Mobile phone numbers in electronic forms

Some time ago Jessica Enders, Principal of Formulate Information Design, conducted some user research for SEEK that involved putting an invitation to participate on the SEEK website. A screen shot of the main page of this invitation appears in Figure 1 below.

In order to participate, users had to provide a contact number in the box titled “Your Phone Number”. As you can see, no constraints were provided on what sort of phone number the form-filler should enter, i.e. mobile (also known as cell) versus landline. We also did not perform any validation on the phone number, letting the form-filler enter it in exactly the way that was most natural for them.

Figure 1: Invitation used to recruit research participants.

Given this design, we can use the phone number data to get a sense of how the majority of Australians provide their phone number, by default. In this post, we’re going to look at mobile phone numbers, with a later post to discuss landlines. Many thanks go to the team at SEEK for allowing access to this invaluable data.

About Australian mobile phone numbers

In Australia, mobile phone numbers are 10 digits long and start with a 0. For example: 0419 123 456.

10 digits makes for a long string, so some people break the number up, as we did in the example above. Here the number is broken into 4 digits, a space, 3 digits, another space and the final 3 digits. Depending on the particular digits in the number, some people might break it up in other ways, e.g. 0417 12 12 12.

What we were interested in knowing was whether there was any pattern to how Australians write their mobile phone number. If there is a stable and consistent pattern, we can use this to decide what sort of field should collect this data in an electronic form. For example, credit card numbers are consistently broken into 4 chunks, e.g. 4133 1234 5678 9012. This provides an argument for collecting such numbers via 4 separate text fields [1].

The research findings: one long string is the clear winner

Out of 2,081 mobile phone numbers provided by interested research participants, a whopping 83% were entered as one long string of 10 digits (i.e. no spaces and no chunking).

Another 11% were broken into chunks of 4, 3 and 3 digits with spaces in between, as per the example number 0419 123 456 above.

The other two common methods, each representing 1% of the total sample, were:

  • chunking into 4 digits followed by 6 digits, e.g. 0419 123456; and
  • including the country code [2] and dropping the leading 0, e.g. +61419123456.

The long tail

There were some 40 other variants on the way people entered their mobile phone number into the text field, but all of these each represented less than 1% of the total sample.

There was a incredible amount of diversity in these variants, including:

  • Different chunking to suit the particular digits in the number,
    e.g. 0417 11 22 33 and 0416 77 1234.
  • Different chunking unrelated to the particular digits in the number,
    e.g. 04 1912 3456 and 04191 23456.
  • Different ways to include the country code,
    e.g. +610419123456, +61 (0) 419 123 456 and 61 0419123456.
  • Putting in the area code,
    e.g. (02) 0419 123 456 or 02-0419123456.
    (This was presumably because we said “please include your area code”, although that was intended for and directed at landline numbers, which is what most people seemed to assume.)

Is this behaviour particular to electronic forms?

Because chunking facilitates memory, we had expected that the chunked versions of the mobile number would be the most common, rather than the unbroken 10 digit string. The research findings were therefore a little surprising to us, although it does make sense that it is faster to type 10 digits in a row than to add in spaces.

This makes us wonder whether the finding is particular to the fact that the mobile numbers were collected via an electronic form. Not only is it faster to type an unbroken string, but the data is much more likely to be accepted during validation. Perhaps form-fillers have learnt this and avoid spaces, parentheses, dashes and other non-numeric characters because they’re unsure whether such characters will be accepted.

As such, we don’t think it is necessarily wise to assume we would find the same pattern of behaviour if we were collecting data from paper forms, where there is no such ‘real time’ validation.


As Russ Weakley of Max Design, points out [3], this research can’t tell us the way Australians prefer to enter their mobile number into an electronic form. What it does tell us, however, is that a single long text field is at the very least going to work for the largest majority of form-fillers.

To allow for the diversity of use this text field should accept at least 12 characters (10 digits plus 2 spaces). The form should also:

  • give an example of formats that will be accepted; and
  • make it clear whether country code is needed.

[1] Knowing that people typically chunk a number when they write it is an argument for collecting that number in the same chunks. However, there are also counter-arguments for doing so. For example, more key strokes/mouse clicks are needed if the form-filler has to tab from one text field to the next. Thus the decision about whether to have separate text fields for chunks in a number is a little complex. More on this in a future article!

[2] As is the nature of the Internet, although the pages of the SEEK website we were interested in promoted jobs in Australia, it is visited by people both inside and outside the country. Also, it is possible that the form-fillers were not sure if the researchers were located in Australia, especially as the incentive was an Amazon gift voucher (there is no Australian Amazon site – people here usually use the US or UK Amazon sites instead.

[3] Russ emailed us shortly after this post was published. In his email, Russ pointed out that if the desire to avoid error — by entering a single unbroken string — is influencing the behaviour of a significant portion of our form-fillers, then it would be dangerous to infer preference from the observed behaviour. As such, we updated the recommendation in this post on 26 September to make it clear that we suggest a single text field as a practical default, rather than asserting it as something that users want.

And just a reminder, if you have a comment on this or any other post of ours, please follow Russ’ lead and get in touch with us. We’re always keen to hear what our readers think.