open your homepage as if you’d never seen it. Count out loud until you can read the headline. That count is what your customers get, and it’s more persuasive than any report.

The precise one: run your homepage through Google’s PageSpeed Insights. Read the mobile tab, not desktop, and look at the Largest Contentful Paint figure. Under 2.5 seconds is good. Between 2.5 and 4 needs work. Over 4 is losing you people who intended to hire you.

Do it on a service page too, not just the homepage. Service pages are often heavier and they’re where the buying happens.

Why this shows up in the lead count and not in the analytics

A visitor who leaves before the page paints often doesn’t register as a meaningful session. They’re a blip. So the analytics look normal, the traffic looks normal, and the only thing that’s wrong is that the phone doesn’t ring as much as the numbers suggest it should.

That’s the failure mode: it isn’t visible in the place you’d look for it. It’s visible in the gap between how much traffic you have and how many enquiries you get, and that gap has ten possible explanations, so it rarely gets attributed to speed.

When to bring somebody in

Image sizing is the first job and it’s the biggest single lever. On most sites it’s a contained piece of work: generate the right sizes, serve the right one to the right screen, and stop shipping desktop photography to phones.

Auditing the third-party scripts is the second, and it needs somebody willing to say out loud that a widget isn’t earning its place. That’s a conversation, not a technical problem.

Where a rebuild becomes the answer is when the platform itself is the limit, or when the page has to be restructured anyway for other reasons. That’s part of how we build websites, and the test we hold ourselves to is the same one above: open it on a phone, on cellular, and count.

Common questions

My site feels fast to me. Why would it be slow for customers?

You are testing on office wifi, on a device that has already cached the site, probably on a desktop. Your customers are on cellular data, on a phone, arriving for the first time with nothing cached. Those are different experiences and only one of them is the one that matters.

Is 2.5 seconds a Google rule or a customer rule?

Both, and the customer one came first. Google publishes 2.5 seconds as the threshold for a good Largest Contentful Paint because that is roughly where people start to disengage. It is also a ranking input, but the reason to care is that visitors leave.

Will a faster host fix it?

Rarely on its own. Hosting matters, and it is usually not the biggest number in the total. Uncompressed images and third-party scripts are, in that order, and neither gets faster on a better server.

How many third-party scripts is too many?

There is no number. The question is what each one costs and what it returns. A chat widget that produces real conversations may be worth 200 kilobytes. A widget nobody has used in a year is not worth anything, and it still loads on every page.

Does this matter if we get most of our work by referral?

A referral still looks you up before calling. That is often the only visit they make, and it is the one where a slow or half-loaded page does the most damage, because they arrived already inclined to hire you.