Hub › Lessons › Content planning, lesson 1
Turn customer questions into a month of content ideas
The hardest part of planning content is often deciding what to write about. Your customers have already told you. The questions they ask by email, in messages and on calls are the questions other people like them are asking too. This lesson shows you how to collect those questions and use one prompt to turn them into four weeks of topics, written in your customers' own words.
What you will be able to do: paste 10 to 20 real customer questions into one prompt and get back the questions grouped by what people are really asking, a four-week plan at your own publishing pace, and a list of the facts only you can add. If you do not have enough questions to fill the month, the prompt tells you instead of making topics up.
Step 1: Collect the questions (10 minutes)
- Look where customers already write to you: emails, chat messages, enquiry forms, comments on your posts and your notes from sales or discovery calls.
- Copy each question as it was asked. "Can I update the menu myself?" is more useful than your tidy summary "CMS training", because it is how the next customer will phrase it too.
- Include repeats. If three people asked about price in different words, paste all three. Repeats are the signal the prompt looks for.
- Aim for more questions than posts. At 2 posts a week you need 8 topics for four weeks. Because some questions will be repeats, start with at least 12.
- Number them, one per line. The numbers let you check that none went missing.
Before you start: protect your data
- Remove names, phone numbers, email addresses, order numbers and anything else that identifies a customer. Keep only the question.
- If a question includes private details ("my landlord raised the rent, can I still afford you?"), rewrite it in general terms ("can I start with a smaller package?").
- Check your AI tool's data settings, including whether chats may be used for training.
Step 2: The content planning prompt
Fill in the first three lines, then paste your numbered questions where it says [paste].
You are helping me plan a month of content from questions my customers actually asked. About me (one line: what I do and who for): How often I can publish (for example "2 posts a week"): Formats I can make (for example "blog post, short video, email, carousel"): Customer questions, one per line, copied as they were asked (names and personal details removed): [paste] Write, in this order: 1. Question groups: group questions that are really asking the same thing. A table: Group | Question numbers | What they are really asking (one sentence in plain words). Mark any group with 2 or more questions "Asked more than once". 2. Content plan: one topic per group, enough topics to fill four weeks at my publishing pace. A table: Week | Topic (a working title in the customer's own words) | Format (one of my formats) | Angle (one line: what the reader will be able to do or decide afterwards) | Answers questions. Put groups marked "Asked more than once" first. 3. What I need to add: for each topic, list the facts, examples or numbers that only I can supply to answer it well, written as [ASK: ...]. Do not fill these in. 4. Leftovers: any question that does not fit a topic, and why. Rules: - Use ONLY the questions I pasted. Do not invent questions, statistics, customer stories or results. - Every question number must appear in exactly one group. - Keep each customer's wording in the working titles where you can, so the titles match how people ask. - If I give fewer topics than I need for four weeks, say so and stop. Do not pad the plan with topics nobody asked about. - Keep it under 300 words, not counting tables.
Worked example
We ran the prompt with Claude on 24 October 2026. The business and the questions are made up for illustration. Your results will differ by tool, model and input.
Input (illustrative)
About me (one line: what I do and who for): I build simple websites for small cafes and home-based food businesses in Singapore. How often I can publish (for example "2 posts a week"): 2 posts a week Formats I can make (for example "blog post, short video, email, carousel"): blog post, short video, carousel Customer questions, one per line, copied as they were asked (names and personal details removed): 1. How much does a basic website cost per year, not just to build? 2. Do I still need a website if I already have Instagram? 3. Can customers order and pay on the website or only see the menu? 4. How long does it take before the site is live? 5. Will my website show up on Google Maps? 6. What do I need to prepare before we start? Photos, menu, logo? 7. Can I update the menu prices myself without calling you? 8. What are the yearly costs after the first year? Domain, hosting, all that. 9. Is Instagram enough or should I really pay for a site? 10. How do I get more Google reviews? 11. Can I take my website elsewhere if I stop working with you? 12. My phone photos look bad. Is that ok for the site or do I need a photographer?
Output (as returned)
1. Question groups
Question groups, as returned. On a phone, scroll the table sideways.
| Group | Question numbers | What they are really asking |
|---|---|---|
| Total yearly cost (Asked more than once) | 1, 8 | What will I pay each year in total, including after the build? |
| Instagram vs website (Asked more than once) | 2, 9 | Is Instagram enough, or is a website worth paying for? |
| What to prepare (Asked more than once) | 6, 12 | What must I hand over, and are my phone photos good enough? |
| Online ordering | 3 | Can customers order and pay on the site? |
| Time to launch | 4 | How long until my site is live? |
| Google Maps | 5 | Will my site help me show up on Google Maps? |
| Updating the menu | 7 | Can I change prices myself? |
| Leaving you | 11 | Can I take my site with me if I stop working with you? |
| Google reviews | 10 | How do I get more Google reviews? |
2. Content plan
The four-week plan, as returned. On a phone, scroll the table sideways.
| Week | Topic | Format | Angle | Answers questions |
|---|---|---|---|---|
| 1 | "How much does a basic website cost per year, not just to build?" | Blog post | Budget the full yearly cost before deciding | 1, 8 |
| 1 | "Do I still need a website if I already have Instagram?" | Carousel | Decide if Instagram alone is enough for your business | 2, 9 |
| 2 | "What do I need to prepare before we start? Photos, menu, logo?" | Carousel | Gather everything needed, and know if phone photos will do | 6, 12 |
| 2 | "Can customers order and pay on the website or only see the menu?" | Short video | Choose between a menu-only site and one that takes orders | 3 |
| 3 | "How long does it take before the site is live?" | Short video | Plan your launch date | 4 |
| 3 | "Will my website show up on Google Maps?" | Blog post | Know what the site does and doesn't do for Maps | 5 |
| 4 | "Can I update the menu prices myself without calling you?" | Short video | Know what you can change on your own | 7 |
| 4 | "Can I take my website elsewhere if I stop working with you?" | Blog post | Know what you own before you sign | 11 |
Eight topics fill four weeks at 2 posts a week, so nothing is missing.
3. What I need to add
- Yearly cost: [ASK: your build price, and each yearly cost (domain, hosting, other) with amounts] [ASK: what the first year includes vs later years]
- Instagram: [ASK: what a website does that Instagram can't, from your experience] [ASK: a real example, if you have permission to share it]
- Prepare: [ASK: your checklist of what you need] [ASK: your photo standard, and when you recommend a photographer]
- Order and pay: [ASK: which ordering and payment options you offer, and any extra cost]
- Time to launch: [ASK: typical timeline and what causes delays]
- Google Maps: [ASK: what you do for Maps, if anything, and what the owner must do themselves]
- Update menu: [ASK: how owners edit prices, and whether this is included]
- Leaving: [ASK: who owns the domain and content, and how handover works]
4. Leftovers
Question 10 (Google reviews) has no slot. Four weeks are already full, and it's about reviews, not the website. It could replace a topic later if you want to cover it.
What we checked
- Nothing went missing. A short script confirmed that each of the 12 question numbers appears in exactly one group, and that the plan has 8 topics, two in each of the four weeks.
- The real repeats were found. Questions 1 and 8 both ask about the yearly cost, and questions 2 and 9 both ask whether Instagram is enough. Both pairs were grouped and put first.
- One group needs your judgement. Questions 6 and 12 were grouped and marked "Asked more than once". Question 12 (are phone photos good enough?) is related to preparing for the project, but it is not the same question. If photos come up often for you, give them their own post.
- Nothing was invented. Every working title is a customer's own question, and no prices, timelines or results were filled in. Those are left as [ASK] items for you.
When you do not have enough questions
We ran the prompt again with only 5 of the questions: two pairs of repeats and one other. It found 3 groups and stopped: "Your questions give 3 topics (one per group), but 2 posts a week for four weeks needs 8. Filling the other 5 slots would mean writing about things nobody asked, so I haven't written the content plan, the [ASK] list or the leftovers." It then offered two options: send more questions, or plan just those three topics. That is the behaviour you want. A plan padded with topics nobody asked about defeats the point of the exercise.
Step 3: Check the plan before you use it (5 minutes)
- Count the numbers. Every question number should appear in exactly one group. If one is missing, ask the AI where it went.
- Read each group. Are the questions really asking the same thing? Split any group that mixes two different worries, as with questions 6 and 12 above.
- Keep the customer's wording in your final titles where it reads well. You can tidy the grammar, but keep their words.
- Answer every [ASK] yourself before you write the post. Prices, timelines and examples must come from your own business, not from the AI.
- Save the plan where you plan the rest of your work, with a publish date for each topic.
Watch out for
- One customer is not a trend. A question asked once can still be a good post, but a question asked by several different customers is a stronger choice. The "Asked more than once" mark helps you see the difference.
- This shows what customers ask, not what people search for. It is a good starting point, but it does not tell you how many people search for a topic. A later lesson in this series will cover checking a planned title against real search results.
- Grouping can go wrong both ways. AI can merge questions that share words but not meaning, or split the same worry asked in different words. Step 3 is how you catch both.
- Keep the questions. Add new ones to the same list each week. Next month's plan is quicker when the list is already there.
Next step
Next: a one-page content brief
Lesson 2 of this series takes one topic from your plan and turns it into a short brief: who it is for, the question it answers, the points to cover and the facts you need to supply. If your customer messages are scattered and unsorted, sort customer feedback into themes first.
Go to lesson 2Keep the plan in Notion
If you already use Notion, a simple database with a publish date and a status works well for a content plan. Lesson 6 of the money admin series shows how to build a database with AI-written formulas and test them, and lesson 4 of this series builds the content calendar itself.
See the content calendar lessonHow we checked this lesson
- On 24 Oct 2026 (Singapore time) we ran the prompt above, filled in with the illustrative input shown, using Claude Code 2.1.310 in print mode. The output is shown as returned, with formatting adjusted for this page.
- We checked the output with a separate script: every question number appears in exactly one group, and the plan has 8 topics across four weeks. We read each group and noted the one we would split.
- We ran the prompt a second time with only 5 questions, to check that it stops instead of padding the plan. It stopped and explained why.
- We tested one assistant on two inputs. AI output varies by tool, model and input, so check the plan each time.
- Last reviewed: 24 October 2026.