What a Slow Website Actually Costs You

Website speed has an image problem.

It sounds like flossing. Technically virtuous, vaguely tedious, and easy to postpone forever because nothing visibly bad happens on any given day. Meanwhile the leaking is continuous and completely invisible, because the people you’re losing never contact you to explain why.

That’s the whole difficulty. A slow website doesn’t generate complaints. It generates absence.

So let’s do the actual math, and let’s be honest about what the evidence supports.

The mechanism is simpler than the metrics

Strip away the acronyms and the causal chain is short:

Your page takes a while. Some fraction of visitors leave before it finishes. Those people were going to be a normal-sized slice of customers, and now they aren’t customers, because they never saw your website.

That’s it. That’s the entire theory. Everything else is measurement.

The reason this feels abstract is that you never meet these people. Nobody calls to say “I was going to book you but your site took four seconds so I hired the other guy.” They simply hit back and forget you existed. Your analytics logs it as a bounce, which is a bloodless word for a lost sale.

What the published evidence actually shows

Google maintains a collection of case studies from companies that measured this. The consistency is the interesting part:

• Lazada improved LCP threefold and reported a 16.9% increase in mobile conversion rate.
• Cdiscount reported a 6% revenue uplift during its Black Friday sale.
• Nykaa improved LCP by 40% and reported 28% more organic traffic.
• Yahoo! Japan fixed layout shift and reported a 15% uplift in page views per session.
• AliExpress improved CLS tenfold and LCP twofold and reported 15% lower bounce rates.
• Vodafone improved LCP by 31% and reported 8% more sales.

Now the caveats, because they matter and most articles quoting these numbers skip them.

These are self-reported by the companies involved. They are all very large sites, mostly e-commerce, mostly in markets with slower average connections than North America. And they are, by definition, the experiments that worked. Nobody publishes a case study titled “we made our site 40% faster and nothing happened.”

So do not take 8% to the bank. What you should take is the direction and the consistency. Across different industries, different countries, and different starting points, faster consistently correlated with better. No published case study found the reverse.

Working out your own number

Forget the case studies. Your own arithmetic is more useful and takes about two minutes.

You need three figures:

1. Roughly how many people land on your site in a month.

2. What fraction of them become a lead or a sale.

3. What a customer is worth to you.

A pest control company getting 800 visits a month, converting 3% into calls, closing half of those, at $450 a year per customer, is running about $65,000 a year through that website.

Now assume speed work recovers a modest slice of the people currently leaving before the page renders. Not 16.9%. Call it 5%. That’s roughly $3,200 a year, recovered permanently, from a fix you do once.

Run your own numbers. The point isn’t the specific answer, it’s that for most service businesses this lands somewhere between “worth a weekend” and “why did nobody tell me sooner.”

And notice which way the leverage runs. Speed work compounds with everything else. Every dollar you spend on ads, SEO, or referrals sends traffic through the same funnel. Widening the top of that funnel makes every other marketing dollar you spend worth slightly more, forever.

The cost you cannot see at all

Here’s the part that doesn’t show up in any dashboard.

Speed shapes what people believe about you before they read a word. A site that appears instantly feels like a company that has its act together. A site that hesitates, then shudders as the layout rearranges itself, feels like a company that might also be slow to return your call.

That judgment happens in the first second, it happens below conscious thought, and it happens to every single visitor including the ones who stay. You cannot measure it and you cannot argue with it. For a home services business asking a stranger to let them into their house, that impression is not a small thing.

Why the fix so often fails

Plenty of people do try to fix this, get nowhere, and conclude the whole topic is nonsense. Usually one of three things happened.

They optimized the wrong layer. Google’s own data shows about 40% of typical LCP time is server response. If your host is slow, you can compress every image on the site and still be slow. You were tuning an engine with the handbrake on.

They tested on themselves. You visit your own site constantly, from the same device, with everything cached. It’s fast for you and only for you. Google grades at the 75th percentile of real visitors, which is a much less flattering audience.

They fixed it once. Speed is not a state, it’s a slope. Every new plugin, tracking pixel, chat widget, and uncompressed photo your team uploads costs a little. Sites don’t become slow through one bad decision, they become slow through eighteen months of reasonable ones.

What good looks like

Google publishes the targets, so there’s no guesswork. Largest Contentful Paint under 2.5 seconds, Interaction to Next Paint under 200 milliseconds, Cumulative Layout Shift under 0.1, all measured at the 75th percentile, on mobile.

Mobile specifically, because Google uses the mobile version of your site for indexing and ranking. Your desktop score is a vanity metric.

If you want to know where you stand right now, run your homepage through PageSpeed Insights and read the real-user section at the top. It takes half a minute and it’s free.

If the numbers are bad, how to make your website load faster is the practical fix list, and Core Web Vitals explained covers what each metric is actually measuring.

If they’re bad and you’d rather just have it dealt with, speed optimization is built into every site we launch. It has never made sense to me to hand someone a slow website and then sell them the cure.