Job help-centre-gap-report-from-your-tickets · revised 12 October 2026
Get a ranked list of customer questions that no help article answers, with example tickets
A ranked list of the questions customers ask that no article answers, each marked missing, outdated or hard to find, with at least three example tickets as evidence.
You might be seeing
- The same question keeps arriving despite a help centre
- The bot gives answers that miss the point
- Nobody knows which articles are missing
No passwords, keys, card details or admin invites needed to start.
What usually happened
Customers write in because the article is missing, out of date or cannot be found, and nobody counts which. Irrelevant bot answers can be a content gap rather than a bot fault.
Who it’s for: Support lead whose bot or help-centre search gives irrelevant answers.
Usually starts when: Customers ask the same things again and again even though you have a help centre, and the bot's answers are technically right but not relevant.
The result: A ranked list of the questions customers ask that no article answers, each marked missing, outdated or hard to find. Each cluster carries a count, a link to the nearest article and at least three example tickets, and billing issues, bugs and feature requests are kept apart.
Check whether this job fits
Answer these without sending chats, tickets, files or logins. Nothing is submitted unless you choose to contact us.
What you get
- The ranked gap report with counts and verdicts
- Example tickets for each cluster
- The article links
- The list of items separated out (billing, bugs, requests)
- A page on caveats
Included
- The last 90 days of tickets or chats, redacted by you, and an export of your help centre
- Clusters of questions with counts, a verdict for each and the article linked where one exists
- At least three example tickets for every cluster
- Billing issues, bugs and feature requests separated from content gaps
- A page on what the counts can and cannot show
Not included
- Writing or changing any article: see the draft articles job
- Changing your bot or help desk
- A forecast of how many tickets new articles would remove
- Chat or ticket data we have not agreed to handle
- Voice or call recordings
- Counts of every theme in a short shop ticket list, without a gap ranking: see the store questions job
How we know it’s done
Agreed with you before work starts. Each check produces evidence you keep.
Your support lead spot-checks 20 clusters (seed recorded) and agrees with the verdict on at least 18 (18 of 20 shows a true agreement rate of at least 71% at 95% confidence, so the report says so).
Evidence: The 20 clusters with your lead's marks.
Every cluster lists at least three example tickets, and the cluster counts reconcile to the ticket export.
Evidence: The report with example tickets, and a reconciliation table.
Billing issues, bugs and feature requests are separated from content gaps, and the separated count is shown.
Evidence: The separated list.
Sign-off. Your lead reads the report and the spot-check result and you sign off in writing. Payment follows sign-off.
If it fails. If your lead rejects more than 2 of the 20 clusters checked, we re-cluster once. If it still fails, we hand over what we have and you decide whether to stop or re-quote.
When it fits, and when we stop
It fits when
- A ticket or chat export of the last 90 days, with personal details removed
- An export or crawl of your help centre
- About 300 tickets or more in the 90 days (100 to 300 gives a smaller report)
- A support lead who can spot-check 20 clusters
We stop and tell you if
- Fewer than about 100 tickets in 90 days: there is no signal
- No export of tickets is possible
- Tickets hold special-category data we have not agreed to handle
What could go wrong
We only read exports and write a report, so nothing in your systems changes. Our copies of your exports are deleted when the job ends.
Scroll the table sideways to read it all.
| Risk | How we handle it |
|---|---|
| A cluster mixes two different questions | Your lead spot-checks 20 clusters and any cluster they reject is split or removed. |
| Counts are read as demand or savings | The report states the period, the export size and that counts show tickets, not future deflection. |
| Personal data in tickets | You redact before sharing, and examples shown in the report are redacted again. |
A separate automated reviewer checks the clusters, counts and verdicts before hand-over, and your support lead on your side marks the sample and decides what goes live. No human supervisor is included unless your agreement names one. At launch the work is largely automated, and we say so.
How we deliver
We arrange the work and independent review, then show you the result against the agreed checks. You keep authority over your systems.
- Agree the exports, the periods and what counts as a gap in writing
- Separate billing issues, bugs and feature requests from content questions
- Cluster the questions and match each cluster to the nearest article
- Mark each cluster missing, outdated or hard to find, with three example tickets
- Independent automated review; your lead spot-checks 20 clusters
- Hand over the ranked report and the caveats
This is a one-off job, not emergency cover or a subscription. We confirm eligibility, the total price, a start window and a delivery date before you accept. Work starts only after agreed inputs, secure access and any needed permissions are in place. Model usage for our build and tests is included in the price; charges on your own platform account, including per-conversation fees after go-live, are yours. An enquiry creates no charge or booking.
Need to keep it working?
Discuss a monthly service that keeps the help centre current and re-lists the gaps.
Ongoing work is separately scoped and quoted: no monitoring, response-time guarantee or automatic subscription is included in this job.
Explore an ongoing managed lane, or mention the responsibility you need in your enquiry.
What you can check
This is a new service. We have not delivered this job for a client yet.
Other ways to get this done
- A help-centre vendor offers a free report of your top 20 knowledge-base gaps, ranked by ticket volume, in about three business days. Try one before paying for this. irisagent.com
Questions
Will new articles reduce my tickets?
We do not promise that. We show which questions have no article, with example tickets.
Do you write the articles?
Not in this job. The draft articles job does that for the top gaps.
Is my data kept?
We delete our copies when the job ends.
How is this different from the store questions-counted job?
That job counts every theme in up to 500 shop tickets, answered or not, and notes which of your pages covers each. It does not rank gaps or mark articles missing, outdated or hard to find. This one needs a help-centre export and does.
Send an enquiry
Send us
- Your help desk and help-centre platform
- Roughly how many tickets arrive each month
- The kinds of question you suspect are missing
- No customer data, logins or keys in the first enquiry
Later, once you agree
- A redacted export of 90 days of tickets or chats through a secure route agreed in writing
- An export of the help centre articles
- Your support lead for the spot-check
You keep the help desk, the help centre and every login. We work from exports you redact and change nothing. You decide which gaps to fill.
Email fallback: open your mail app
If website submission is unavailable, review and send the fallback email yourself. An email fallback is not a website receipt. Or write to hello@syntheticindustry.ai with “help-centre-gap-report-from-your-tickets” as the subject.