23ISO 25000-23Part 23 measurement record

Home / The record / 6.5 Interaction capability measures

ISO/IEC DIS 25000-23.2, 6.5.8 ยท Heading only

6.5.8 User assistance measures

User assistance asks how well the product helps a user who is stuck. A measure relates successful help to the help offered, or support demand to the tasks performed.

The free preview lists this heading and its subheadings but not the requirement text. This page states what the heading requires in our words and never invents the wording.

Not legal advice; does not certify. This page states our reading of a public draft, not the standard's wording, and it is not an official interpretation.

What this heading covers

Help takes many forms: inline hints, a help centre, a chatbot, a person. The family is about the product's assistance, so the measure should start with the point where a user needs it. Two useful shapes are the proportion of help interactions that resolved the problem, and the support contacts per unit of use. A help centre with excellent articles that nobody finds is not good assistance. The measure should record how the user reached the help and whether they returned to the task.

Record position: family 15 of 40 in clause 6, under 6.5 Interaction capability, which carries 8 families. It is one of the 40 families a programme has to map its measures to, and its elements are listed below. The heading is 6.5.8 in the public contents; the text behind it is not public.

How a measure for this family is built

Count help interactions that ended with the user completing the task as the numerator over help interactions in the period. Alternatively, divide support contacts by active users or by tasks performed, and read it as a demand measure with the caveat that lower is not always better.

Our worked formula

Help interactions that ended with the task completed divided by help interactions in the period.

Measure elements that can feed it

Number of user actions, Number of responses, Number of events

Chosen by us from the thirty-one quality measure elements named in Annex B. The draft does not publish which elements feed user assistance, so the choice above is a worked example, not the draft's assignment.

What to put in place before publication

  1. Define resolved from the user's task, not from the help article being opened.
  2. Record the help path, including search and chatbot.

Evidence to keep

KindWhat it looks like
Help pathHow the user reached the help and what was offered
OutcomeWhether the task was completed after the help and how long it took

The test a reviewer asks

  • Does the measure start from the user's problem or from the help content inventory?
  • Is a low support volume distinguished from good assistance?

Two questions to ask

  • Does resolved mean the user completed the task, or that the help was opened?
  • How did users reach the help: search, inline, chat or a person?

Where teams usually fall short

  • A count of help articles published, presented as assistance quality.

What the number cannot tell you

It cannot tell you whether the help was needed at all. Good assistance can mask an interface that should have explained itself.

What happens next

Self-descriptiveness is the design property that reduces the need for assistance in the first place.

Source

ISO Online Browsing Platform, free preview of ISO/IEC DIS 25000-23.2 https://www.iso.org/obp/ui#iso:std:iso-iec:25000:-23:dis:ed-1:v2:en read 22 October 2026

ISO/IEC 25010:2023, Product quality model https://www.iso.org/standard/78176.html read 22 October 2026

ISO/IEC 25023:2016, Measurement of system and software product quality https://www.iso.org/standard/35747.html read 22 October 2026

The rules this is built from

Read from the real rules, not invented.

This page is read from the public record for 6.5.8: the heading in the free contents, and where public, the text the preview carries.

  • Read from the real rules
  • 63 record pages behind it
  • 116 public headings checked
  • Check any line yourself

The measure set is read from the public contents; the model table is read from the public 25010:2023 model. Where the draft text is not public, the page says so instead of filling the gap.

Nothing here is guessed while you wait. Each page is written from the public record and points to the exact page it came from. Record updated 22 October 2026