Turn 10 Support Tickets Into a Help Center Article (Free Prompt)

Your support team already has the best help center content you will ever write. They write it every day - 50 times a day - answering the same 20 questions.

The problem is that it is buried in Gorgias tickets, Zendesk threads, and Richpanel conversations where nobody else can find it. Meanwhile, your help center is either empty, full of generic copy your marketing team wrote without talking to support, or does not exist at all.

I built a full CX Ticket Help Center skill that processes hundreds of tickets, clusters them by topic, scores deflection potential, and generates complete help center systems with bot Q&A pairs and agent macros. But you do not need hundreds of tickets to get started. You need 10.

Here is the prompt. Pick one topic your team answers over and over, grab 10 tickets about it, and paste them in.

---

The Prompt

Copy everything below. Replace the bracketed sections with your actual data.

You are a CX content specialist for an ecommerce brand. I am going to give you 10 real support tickets about the same topic. Your job is to analyze the customer questions and the agent responses, then write a single help center article that answers the question so well that customers never need to open a ticket about it again.

BRAND CONTEXT:
- Store name: [e.g., "Vitality Supplements"]
- What you sell: [e.g., "Premium supplements for sleep, focus, and recovery"]
- Support platform: [Gorgias / Zendesk / Richpanel / Intercom / other]
- Tone of voice: [e.g., "Friendly, direct, no corporate speak"]

TICKETS (paste 10 tickets about the same topic - include both the customer message and the agent response for each):

Ticket 1:
Customer: [paste customer message]
Agent: [paste agent response]

Ticket 2:
Customer: [paste customer message]
Agent: [paste agent response]

[... repeat through Ticket 10]

FROM THESE TICKETS, GENERATE:

1. HELP CENTER ARTICLE
Write a complete, publish-ready help center article that includes:
- A clear, searchable title (what a customer would actually type into a search bar)
- A one-sentence TL;DR answer at the top (for scanners)
- Step-by-step instructions if the topic involves a process (returns, exchanges, account changes)
- Edge cases and exceptions your agents mentioned across the 10 tickets
- The exact language customers use to describe this problem (not internal jargon)
- A "Still need help?" footer with a link to contact support

Format the article in markdown with proper H2/H3 headings. Keep paragraphs short - 2-3 sentences max. Customers scanning a help center do not read walls of text.

2. QUICK-ANSWER VERSION (for chatbot)
Write 3 variations of how a customer might ask this question (different phrasing), plus a 2-3 sentence answer a chatbot could send immediately. Include a handoff trigger - when should the bot escalate to a human instead of answering?

3. AGENT MACRO
Write a templated response the support team can use for this topic going forward. Include [BRACKETS] for personalization variables like [CUSTOMER_NAME], [ORDER_NUMBER], [PRODUCT_NAME]. The macro should sound human, not robotic - use the tone from the best agent response in the tickets.

4. PREVENTION RECOMMENDATION
Based on the tickets, suggest one change to the website, product page, order confirmation email, or packaging insert that would prevent this question from being asked in the first place. Be specific - not "improve communication" but "add estimated delivery date to the order confirmation email subject line."

---

What You Will Get Back

Here is a condensed example from a supplement brand. The topic: "Where is my order?" (the number one ticket for almost every DTC brand).

Help Center Article (excerpt)

Title: Where Is My Order? How to Track Your Shipment

TL;DR: You can track your order using the tracking link in your shipping confirmation email. Most orders ship within 1-2 business days and arrive in 3-5 business days.

How to find your tracking link:

1. Check your email for a message from [Store Name] with the subject "Your order has shipped"
2. Click the tracking link in that email
3. If you do not see it, check your spam folder - shipping emails sometimes land there

My tracking says "pre-transit" or "label created": This means we have printed the shipping label but the carrier has not scanned the package yet. This is normal and usually updates within 24-48 hours...

Chatbot Q&A

Customer might ask:

• "Where is my order?"
• "I have not received my package yet"
• "My tracking number is not working"

Bot response: "Your tracking link is in the shipping confirmation email from us - check your spam folder if you do not see it. Most orders arrive in 3-5 business days after shipping. If your tracking has not updated in 48+ hours, reply here and I will look into it for you."

Escalate to human when: Tracking shows delivered but customer says they did not receive it. (This requires investigation, not a canned answer.)

Prevention Recommendation

"Add the tracking link and estimated delivery date directly to the order confirmation email - not just the shipping confirmation. 6 of the 10 tickets came from customers who ordered 2-3 days ago and had not received a shipping email yet. They were not actually late - they just could not find tracking info because it had not shipped yet. Setting the expectation in the order confirmation ('your order will ship within 1-2 business days and you will receive a tracking link by email') prevents the ticket entirely."

---

Why 10 Tickets Is the Magic Number

Fewer than 10 and you get an article based on one customer's edge case, not the common pattern. More than 10 and you are doing the work of the full CX skill - clustering, scoring, bulk generation - which handles 200+ tickets at once.

10 tickets about the same topic gives you:

• The 3-4 ways customers phrase the question (for your chatbot training and article title)
• The edge cases that come up 20-30% of the time (not just the obvious answer)
• The best agent response to use as a template (there is always one agent who nails it)
• Enough signal to spot the root cause (the prevention recommendation)

---

How to Pick Your First 10 Tickets

If you use Gorgias, go to Statistics > Top ticket tags. Your top tag by volume is your first article. Export 10 recent tickets with that tag.

If you use Zendesk, check Support > Explore > Tickets by topic.

If you do not tag tickets (you should start), ask your support team: "What question do you answer most often?" They will know immediately. Search your helpdesk for that phrase and grab 10.

The topics that deflect the most tickets are almost always the same for ecommerce:

1. "Where is my order?" - tracking, shipping times, delivery expectations
2. "How do I return this?" - return process, exceptions, refund timelines
3. "Which product is right for me?" - product selection, comparison, use cases
4. Subscription management - skip, pause, cancel, change frequency
5. "Does this work with...?" - compatibility, ingredients, contraindications

Write one article per topic. Five articles covering these five topics will deflect 30-50% of your ticket volume. I have seen this pattern on every DTC brand I have worked with - the top 5 topics account for half the inbox.

---

What the Full Skill Adds

This prompt handles one topic at a time. The CX Ticket Help Center skill in the DTC Stack processes your entire ticket export and:

• Clusters 200+ tickets into 20-30 topic groups automatically
• Scores each cluster by deflection potential (which articles will reduce the most tickets)
• Generates a complete help center - all articles, all bot Q&A pairs, all macros, in priority order
• Maps tickets to customer journey stages so you can see where in the buying process friction happens
• Writes proactive content recommendations for product pages, emails, and checkout that prevent tickets from being created

But 10 tickets and this prompt is where you start. Pick your worst topic. Paste the tickets. Publish the article. Check your ticket volume for that topic in 30 days. You will see the drop.

---

FAQ

What format should I paste the tickets in?

Plain text works fine. You do not need to format it perfectly. Just include the customer's message and the agent's response for each ticket. If the ticket had multiple back-and-forth messages, include the first customer message and the final resolution from the agent - that is usually the most complete answer.

Does this work with any helpdesk platform?

Yes. The prompt does not care where the tickets came from. Gorgias, Zendesk, Richpanel, Intercom, Freshdesk, even email threads you copy from Gmail. As long as you can paste the customer question and the agent answer, it works.

My support team writes inconsistent responses. Will the article be inconsistent too?

No - that is actually why this works well. The AI reads all 10 responses and synthesizes the best elements from each one. If Agent A gives great step-by-step instructions and Agent B catches an edge case that A misses, the article combines both. You end up with a response better than any single agent wrote.

How do I measure if the help center article actually reduced tickets?

Tag the tickets for this topic in your helpdesk before publishing the article. Check the count 30 days after publishing. For "where is my order" style questions, most brands see a 25-40% reduction within the first month - but only if the article is findable. Put it in your help center, link it from your order confirmation email, and make sure your chatbot surfaces it.