Good Citizens · Usability Harness · Simulator ·
Slow connections: when every click waits
On a fast connection in an office, a page answers before you have finished moving your hand. For a great many people it does not.
Beside this, every click waits before anything happens. Try it at each delay and notice what your hand wants to do while it waits.
I · How waiting feels
A tenth of a second, a second, ten seconds
Under about a tenth of a second, a response feels instant. Under a second, your train of thought survives. Past ten seconds, your attention wanders off.
Those three limits are usually credited to the usability writer Jakob Nielsen, but he was summarising1 work by Robert Miller in 1968 and by Stuart Card and colleagues in 1991. In 1982 two researchers at IBM, Walter Doherty and Ahrvind Thadani, found that programmers got twice as much done2 when the computer answered in 0.3 seconds instead of 3: about 371 transactions an hour instead of 180.
When nothing seems to happen, people do the natural thing and click again. Analytics companies even have a name for it, the “rage click”. On a slow connection, that second click can order twice, pay twice, or send the person somewhere they did not mean to go.
II · Who it affects
The slow web is most of the world’s web
According to the International Telecommunication Union, 2.2 billion people6 are still offline, and 96% of them live in low- and middle-income countries.
Coverage is not the main problem; price and quality are. In nine of ten low-income countries, a modest 5-gigabyte mobile plan costs more than a tenth7 of a month’s income. Every heavy image is paid for by the person who downloads it.
It is close to home, too. In the United States, about 24 million people lack fast home broadband, including 28% of rural residents, and among households earning under $30,000 a year, about one adult in three9 is online only through a phone. And everyone has had one bar of signal on a train.
III · The numbers people quote
What the famous statistics really say
Some of the best-known figures about speed are weaker than they sound, and it is worth knowing which.
“53% of mobile visits are abandoned after three seconds” comes from a 2016 Google report3 built on its own analytics data, with no published method. “Every 100 milliseconds costs Amazon 1% of sales” is a former engineer’s recollection4, not a published study. The direction they point is supported; the exact numbers are not.
The firmer targets today are Google’s Core Web Vitals5: the main content should appear within 2.5 seconds, the page should respond to a tap within 200 milliseconds, and nothing should jump around while it loads, all measured for three visits in four.
IV · What to do
What it means for your design
- 01
Answer every click at once
Show that a click landed within a tenth of a second, with a pressed state or a small spinner on the button itself, even if the result takes longer.
- 02
Make a second click harmless
Disable a button after the first press and make the action safe to repeat on the server, so an impatient double click never orders twice.
- 03
Show progress for long waits
For anything over about ten seconds, show how far along it is. In a 1985 study, 86% of people preferred a progress bar even when it saved no time.
- 04
Test on a slow connection
Set a performance budget and test on a throttled phone connection, not office Wi-Fi. In Google’s 2016 tests, the average mobile home page took 19 seconds to load on 3G.
- 05
Treat page weight as the reader’s cost
Compress images, send small ones to small screens and drop what you do not need. For many people, every megabyte is money.
- 06
Be honest about loading
Placeholder “skeleton” screens help some people and not others: one study found they felt slower than a spinner. Keep them quiet, and never fake progress.
V · Prototypes
Works on prototypes too
This tool changes how your clicks and taps land, and it catches pointer presses as well as clicks, so it works on prototypes that are drawn as a picture, such as a Figma prototype link, as well as on live pages.
A few apps listen for presses in ways no bookmark can reach; if nothing changes when you click, the app is one of them.
VI · Myths
What people get wrong
- IBM proved a 400-millisecond threshold
- Doherty and Thadani’s paper compares 0.3 and 3 seconds and says productivity rises sharply below one second. It does not name 400 milliseconds.
- Nielsen discovered the 0.1, 1 and 10 second limits
- He summarised research by Robert Miller (1968) and Card, Robertson and Mackinlay (1991).
- The problem is coverage
- Most people live within reach of a 3G signal or better. The barriers are cost, quality and devices.
- Skeleton screens always feel faster
- The evidence is mixed. In one test they felt slower than a plain spinner.
Click it on any page and every click waits 0.4 seconds before it happens. Click again for 1 second, then 3, and once more to switch it off.
Works in Chrome, Safari, Firefox and Edge on a computer.
Or all 10 at once
Download the bookmarks fileOne Good Citizens folder with every tool in it. Download it, then import it:
- Chrome: Bookmarks › Import bookmarks and settings › Bookmarks HTML file
- Safari: File › Import From › Bookmarks HTML File
- Firefox: Bookmarks › Manage bookmarks › Import and Backup › Import Bookmarks from HTML
Every time you check a design against the people who will use it, you are doing what Good Citizens is for. Thank you for that.
More tools: Fovea · Colour vision · Ageing eyes · all 10
VII · Sources
The research behind this page
Every figure above comes from one of these 11 sources. The small numbers in the text point here, and each source points back.
Guidance
Response times: the 3 important limits·
Nielsen Norman Group
Report
The economic value of rapid response time·
IBM
Vendor data
The need for mobile speed·
Think with Google
Article
Marissa Mayer at Web 2.0·
Geeking with Greg
Guidance
Web Vitalsweb.dev
Data
Facts and Figures 2025·
Report
Global Connectivity Report 2025: affordability·
Report
Broadband deployment report·
Survey
Internet, broadband fact sheetPeer-reviewed study
The importance of percent-done progress indicators for computer-human interfaces·
CHI ’85
Article
A bone to pick with skeleton screens·