Prioritize documentation work with Insights

Documentation teams often know they should improve something, but not what will create the biggest impact first. This guide shows how to use Insights to prioritize documentation work using actual reader, assistant, search, and support behavior rather than intuition alone.

Who this is for

• documentation leads running an active docs program
• support and product teams trying to reduce avoidable questions
• teams with too much documentation work and not enough time

What Insights measures

The Insights surface includes sections for:

• Content performance
• Assistant activity
• Search behavior
• Reader engagement
• Team adoption
• Sentiment & responses
• Link health
• Missing pages
• Support impact
• Support tickets

This is enough to build a strong prioritization model for documentation maintenance and expansion.

Step 1 — Start with the right scope

Open Insights and set:

• a time range such as 7d, 30d, 90d, or All time
• a Project filter if you only want one docs property
• Compare when you need to look at relative movement across time or projects

For most operational review cycles, 30d is the best starting point because it balances recent behavior with enough volume to reveal meaningful patterns.

Step 2 — Identify high-traffic content first

Begin with Content performance.

Use it to find:

• the pages that receive the most traffic
• the entry pages readers see first
• whether high-traffic content has weak completion or poor engagement

These pages deserve first attention because even small improvements there affect more readers than low-traffic pages ever will.

Step 3 — Use search behavior to find missing or mismatched content

Move next to Search behavior.

This section helps answer:

• what readers are searching for
• which terms fail to return useful results
• whether the site vocabulary matches reader vocabulary

Strong signals for documentation work:

• repeated zero-result queries
• search terms that imply a page should exist but does not
• search phrasing that does not match your internal naming

These are often some of the highest-value docs opportunities because they reveal explicit reader intent.

Step 4 — Check assistant activity for unresolved demand

Use Assistant activity to understand:

• where readers rely on the assistant heavily
• which pages are associated with assistant conversations
• which intents or question types are most common

This helps you distinguish between:

• pages that are already helping readers through normal reading
• pages or topics where readers still need conversational help

If the assistant repeatedly handles the same topic, that topic may need:

• a clearer article
• a better quickstart
• stronger troubleshooting content
• more direct explanation near the top of the relevant page

Step 5 — Use sentiment and feedback to evaluate experience quality

Then inspect Sentiment & responses and any feedback-related signals.

These metrics help answer:

• whether readers find pages helpful
• which topics generate frustration or confusion
• whether feedback is positive, neutral, or negative

This is especially valuable when traffic alone would mislead you. A busy page is not necessarily a good page.

Step 6 — Watch link health and missing pages

Use Link health and Missing pages to catch structural issues that interrupt reader flow.

Prioritize:

• broken links in high-traffic pages
• redirect chains that create friction
• missing URLs with repeated visits
• referrers that point readers into dead ends

These are often easier to fix than content rewrites and can remove a surprising amount of user friction quickly.

Step 7 — Bring support signals into the prioritization model

Use Support impact and Support tickets to find where the docs are saving time and where they are failing to do so.

Questions to ask:

• which pages are actively deflecting support demand?
• which topics still generate tickets?
• are there repeated ticket categories that should become articles?
• are unresolved assistant or feedback flows turning into avoidable support work?

This is where documentation can demonstrate business value, not just editorial cleanliness.

Step 8 — Turn the data into a weekly priority list

At the end of each review cycle, create a flat priority list with categories such as:

1. high-traffic pages with weak quality
2. repeated zero-result or weak-result search terms
3. broken links or missing destinations affecting active journeys
4. common assistant questions that deserve clearer source pages
5. support-generating topics that should be documented better

Keep the list short enough that the team can actually finish it before the next cycle.

A practical prioritization model

Use this simple rule:

• highest priority = high traffic + high confusion + clear fix
• next priority = repeated missing demand + easy article creation
• then = structural issues such as links, redirects, and teaser handling
• then = lower-volume refinements

Step 9 — Re-check results after changes

After you make the improvements:

• return to the same Insight sections
• compare the same time windows where possible
• look for reduced failure signals and stronger engagement

This closes the loop and helps the team learn which kinds of documentation work create the biggest returns.

Related pages