What does it do?
Lists contact details (emails, phone numbers and postal addresses) within a website.
Why is it important?
This information can be useful, particularly for:
- Checking these details are correct
- Updating these details across a large site
- Identifying details which spammers may exploit
How is it measured?
This test never awards a score, and is for informational purposes only.
The entire text of each page, including invisible text (metadata, content of fields, alternative text) is analyzed for each page.
Sitebeam uses various complex means to uncover contact details, including:
- Email addresses are identified as text containing an @ symbol (they don’t need to be a mailto: link). This therefore identified email addresses inside Alternative text, metadata etc. Phone numbers are contiguous sequences of numbers and supported symbols (brackets, dashes) with various positive and negative validation rules applied. Discriminating factors include length, prefix and context.
- Where possible, the context of a phone number tells Sitebeam that it is likely a fax, mobile or phone number.
- Addresses match based on a large range of complex criteria, intended to recognize most international addresses.
- The precise details of these algorithms are a protected trade secret and are updated over time as we identify potential improvements.
Being an automated test, context and subtleties of each language are not always fully appreciated. The test may not be certain where a sentence ends and a postal address begins, and the line may be equally confusing for some human beings. Consider:
Meet us at The Example, Example Street, London, AB1 DE2.
To a computer, the first part (“Meet us at the Example”) appears part of the address, and without a full grasp of every natural language in the world, over-extensions of this nature are inevitable, but rare.
Similarly, some phone numbers are hard to distinguish from natural numbers:
1332-224-800 (phone number) 1332-224-800 (mathematical sum) 1332-2007 (date range)
In practice we use a lot of artificial intelligence to reduce instances of this, and to suggest ‘possible’ matches where appropriate, but without human-level intelligence an automated test will occasionally get these wrong.
How to use this test effectively
This test is most useful for infrequent checks that contact details are present and correct on a website, or when contact details need to be updated.