Customer Needs Analysis from Conversations: What Do Your Customers Ask For Most?
You do not need a survey to find out what your customers want most. The answer is already in your WhatsApp conversations, but you have to count it properly. Collect needs in the customer’s own words, group similar phrases, count customers rather than messages, and keep evidence next to every topic. This guide walks through it step by step with an illustrative example.
How needs show up in a chat
Customers do not always state their need directly. In a chat it appears in three forms, and you need to collect all three.
- Explicit requestThe customer says exactly what they want: “Do you have a winter snowsuit for a three-year-old?”, “Do you deliver outside the city?”. This is the easiest form to spot, and usually the only one teams collect.
- Problem descriptionThe customer describes a situation, not a product: “My child grows so fast I have to buy new clothes every season”, “the last size I ordered was too small”. The need is hidden here: sizing advice, a model with room to grow.
- Comparison criterionThe customer says what they look at when choosing: “It has to be cotton”, “can I return it?”. This is the quality side of the need: not what they want to buy, but how they want it to be.
Many teams collect only the first form and lose the other two. Yet problem descriptions often point to new product or service ideas, and comparison criteria show how product descriptions should be written.
Grouping topics
The phrases you collect will be scattered. When you sort them into topics, follow three rules.
- Synonyms“Do you ship to other towns?”, “is there delivery to Manchester?”, “regional shipping” are the same need: delivery outside the city. Merge them into one topic, but keep the original phrases.
- Separate needs“When will delivery arrive?” and “is delivery free?” start with the same word but are different needs: one is about speed, the other about cost. Merge them and you will not see which one is the problem.
- Unclear meaningSome phrases are ambiguous: “Don’t you have another option?” Colour, size, price? Do not group these by guesswork. Keep them in an “unclear” category. If that category is large, your team is not asking clarifying questions.
Choosing a counting unit
The most common mistake in a needs map is the counting unit. There are three options, and each gives a different result:
- Messages: if one customer asks the same question five times, it counts five times. This inflates one customer’s concern rather than showing how widespread a need is.
- Conversations: one customer may have several separate chats. This gives a result slightly higher than the customer count.
- Customers: how many different customers expressed the need at least once. The most honest unit for showing how widespread a need is.
Recommendation: use customer count as the main measure, and keep message count separately as “intensity”. A customer asking the same thing again and again means they were not answered, or the answer did not satisfy them. That is a service problem, not a need.
Turning the map into decisions
The map below is an illustrative example: one month of chats at an online children’s clothing store. Each topic has four fields: number of customers, who it shows up with, evidence and the next test.
- Delivery to other townsCustomers: highest. Who: customers outside the main city. Evidence: “Do you ship to Leeds?”, “how many days to Bristol?”. Test: show regional delivery clearly on product pages and watch whether the number of questions changes.
- Sizing adviceCustomers: high. Who: first-time buyers. Evidence: “Which size for a four-year-old?”, “the last one was too small”. Test: a simple size chart by age and height.
- Return policyCustomers: medium. Who: mostly those looking at pricier items. Evidence: “Can I return it if it doesn’t fit?”. Test: show the return policy next to the price.
- Larger sizesCustomers: few, but every time the answer was “we don’t have it”. Who: ages 10–12. Evidence: “Do you have size 146?”. Test: try a small batch. The requests are few, but the demand is going unmet.
- UnclearCustomers: medium. Evidence: “Don’t you have another option?”. Test: coach the team to ask “another option in what way?”.
Notice there is no “market demand” on the map. These are the needs of customers who wrote to you. What customers who never wrote, and went to another store, wanted is not visible here. That is why each topic has a test next to it, not a decision.
Who uses the map, and how
A needs map should not be one person’s report. The same information leads three teams to different decisions:
- MarketingCustomers’ own words are ready-made material for ad copy, a website FAQ and product descriptions. If “how many days to other towns?” comes up often, the answer should appear in your ads and on product pages.
- Product and purchasingRequests that get “we don’t have it” every time are a signal for your range. The number of customers may be small, but unmet demand is visible. Testing with a small batch is the safer way to decide.
- Sales managerThe most repeated questions are candidates for ready answers. The “unclear” category is a coaching topic: it shows where reps are not asking clarifying questions.
Update the map once a month and compare it with the previous one. Needs change with the season: winter clothing in autumn, delivery speed in summer. If a topic stays high for several months in a row and has not been tested, it should be first on the agenda at your next meeting.
Competitor names will show up in your needs map too, so keep them separately using competitor mentions in sales conversations. If you want to write questions for the map, 15 business questions to ask your chats gives ready examples. And before building a first map on older history, record the coverage of your first audit.
How to check your own chats with Chatdue
For every chat, Chatdue automatically extracts what the customer asked for, a summary, objections and competitor mentions. You can create your own analytics fields to sort needs into categories like the ones above: “delivery”, “sizing”, “returns”. Voice messages, images and documents are converted to text and included.
You can export filtered chats with all AI fields to Excel and build the map by customer count. Because older conversations are loaded too, your first map can cover several months of history.
Pull your customers’ recurring needs from your own chats with Chatdue. The 30-day free trial analyses 1,000 messages, and no card is required.
Frequently asked questions
- Is the most frequently mentioned request the most important?Not necessarily. Frequency shows how widespread a need is, not how valuable it is to the business. A need that few customers raise, but that goes unmet every time, may be a bigger opportunity. Always read frequency together with “were we able to answer it?”.
- How should I count one customer’s repeated messages?In the needs map, count each customer once. Record their repeat messages separately as intensity or as a service problem. The same customer asking the same thing three times usually means the first two answers did not satisfy them.
Analyze your WhatsApp conversations
Connect your number with a QR code — the first 30 days are free.
Start free