The Link Assistant uses the "best guess" to estimate the capacity in characters. It uses the characters and not words, as the words count would be even less precise that the characters count. The characters count is not precise as the "i" letter takes three time less space than the "m" letter, and two times less space than the "n" letter. The frequency of letters in a text is of course unpredictable for any given text, although it is statistically predictable. The most frequent twelve letters in English are ETAON RISHD LF and they cover of about 80% of total text. More information about this subject can be found here: https://en.wikipedia.org/wiki/Letter_frequency

From the text length point of view, the situation is much more complicated. The uppercases generally take more space than the lowercases. The uppercase E width in one font may be 61.9% of EM-space, while in another font can be 59.7%. The lowercase e width can be 50,3% or 51.5%. If the tracking and kerning is applied on the font, it changes the width of the overall text or just some pairs.

As the software had to get to an anchor point, it was decided to use the EN space width, that is an easily readable font-related value. Please note that it was a good or bad decision as any other. The EN width in a test font was 59.1% of EM space, which is closer to the uppercase E width (61.9%) than to the lowercase e width (59.7%). The uppercase T is 63.6%, the lowercase t is dramatically narrower, as 37.9%.

Therefore, the "best guess" in many cases can be a "quite bad guess", depending on the text, on the font, on the type size and leading.

In order to get a better precision, You can customize some measuring methods and adapt them to your situation. This chapter explains how and where to.

This is a super-user or IT job.

These functionalities are available in the GN4 version 2.5 or newer.