The 20 Minute Prompt That Turns Closed Tickets Into a Published FAQ

One thing before that CSV goes anywhere: strip the identifying details out of it. Tickets carry names, email addresses, order numbers and sometimes personal circumstances, and pasting them into a third-party AI tool is a disclosure of personal information under the Australian Privacy Principles. The OAIC's October 2024 guidance on privacy and developing or using AI is the plain-English version of what that means. For this job you only need the subject, the category and the count, so delete the name, email and company columns before you upload, and redact anything identifying that has ended up inside a subject line. Worth reading first if your ticket data is patchy: scoring your tickets for AI readiness before you start.

Step 3: Read the Output

You'll typically find that three or four topics account for the majority of your ticket volume.

That's not surprising, most support queues follow a power law distribution where a small number of issues generate most of the load. What changes when you can see it in a table is that you now have a prioritised publishing list, not a vague sense that "billing questions seem common."

Look for two things specifically:

The high-priority flags. Any topic with five or more tickets in 30 days is generating at least 60 tickets a year. If there's no self-service answer published anywhere, that's a predictable cost you're choosing to absorb manually.

The quality of the draft answers. Claude writes from the ticket subjects and categories, not from the actual ticket content, since those are rarely in the export. The drafts will be directionally correct but will need a review pass from someone on your team who knows the actual resolution. Don't publish them raw.

Step 4: Publish the Top Answers

You have two natural places for this content: your HubSpot Knowledge Base, or an FAQ section on your website.

HubSpot Knowledge Base (available in Service Hub Professional and Enterprise) is the right home if you want articles searchable inside your chat widget, accessible to your support team while they're inside tickets, and potentially picked up automatically by Breeze Customer Agent when it handles inbound queries. Go to Service > Knowledge Base > Create Article, paste in the FAQ answer, write a descriptive title that mirrors the question a customer would actually type, and add a meta description and URL slug before publishing. HubSpot indexes these articles externally, so they can rank in search.

A website FAQ page is the right home if you want the content public-facing and optimised for organic search from the start. Either way, start with the high-priority topics. The rest can follow.

One practical note: publish the content first, worry about the perfect format second. A published 50-word answer that's 80% right beats a beautifully structured article that's still in draft three months from now.

One forward note if any of this content ends up feeding an automated customer-facing decision rather than just answering a question: the OAIC's automated decision-making transparency obligation commences on 10 December 2026, and you have to be able to explain the decision if a customer asks.

Run This Monthly

The value of this hack adds up when you run it on a schedule.

Once a month, export the previous month's closed tickets, redact them the same way, and run the same prompt. You're looking for two things: new topics appearing in the high-priority category, and existing published topics that are still generating tickets, which means either the answer isn't easy enough to find, or it's not answering the actual question well enough.

That second one matters. A knowledge base that nobody uses isn't a support tool. It's a document library. The ticket data tells you whether your published content is actually doing its job.

If you want a second monthly check that takes about as long, try auditing your response time in five minutes using AI.

What to Do With What You Find

If this turns up a deeper problem, ticket categories that suggest a product or onboarding issue, volume patterns that point to a process gap, or a knowledge base that needs a full rebuild rather than a few new articles, that's the kind of work we dig into at Neighbourhood. Some of it is fixable the same afternoon, like automating follow-up and closure for stale tickets.

We're a Diamond HubSpot Partner and we've helped a lot of Service Hub customers turn a reactive support operation into one that actually scales. This prompt-and-review pattern is the small version of how NBH builds with AI engineering. If your support queue is running you rather than the other way around, get in touch.

Talk to us about your HubSpot setup →

Stay in the Loop

If you want more hacks like this one, practical, no-fluff, built for teams who actually run HubSpot day-to-day. Here's where to find us:

Ops Intel, our newsletter for HubSpot admins and RevOps teams, arrives in your inbox with exactly this kind of content. No padding, no roundups of stuff you've already seen. Subscribe to Ops Intel →

We're also on Facebook and YouTube if you'd rather watch someone click through it than read about it. Follow us there for more updates.

Happy HubSpotting.