DESIGN
The story of touch-tone keypads.
Telephone illustration by the author.
Every time you pick up your mobile phone to make a call you’re presented with the layout below. But has it ever occurred to you to wonder why the numbers are arranged like this? Or why the calculator on your phone is arranged in the opposite direction?
It turns out that the reasons behind these layouts are a useful illustration of some key design principles that apply across the board.
Keypad vs calculator, screenshots by the author.
The tale begins at Bell Labs in the 1950s. Researchers were looking for a faster and more accurate way for users to dial, using buttons instead of a rotary dial. The idea was that increased accuracy would reduce the amount of wrong numbers, while dialling faster would reduce the connection time and thus make it easier for telephone exchanges to handle the increasingly large number of phone calls.
This two-minute video offers a bit more background information:
https://medium.com/media/9ce1597109c522e7544daf94ccb24b43/href
The team at Bell Labs tested a number of possible layouts for the keypad of their new touch-tone phones. Most of these look wrong to modern eyes, because we’re so used to seeing the “winning” layout on our products. But at the time all of these alternatives were equally valid.
Bell System Technical Journal, July 1960, p. 1000. Annotations by the author.
It might seem sensible just to use the familiar calculator layout for phones, but this was tested and rejected at an early stage because it was unintuitive and failed to perform significantly better on any of the metrics studied.
It’s worth bearing in mind that the study tested data entry using telephone numbers, which are essentially random strings of numbers. The calculator layout is not intended for entering random numbers, but for entering real-life data — and according to Benford’s Law it’s more likely that in a real-life data set the leading digit will be smaller. A real-life data set is also likely to have more zeroes for various reasons, including numbers being rounded up or down. It therefore makes sense for frequently used keys such as 0, 1, 2, 3 to be placed at the bottom of the keypad, closer to the hand, where (according to Fitt’s Law) they can be pressed with greater accuracy. This does not apply to “non-natural” numbers such as telephone numbers.
The findings of the Bell Labs study were interesting and in some cases unexpected:
Testers were fastest at dialling using the Telephone layout which replicated the layout of the rotary phones they were used to.Testers produced the fewest errors when using the Vertical Columns layout, but despite this they hated this layout the most.Testers loved the Horizontal Rows layout, even though it was the slowest to use.
Interestingly, Bell Labs did not select the fastest, least error prone or most popular layout. The chosen option was actually the one which was the most logical for the majority of people who participated in the study. English writing tends to go from left to right, top to bottom, so most people (55%) thought the Three-By-Three-Plus-One layout was the correct design for a keypad. This preference also applied when letters were added to the numbers on the keypad.
Supposedly this layout also offered some engineering advantages, so Bell Labs decided to go with this option, which has since become the expected standard on all products from phones to ATMs and card machines (although calculators have still retained their traditional layout).
Usability pioneer Jakob Nielsen estimates that this keypad has been used approximately 40 trillion times since it was designed, and the world has saved a collective one million person-years of time.
This leads us to a number of key points for design:
What people say they prefer is not necessarily what works best. It’s important to note that had Bell Labs relied on users’ self-reported preferences, they would have gone with the Horizontal Rows layout which is also the slowest option. This highlights the need for gathering feedback and insights by watching real users completing tasks.What people find easiest to use is not necessarily what they prefer. Preferences are influenced by individual tastes, habits, cultural background, aesthetic preferences and emotional responses. A design may be preferred by users if it aligns with their personal preferences, even if it is not the most usable option.The fastest solution for completing the task is not always the best. Other factors such as accuracy, ease of use, enjoyment and emotional engagement also contribute to the experience of using the product.The solution varies depending on the specific task. Typing numbers on a telephone seems superficially similar to typing numbers on a calculator, but when the tasks are studied in more depth it becomes clear that the data being entered is entirely different, thus warranting a different design solution.Any improvement is better than none. The chosen keypad layout was not the fastest or most accurate, but it has still saved an enormous amount of time and had a huge impact on the modern world.Familiarity often trumps other considerations. Even if the approach that is currently in use is less than ideal, it may be worth persevering simply due to the benefits of consistency and standardisation.In some cases all of these factors are superseded by engineering constraints or other issues. It’s important to take a holistic approach to design which balances various factors including usability, preferences, task requirements and engineering constraints.
One final point which is of particular interest: the most important factor in keying times turned out to be the individual themselves. Specifically, average keying time is correlated more with how long the person pauses between typing each digit, versus how long it takes them to key each digit. Those who memorised the entire number and keyed it in one go were faster than those who typed half of the number, paused to refer back to the written number, then keyed the remaining digits.
In other words, human cognition has a far greater effect on performance than any physical-design-related factors. I often say “The most important thing about a product is who will use it”, and this is a neat demonstration of how that principle applies in practice.
What telephones can tell you about good design was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.
