How to find and fix help center gaps
Find help center gaps by grouping support tickets into customer questions, checking article coverage, and deciding which content to write or revise.
Free app. Runs on your Mac with the Claude or Codex plan you already have.
What help center gaps is
Help center gaps are places where customers need an answer they cannot get from your existing self-service content. The answer might be absent, too narrow, out of date, contradictory or buried under unfamiliar language. A knowledge base gap analysis pairs what customers actually ask with what the relevant articles actually say. It ends with a ranked list of content changes the team can assign.
Support tickets give you a record of questions customers brought to the team. Start with a dated ticket export and the help content for the same product and audience. Group tickets by the question a reader needs answered, then read the closest articles as that reader would. A single word can match five pages while the answer is missing from all of them. The unit of the audit is the question and its coverage.
What it does for the business
- Spend writing time on questions customers bring to supportRank gaps by how many distinct people ask so the backlog reflects the reach of each problem. Give the first items a writer and a product reviewer with the source tickets already attached.
- Correct conflicting instructions in one passPut both passages in front of the policy or product owner so they can choose the supported answer. Then update linked guidance together to give the support team and customers a consistent explanation.
- Make existing answers easier to findUse customers' own search terms to revise a title, heading or opening paragraph. Agents can point readers to a familiar phrase, and customers can recognize the right page when they see it.
- Improve support replies while content is pendingGive agents the confirmed facts and the exact questions the content team is covering. The team can answer open tickets and share the revised article once it passes review.
Help center gaps, step by step
Turn tickets into specific customer questions
A group should contain questions that the same answer would resolve.
What to do
- Choose a date range, product area and audience. Export ticket IDs, subjects, messages, requester IDs and creation dates. Keep account IDs if they help you see whether several people from one account encountered the same problem.
- Separate product questions from sales inquiries, spam and outage reports. Document the exclusions and their counts. Keep restricted details out of example wording while retaining ticket references the reviewers can follow.
- Start groups from shared wording, then read the members. Merge phrases like pause recurring billing and stop my invoices temporarily when they ask for the same action. Split questions about canceling a subscription from questions about canceling one invoice.
- Name each group as a question customers would recognize. Track tickets and distinct customers separately so repeat contact from one requester stays visible without overstating the reach of the problem.
- Review borderline tickets where one message asks several things. Decide how you will record those cases and apply the same rule across your scope. Keep the membership list so another reviewer can trace the totals to the export.
Why
Good groups give the content reviewer a specific answer to look for. Broad labels such as billing hide the difference between changing a payment method and pausing a schedule. Broad groups can also make a large count appear to support a page that would answer only part of the demand.
Read the article sections that might answer each question
A matching phrase gives you somewhere to read and the passage tells you whether the question is answered.
What to do
- Search for the question in customers' words and for the terms used in product documentation. Open likely matches and read the section around each hit. Record the article title, heading, audience and update date.
- Check that the instructions work for the intended reader. Compare plans, roles or versions when they change which steps are available. A correct answer for an administrator can still leave an account member without guidance.
- Classify article coverage as missing, partial, conflicting, hard to find or covered. Missing help articles leave a question unanswered in the material reviewed. A partial answer gives some of the needed steps. A hard-to-find answer is usable once the reader reaches it.
- For a conflict, save both passages and their dates. Check whether they address the same scenario before calling them inconsistent. For a findability problem, save the customer's wording alongside the article's title or heading.
- Keep a note for covered questions too. Customers may still contact support when a page is clear. Review those cases separately for clues about where people encounter the link or how agents use it.
Why
Matching by title alone can send writers toward the wrong fix. An article on refunds might explain eligibility while leaving timing unanswered, or a second article might give a different answer for the same purchase. Reading the passages together shows which change the writer should make.
Rank the gaps and specify the fix
A ranked gap earns its place by showing who is affected and what should change.
What to do
- Rank by distinct customers, then by tickets and recency. Put contradictions about money, account access or security ahead of routine wording fixes when the wrong instruction could have serious consequences.
- Choose a new article when the answer has no suitable home. Write a title in customer language, list the questions it must answer and outline the sections. Give the writer example wording so the scope stays close to the tickets.
- Choose a new section when there is a useful article with an incomplete answer. Name the article and heading and state which missing decision or step the section must cover. Include prerequisites where they affect the answer.
- For a correction, record the old wording, the proposed wording and the source that supports the change. For findability, specify the heading or title to change and the customer words to include.
- Mark any fact that needs confirmation and name the person or team that can settle it. A page about changing a limit can have a useful outline while the exact rule is still open.
Why
A clear fix turns evidence into work a writer and reviewer can divide between them. Frequency gives routine items an order and consequence helps the team spot low-volume problems that deserve attention. Keeping open facts visible lets other fixes proceed while someone checks the uncertain rule.
Check the findings and hand off the fix list
The next writer should be able to verify the gap without repeating the whole audit.
What to do
- Ask a second reader to check counts against the final group membership and test the missing findings against the help content. Have them check that the citations resolve and both passages are present for each conflict.
- Check that covered questions are accounted for outside the ranked gaps. Confirm that partial gaps identify the information the page lacks, and that suggested wording draws on facts the approved source material supports.
- Put the first decision at the top of the report. State the period, tickets and articles read, which records were excluded and why, and any audience or content that the review could not cover.
- Give each fix a destination and a next action. Distinguish changes ready for drafting from ones that need a product owner. Carry the reviewed order into the writing queue so the most important changes get an owner first.
Why
The review establishes how far the team can rely on the ranking. When a product area or article set is outside scope, the report should make that visible before a reader uses it to set priorities. Clear open questions also prevent draft text from turning a working assumption into an instruction.
Common mistakes
- Grouping by a product label aloneRead the member tickets and split groups whenever the customer would need a different answer.
- Equating ticket count with customer countKeep distinct requester counts next to ticket totals so a long back-and-forth stays in proportion.
- Calling a question missing after reading one matchCheck the other plausible sections and search with both customer and product wording.
- Treating every repeated question as a content gapKeep clearly answered questions visible outside the gap list and look separately at how customers reach those answers.
- Writing an instruction from a guessRecord the open product fact and assign it to someone who can verify the instruction.
What a good result looks like
- Each gap has a customer question, example wording and traceable ticket references.
- Both ticket count and distinct customer count come from the final grouping.
- Each classification points to the relevant article section or explains the search for an answer.
- The suggested fix identifies an article, a section or a proposed new page.
- The report states what sources, period and audience it covers and what it excludes.
And then what? Using the result
- Assign the first fixes to writers and route each unresolved product fact to the responsible owner.
- Brief new articles with the example wording, exact question and outline from the gap list.
- Coordinate changes to conflicting articles so both pages reflect the same confirmed rule.
- Prepare support replies from verified information while the articles go through review.
- After the changes go live, watch for the same questions in later tickets and check whether customers can find the revised answer.
When to run it
- When the same questions keep reaching supportGroup the wording first, then check whether the repeat contact reflects a missing answer, unclear guidance or a discovery problem.
- When a product or policy change affects several pagesCompare the old instructions against the current rule and identify every passage that needs review.
- Before planning documentation workUse the ranked gaps to decide which articles to write, which sections to amend and which facts require a product decision.
Questions people ask
How do you find help center gaps?
Group a dated ticket export into questions, then read the relevant help content to see which questions receive a usable answer. Rank unanswered and poorly answered questions by reach and consequence and specify one content change for each.
What is a knowledge base gap analysis?
A knowledge base gap analysis compares customer needs with what your documentation actually answers. It produces evidence for adding, correcting or making content easier to find, along with a scope statement that records what was reviewed.
How do you prioritize missing help articles?
Start with the number of distinct customers who asked, then use ticket count and recency to order similar gaps. Move advice that conflicts on a consequential issue such as payment or account access ahead of routine gaps for prompt review.
Can you find documentation gaps without search analytics?
Yes, match questions in support tickets against the article text. This tells you what the people in the export asked and how well your content serves those questions. Search behavior and visits before a ticket remain outside that evidence.
How do you tell if an answer is hard to find?
Compare the terms in the tickets with the titles and headings a customer would see while looking for the answer. If the instructions are clear once found, save the wording mismatch and show where a title or heading change would help.
What if the team cannot confirm the correct product answer?
Keep the fix as an outline or correction request with the missing fact clearly marked. Assign a product or policy owner to confirm it before anybody publishes a specific instruction.
More jobs in Support
Find gaps in the help center, on your own Mac.Free app. Runs on your Mac with the Claude or Codex plan you already have. Your files stay on your computer.
Download for Mac