Use Knowledge Base Reports to review article use, feedback, reuse and maintenance

Knowledge Base Reports helps you see how knowledge is used and where it needs attention. Open Reporting → Knowledge Base Reports, select a Report Type and scope, then choose Apply. Use the detailed events to investigate an article before changing it. A view count is not a unique-reader count, a Case link is not proof of resolution, and no views in a period is not enough evidence to archive a guide.

Confirm access and choose a question  ·  Run your first Knowledge Base Analytics report  ·  Understand calendar ranges, review dates and result limits  ·  Choose the report that matches the maintenance decision  ·  Investigate an article’s readership or Case reuse  ·  Interpret feedback without hiding the sample size  ·  Turn review, unused-content and draft lists into an owned queue  ·  Worked example: improve a recurring equipment troubleshooting article  ·  Troubleshoot missing reports and misleading comparisons

Confirm access and choose a question

You need the account’s Knowledge capability and the role permission View knowledge base analytics. Availability can also depend on your account’s current feature access. The reporting page and its data requests enforce access separately; a visible link does not grant unrestricted reporting. Team and Division choices are limited to permitted options.

1. Choose one question: which articles are read, which search terms deserve investigation, where readers report a problem, which procedures are reused in Cases, or what needs review.
2. Decide whether the question concerns customer/community consumption or internal agent activity. Use the appropriate Source when the selected report offers it.
3. Agree on a reporting period and scope before comparing results with another person or an earlier report.
4. Identify who can review the underlying article and who can change or publish it. Reporting permission is not the same as editing or public-publication permission.
5. Use a controlled maintenance process. An unfamiliar low-use article may serve a rare but consequential task, a restricted audience or a seasonal requirement.

Run your first Knowledge Base Analytics report

1. Open Reporting → Knowledge Base Reports. Confirm the page is Knowledge Base Analytics.
2. Under Report Type, choose the question you want to investigate, such as Article views. The overview tiles are a starting summary; selecting a report provides detailed rows.
3. When Source is shown, choose Community users for community activity or Agents for internal activity. Some reports have a fixed meaning and hide this selector.
4. Under Date Mode, choose Quick range or Custom (From / To). In Quick range, review Starting From and Time Frame. In Custom, enter both From and To.
5. Choose permitted Teams (optional) and Divisions (optional) when the question needs a narrower scope. Team filters apply to internal activity; Division filters target articles in those Divisions.
6. Choose Top results: 25, 50, 100, 250 or 500. Treat this as a result limit, not a promise that the displayed rows represent the complete library.
7. Choose Apply, wait for the result and inspect the active filters before interpreting counts.
8. Select a linked article ID, person, agent, search term, Case or Division where the report offers a drilldown. In Top authors, select the agent to open Articles created by this agent; in Top case linkers, select the agent to open Links created by this agent. Read the detailed evidence, then open the actual article or Case when needed.

Expected result: the chosen report reflects its supported source, period and authorized scope. Keep these selections with any conclusion you share. The current report surface does not provide an export/download control; do not direct a colleague to an invented Export button or treat another report’s export as identical.

Understand calendar ranges, review dates and result limits

Quick range offers 24 hours, 7 days, 14 days, 30 days, 90 days and 180 days. The date inputs use browser-local calendar dates that are converted for the report. Do not assume that the label 24 hours means a rolling window starting at the exact current minute. Inspect Starting From and the actual reporting context.

A Custom range includes the full To day. Use the same dates and local-time context for a repeat comparison. A period can be incomplete when it includes today; do not compare a partial day with a completed day as if they were equivalent.

If the numbers differ, compare the measure, source, dates, scope and limit first. An overview tile, a ranked list and an event drilldown can answer related but different questions.

Choose the report that matches the maintenance decision

Source is offered for Article views, Searches, Top commenters, Top voters and Feedback health. Reports such as Top authors, Cases linked, review queues and Division scorecard have their own defined meaning; do not assume changing a source elsewhere reinterprets those measures.

Investigate an article’s readership or Case reuse

1. Run Article views with the appropriate Source, dates and Division.
2. Select the article ID to open Article view events. Review the event dates and available person information; repeated events are not necessarily different people.
3. Open the article and check whether the procedure is current, complete and accessible to its intended reader. A popular article can still contain a missing or incorrect step.
4. Run Cases linked for a comparable scope when you need reuse evidence. Inspect Linked By and Linked On, then drill into the Case or article.
5. Read the Case’s actual outcome. A link may identify a useful starting procedure, an unsuccessful attempted resolution or a reference shared for context. It does not by itself prove the issue was resolved.
6. Record a specific improvement action if the evidence identifies a gap. Use the article’s feedback or authoring workflow rather than replacing it with a near-duplicate.

In B2B support, a low-volume procedure can be essential for a specialized asset or a restricted customer group. Judge usefulness in its actual context instead of comparing every article against a widely used onboarding guide.

Interpret feedback without hiding the sample size

In Feedback health (thumbs up/down), Score is the percentage of positive votes among positive plus negative votes, rounded to a whole percentage. For example, 8 Up and 2 Down gives 80%. The zero-vote calculation displays 0, which must not be interpreted as proof that readers disliked the article. Always read Up and Down alongside Score.

1. Choose the intended Source and period, then Apply.
2. Compare the vote counts before comparing scores. One positive vote and a large positive sample can both produce 100% but carry different evidence.
3. Select the article ID for the feedback events. Review dates, people where available, vote and comment.
4. Find the actual problem in the article: an incorrect control, unsupported version, missing prerequisite, unclear recovery or access mismatch. Do not infer the cause solely from a negative vote.
5. Create a precise correction request or review the edited article through the permitted authoring route. After publication, retest the procedure and compare a later equivalent reporting period.

Turn review, unused-content and draft lists into an owned queue

1. Use Review due / overdue and deliberately select or clear Overdue only. Read the owner and review date for each item you take on.
2. Open the article and review the actual procedure, audience, links and media. Record the correction or the evidence supporting retention; merely extending a review date is not a content review.
3. Use Unused articles (0 views) to identify investigation candidates. Test its public/customer access and findability, and check whether the chosen period matches the task’s likely use.
4. Compare overlapping articles in full before consolidation. Preserve distinct supported steps and publish/verify the useful replacement before retirement or redirection.
5. Use Draft articles to coordinate unfinished work with its owner. Inspect the existing draft/review rather than creating another article to avoid the workflow.
6. Record the next action and responsible editor/reviewer through your team’s supported workflow. Revisit the list after the changes and verify the current saved result.

Worked example: improve a recurring equipment troubleshooting article

A knowledge manager sees controller-alarm searches, a highly viewed troubleshooting guide and several negative votes. The task is to learn whether the procedure is useful, not to assume that more views prove success or that negative feedback proves the product failed.

1. Run Searches for the agreed period and confirm the actual phrases. Use manual search as the intended audience to check which current guide is found.
2. Run Article views with Community users and the same applicable scope. Open the article events and then the guide itself.
3. Inspect Feedback health and its comments. A note that step 4 uses an obsolete control is actionable; an unexplained vote needs further investigation.
4. Review Cases linked to that article. Check the product/version and whether the documented verification actually succeeded for the customer.
5. Assign the specific content correction, preserve the canonical article identity and complete review/publication. Verify access, steps and media after saving.
6. Repeat the same search and procedure test. Later compare equivalent reporting windows, keeping vote counts and case outcomes separate from any claimed benefit.

Troubleshoot missing reports and misleading comparisons

Continue learning

• Understand original authorship and article-count differences
• Interpret Dashboard knowledge widgets separately
• Run the KCS capture and improvement process
• Flag a precise article correction
• Review and publish article changes
• Test article findability and AI answers
• Configure knowledge topics and their audiences