What I Was Trying to Accomplish
We get a lot of customer questions.
Most of them are repeats. "How do I reset my password?" "Where can I find the invoice?" "Can I change my plan?" The same questions, over and over. We answer them in emails, in chat, in support tickets. But we don't have a FAQ.
I wanted to fix that. I wanted to take 50 recent customer questions and turn them into a searchable FAQ. Not a perfect FAQ. Just a starting point.
I didn't want to learn to code. I didn't want to build a custom tool. I wanted a no-code workflow that I could run in an afternoon.
Here's what I built.
What I Used
Tools: A general-purpose AI assistant (free tier) and a spreadsheet (Google Sheets or Excel)
Input: 50 customer questions, exported from our support inbox
Time available: About 1 hour
Goal: A searchable FAQ draft
The Workflow
The workflow has five steps.
Step 1: Collect. I exported 50 customer questions from our support inbox. Just the questions, not the answers, not the customer names.
Step 2: Clean and anonymize. I removed names, order numbers, and anything personal. I replaced them with placeholders like [CUSTOMER] and [ORDER].
Step 3: Cluster and categorize. I pasted the questions into the AI tool and asked it to group them by topic.
Step 4: Draft answers. I asked the AI to draft a short answer for each cluster.
Step 5: Build the FAQ. I pasted the questions and answers into a spreadsheet. One column for question, one column for answer, one column for category. Then I made it searchable.
Five steps. About an hour.
The Prompts
Here are the exact prompts I used.
Prompt 1: Cluster and categorize
"Here are 50 customer questions. Group them into 5-10 categories. For each category, give a short label and list the questions that belong to it. Do not answer the questions yet. Just group them."
Prompt 2: Draft answers
"For each category, write a short answer (under 75 words) that addresses the most common question in that category. Use plain language. Do not promise anything we cannot deliver. If the answer depends on the customer's account, say 'contact support for account-specific help.'"
The Result
Here's what the workflow produced.
Categories:
Account & Login
Billing & Invoices
Plan Changes
Technical Issues
Data & Privacy
Other
Sample FAQ entries:
Question: How do I reset my password?
Answer: Go to the login page and click "Forgot password." Enter your email address. We'll send you a link to reset it. If you don't see the email, check your spam folder.
Question: Where can I find my invoice?
Answer: Log in to your account and go to Billing. Your invoices are listed by date. You can download them as PDFs.
Question: Can I change my plan?
Answer: Yes. Go to Account Settings and select "Change Plan." You can upgrade or downgrade at any time. Changes take effect at the next billing cycle.
Question: Why is the app running slowly?
Answer: Slow performance is usually a connection issue. Try restarting the app and checking your internet connection. If the problem continues, contact support with your device and operating system.
Question: How do I delete my account?
Answer: Go to Account Settings and select "Delete Account." This action is permanent. Contact support if you need help exporting your data first.
Searchable format: I pasted the questions and answers into a spreadsheet. I used the spreadsheet's search function to make it searchable. It's not a website. It's not a polished FAQ. But it works.
What the Workflow Does Well
It's no-code. I didn't write a single line of code. I used an AI tool and a spreadsheet. That's it.
It's fast. Collecting, cleaning, clustering, drafting, and building took about an hour.
It's organized. The AI grouped questions into categories I might not have seen. It found patterns in the noise.
It gives a starting point. Instead of staring at 50 questions, I have a categorized, answerable FAQ draft.
It's repeatable. I can run this workflow every month. Same prompts. Same process. Updated FAQ.
What Still Needs My Attention
It missed context. The AI didn't know our product. It didn't know our policies. Some answers were generic. I had to fix them.
It missed the hard questions. Some customer questions don't have simple answers. The AI didn't know which ones were sensitive. I had to flag those myself.
It missed the priority. Some questions are asked more often. Some are more urgent. The AI treated all questions equally. I had to prioritize.
It missed the brand voice. The answers were fine. But they didn't sound like us. I had to rewrite some of them.
It missed the accuracy check. The AI doesn't know if an answer is correct. I had to verify every answer with our product team.
The Math
Without the workflow: Writing a FAQ from scratch would have taken about 4 hours.
With the workflow: Collecting, cleaning, clustering, drafting, and building took about 1 hour.
Time saved: About 3 hours. Not 4. But the worst 3 hours—the staring, the sorting, the blank page.
What I'd Change
I'm still refining the workflow. Here's what I'm considering.
Adding a priority column. Something like "How often is this asked?" That would help me know what to answer first.
Adding a sensitivity column. Something like "Is this a sensitive topic?" That would help me know what to review carefully.
Adding a "contact support" rule. Some questions can't be answered in a FAQ. I want to flag those automatically.
Using a better search tool. The spreadsheet search works, but it's not great. I'm testing a few no-code FAQ tools.
My Question for You
Have you built a FAQ from customer questions?
What does your workflow look like? Do you use AI? Do you use a spreadsheet? Do you use a no-code tool?
I'm curious what works for other people. I've been using this workflow for a few weeks. It's not perfect, but it's better than answering the same question 50 times.
