How Fast Should a Website Load?

Every owner has had this moment. Someone at a barbecue mentions that the website took forever to load, and the owner pulls it up on the office Wi-Fi the next morning to prove them wrong. It loads at once. The person at the barbecue was on a phone in a backyard with two bars, and was right.

How fast should a website load? It depends which of those two people you ask, and the second one pays you. So the target is set for the phone in the backyard, not the desk in the office.

Here is the number. The main content of a page should be visible within about two seconds on a phone over an ordinary cellular connection, and under one second on a desktop. Three seconds on a phone is losing people. Five is losing a lot of them. The targets line up with how long people wait before deciding something is broken, and with the thresholds Google uses to grade sites. The why is my website slow guide covers what is usually in the way. This one covers the target and what missing it costs.

"Loads" is a vague word, and the industry uses the vagueness to sell scores. The browser finishing, the headline becoming readable and the button answering a tap are three different moments, and a visitor cares only about the last two. Here are the targets, stated as what the visitor experiences.

The second row matters most. A visitor decides whether to stay when the headline and main image are readable, not when the last tracking pixel finishes, which is why Google built its main measure around the largest thing on the screen. Hit two seconds there on a phone and most of the rest follows.

The cost of a slow second is easy to overstate with a statistic and easy to understate with a shrug. Here is the reasoning instead.

A visitor from a search has other options one tap away; the back button is a list of your competitors. Patience depends on need and trust, and a stranger from a search has no trust yet. So the visitors who leave first are not the casual browsers. They are the people with the cracked windshield, the flooded kitchen, the toothache: the ones most ready to call and least willing to wait.

Google has published research on this for years, and the shape is always the same: the share of visitors who give up rises sharply between one and three seconds and keeps climbing. As a rough rule, each added second on a phone costs around a tenth of the people who would otherwise have stayed, and the losses compound. A page that goes from two seconds to five can lose a third of its visitors before it paints.

Put it in your own numbers. Say 1,000 people a month reach a service page from search and 5 percent call. That is 50 calls. If the page takes five seconds instead of two and a third of the visitors leave before it paints, that is roughly 17 calls gone every month, from people who had already chosen you from the results. At even $500 a job, the slow page costs more each quarter than fixing it costs once. The dollar value of a fast website runs the math with more care; the conversion guide covers what to do with the visitors you keep.

Does website speed affect Google rankings? Yes, directly and indirectly, and the indirect part is larger.

Directly: Google has used page speed as a ranking signal on desktop since 2010 and on mobile since 2018, and since 2021 the Core Web Vitals have been part of its page experience signals. The effect is real and modest. Speed is a tiebreaker between pages with comparable content; a fast thin page does not beat a slow thorough one. But ties are common in local search, where twenty plumbers have twenty similar service pages, and the tiebreaker decides who is third and who is eleventh.

Indirectly: a slow site is crawled less, because Google budgets crawling by how quickly a site responds. Visitors who bounce back to the results signal that the result did not satisfy. And a slow page converts fewer of the visitors it gets, so whatever the ranking, it produces less. The does website design affect SEO guide places speed among the other design decisions that touch ranking; the SEO basics guide covers the content side that speed only supports.

The practical reading: get into the green on the three Core Web Vitals so speed stops counting against you, then put the effort into content and links. A score of 95 does not rank better than 90. A score of 40 does rank worse.

Run a speed test and you get two sets of numbers, and they often disagree. The top set, labeled something like "what real users are experiencing," is field data: measurements collected from actual visitors on their actual phones and connections. The bottom set is lab data: one simulated load on a throttled connection from a server somewhere.

They measure different things. Lab data is one visit under fixed conditions, useful for diagnosing what is slow. Field data is thousands of visits under real conditions, which is what customers experience and what Google ranks on. A site can score 90 in the lab and be red in the field, because the lab ran on a fast simulated phone and your visitors are on three-year-old Androids in parking lots. The reverse happens too: a middling lab score with green field data, because most visitors are on good connections with the page cached.

When they disagree, believe the field data. It is the customer. Use the lab data to find out why. A small site with too little traffic for field data has to be judged by the lab, and the correction is to assume the real experience is worse than the lab shows.

Most owners test wrong, which is why so many believe their site is fine. Here is how to test in a way that matches what customers see.

Over three seconds on a phone, and the fix is usually cheaper than owners expect; the free site audit lists the causes in order. Under two, and the page is fast enough: stop worrying about speed and read the conversion guide, because the leaks are elsewhere. Either way, conversion tracking is how you see the change in calls, and without it the improvement is a score rather than a result.