Website Load Time: Why It Matters and What Improves It

Why load time matters for a business

A prospect searches for a service on the go, taps the result and waits. The image at the top builds up slowly, the text jumps, a cookie notice slides over the copy. After a few seconds they tap back and choose the next result. The business never finds out. Its analytics show only a visit that ended after a short time, or not even that.

This piece asks: Why does load time matter for a business’s inquiries, how is it measured, and which measures improve it with reasonable effort? To answer it, we first look at a model from research on waiting times and the terms Google uses to measure load time. Then we weigh how important speed is compared with content, and go through the measures by impact and effort.

The fastest page is not the one with the best technology, but the one that only loads what the visitor needs.

The thinking model: three time limits of attention

Research on computer response times described three limits early on that still serve as a guideline today. Fiona Nah (2004) summarizes them in her review, following Jakob Nielsen: up to about 0.1 seconds, a user perceives the response as instant. Up to about 1 second, their flow of thought remains intact, even if they notice the delay. Up to about 10 seconds, their attention stays on the task; after that they turn to something else. As early as 1968, Robert Miller had concluded from the limits of short-term memory that waits of about two seconds or more disrupt processing.

Nah tested in an experiment how long users actually wait. 70 students were asked to look up information through a series of links, some of which never loaded. The researchers measured when they gave up waiting. One group of 34 participants saw a moving progress bar, the other group of 36 did not. At the first link that did not load, participants without the bar waited 13 seconds on average, those with the bar 38 seconds. Feedback therefore extends patience considerably. For information retrieval overall, Nah arrives at a tolerable waiting time of about two seconds.

The study has limits that need to be known. It dates from a time of slow connections, the participants were students in a university lab, and the pages either loaded instantly or never. Today’s users are used to faster pages, so their patience is likely to be shorter. More recent studies confirm the direction, however. Lee, Chen and Ilie (2012) showed that long waits online worsen the evaluation of a website and its provider and lead to abandonment, but that well-designed fillers can shorten the perceived waiting time.

Terms: how Google measures load time

Google measures the user experience during loading with three metrics, the Core Web Vitals. They are based on measurements of real visits with the Chrome browser. A page counts as good if all three values are in the green range for at least three quarters of visits. Technically, this means: at the 75th percentile.

The article A website that generates inquiries sets the benchmark that the visible part of the page is in place on a smartphone in under three seconds. Google’s threshold of 2.5 seconds for the largest content is somewhat stricter and refers to three quarters of all visits, including slow connections. The two values do not contradict each other. The benchmark is a rule of thumb for your own test; Google’s threshold is the yardstick for measurement on real visitors.

How important is speed compared with content? Two views

The first view says speed is the foundation of everything. A slow page loses visitors before the content can take effect, and fast pages are rated better. The studies on waiting times support this. The second view objects: in business-to-business markets, visitors search with a concrete concern. Someone looking for a solution to an expensive problem will wait a second longer if the page promises the better answer. Smith and colleagues (2019) found in two experiments that participants accessed content with long load times less often and chose faster content instead. Those who have a choice therefore avoid waiting.

I consider both views justified, with a clear order of priority. Speed is a hygiene factor in the sense of the model described in the article Planning a website relaunch: if it is right, nobody notices. If it is not, it devalues the best content. A page therefore does not have to be the fastest in its industry, but it must not be slow. Anyone in the green range of the Core Web Vitals gains little from further tenths of a second. Anyone far above it loses a lot.

Measuring load time: three tools, one routine

For a small business, three free ways to check load time are enough. What matters is using them in the right order.

1. Search Console, Core Web Vitals report

   Shows which groups of pages perform well, need improvement or perform poorly for real visitors. This is the starting point because it reflects real visits. How Search Console is structured is described in the article Understanding Google Search Console.

2. PageSpeed Insights for individual pages

   Shows the values from real visits for a URL, if there is enough data, and a lab test with concrete hints as to which files are slowing things down.

3. Your own test on a smartphone

   Open the most important service page over the mobile network, not over Wi-Fi. Time when the headline and image appear and when the button responds.

4. Record the result

   Note the values per page and date. Only then can you see after a change whether it worked.

What improves load time: measures by impact and effort

For most small business websites, the cause lies not in the server technology but in what sits on the page. The following worked example shows what the weight of a typical service page is made up of.

The order of the table is a recommendation: first the measures with high impact and low effort. Many of them also help with data protection, because every third-party service that does not load transmits no data either. The article A website that generates inquiries therefore recommends hosting fonts and images locally. A rebuild only pays off once the simple measures are exhausted and the technology itself is the brake.

Who is responsible for load time

Load time rarely deteriorates all at once, but gradually. A new hero image at full camera resolution, an additional service for reviews, a video on the homepage. Each change is harmless on its own; together they make the page slow. That is why load time needs someone responsible for it.

• Rules for content. Whoever uploads images knows the maximum size. Videos are not embedded directly.
• New services only after review. Every new third-party service is checked for benefit, data protection and load time before it is added.
• Quarterly measurement. A look at the Core Web Vitals report and a test of the most important page.
• A clear agreement with the agency. Target values for the Core Web Vitals belong in the contract, not just a nice design. Which questions to ask beforehand is described in the article Choosing an agency.

Anyone who does not want to look after website, load time and findability themselves can hand this over to a service provider, for example quik-services, a business partner of Owners Council that offers websites, SEO and lead generation. In any case, the business should be able to see the measurements itself.

What can be generalized

Load time is a good example of a quality you only notice when it is missing. No customer praises a fast website, but many leave a slow one. This leads to a sober order of priority for work on the website: first it has to be fast enough for the content to be seen, then the content has to convince. How to measure whether it convinces is described in the article Measuring the website’s inquiry rate.

The limits are clear. Some of the studies on waiting times are older and based on lab experiments with students. Google’s thresholds are conventions, not laws of nature, and Google weighs them in ranking only as one of many signals. How many inquiries a faster page brings depends on industry and starting point and cannot be predicted in general.

Conclusion and a first step

The question at the start was
load time matters for a business’s inquiries, how it is measured and what improves it. The answer: it matters because, according to the three time limits of attention, visitors lose interest quickly and switch to faster offers. It is measured with the Core Web Vitals on real visits, with 2.5 seconds for the largest content as the threshold for good. It is improved mainly through smaller images, fewer third-party services and videos that are not embedded directly, not primarily through new technology.

• Open the Core Web Vitals report in Search Console and note which pages are red or yellow.
• Check the most important service page in PageSpeed Insights and note the three biggest brakes.
• Open the page on your own smartphone over the mobile network and time it.
• Re-upload the largest images at the right size and in a modern format.
• List all third-party services and remove those that are not needed.
• Set a maximum image size and tell everyone who maintains content.

A fast website does not win an order on its own. A slow one loses it before the conversation can begin.