Machine Readability

Machine readability is whether the meaning of your page survives being read by software. Not whether the page looks good, loads quickly or wins design awards — whether a system that requests it, takes the text and moves on ends up holding an accurate idea of what you do.

It is the plainest test in this whole territory, and most failures are failures of access rather than of writing. Text living inside an image cannot be read. Text inside a downloadable PDF is largely invisible to the thing deciding whether to recommend you. A page that assembles itself only after code runs may arrive empty. A crawler refused at the door reads nothing at all. In every case the words exist and simply do not travel.

The rest is legibility rather than access. A homepage of adjectives — passionate about excellence, delivering integrated solutions — is perfectly readable and says nothing a machine can use. Facts are readable in a way adjectives are not: family-run since 1994, three thousand roofs repaired, offices in two cantons. Each of those can be lifted, checked and repeated. None of the adjectives can.

This is what makes the discipline finite rather than endless. There is a fixed list of ways meaning fails to survive the trip, most of them cheap to correct, and once corrected they stay corrected. It is also why the work rewards checking over guessing: not one of these failures is visible from inside your own browser.

The human translation is the difference between what you said and what was heard. You cannot improve the hearing. You can only say it in a way that survives the room.