Voice of customer product insights: detecting new demand in calls and chats

Short answer

To detect new product demand from calls and chats, collect as a separate field the things customers repeatedly ask for that you do not have: "what did the customer ask for that we don't offer?". Then put those requests through four tests: frequency (how often), segment (who asks), seriousness (would the customer buy because of it) and persistence (a one-off wave or steady demand).

Conversations give a strong demand signal because customers state their need in their own words without being asked. But they do not prove market size or that customers will actually pay. So the result is a product hypothesis, not a product decision.

Why conversations are a good source

In surveys you show customers options you prepared, and they choose among them. In conversations customers write for themselves: "do you have a children's size?", "can I make the subscription a family one?", "do you come and install it?". Such questions can reveal demand you never thought of.

A further advantage: these customers have already contacted you, so the interest comes from your real audience, not random respondents.

Demand vs curiosity

Not every question is demand. "Do you have online classes?" is very different from "if you had online classes I'd sign up right away, I can't come in person". The first is curiosity; the second is demand — the customer says they are not buying because the thing is missing.

• Curiosity: the customer asks, and when told "no", carries on with another product.
• Conditional demand: the customer says "I'd buy it if you had it", but carries on with the current product.
• Hard demand: the customer refuses, or goes to a competitor, because the thing is missing.

Recording these three levels as separate field values is the most important information for a product decision.

Which fields you need

1. 1What was asked forA closed list: the 8–12 products, features, sizes or service types customers ask for most today, plus "other". Build the list by reading.
2. 2SeriousnessCuriosity / conditional demand / hard demand / unclear.
3. 3SegmentChannel, product category, new or existing customer — often not a field but data the conversation already carries.
4. 4OutcomeDid the conversation end in a sale, was another product bought — this comes from the CRM.

Four tests

1. 1FrequencyIn how many conversations a month does the request appear? Is the share rising steadily? A one- or two-week wave may be a campaign or season effect.
2. 2SegmentWho asks? If it comes from one channel, region or product category, it may be a small niche or the effect of a marketing message.
3. 3SeriousnessWhat share is hard demand? Are customers refusing, or just asking?
4. 4PersistenceDoes the request hold for three months? Steady new demand is a basis for a product decision.

A demand evidence card

For every request sent to the product team, prepare a one-page card. It shortens debate because number and evidence sit in one place:

• The request: one sentence, in the customer's language.
• Period, number of conversations analysed, coverage rate.
• Frequency and share, the three-month trend.
• Seriousness split: curiosity / conditional / hard.
• Segment: channel, category, new or existing customer.
• 5–10 links to evidence conversations.
• Unknowns: price sensitivity, market size, technical feasibility.

What conversations cannot tell you

• Market size: conversations only see those who write to you.
• Willingness to pay: saying "I'd buy it" is easy; paying is harder.
• Silent demand: people who know you lack the product and never write.
• Competitors' offers and prices: customers may mention them, but accurate data must come from elsewhere.

How to complement a conversation hypothesis with open market data is covered in customer inquiry analysis and the articles on market research.

From hypothesis to decision

1. 1A small testInstead of a full product, a minimal version: pre-orders, a waiting list, a pilot service.
2. 2Go back to the same customersShow the new offer to the customers who expressed the demand (where you have permission) and measure the response.
3. 3IndicatorPre-order count, pilot sales, a fall in the refusal share.
4. 4Decision dateWrite down in advance when the test ends and which number means "go".

Illustrative example

This is an illustrative example. A flower delivery service sees the value "corporate subscription" in the "what was asked for" field grow slowly over three months. Most conversations are from office managers who want to order flowers for the office every week. In the seriousness field, hard demand is low: customers still order one at a time.

The decision is not a big product but a small test: a simple form is added to the website, and a weekly offer is prepared by hand for offices that send a request. Two months later, the number of offices that switched to a subscription is measured.

Typical mistakes

• Treating a few loud customers' wishes as market demand.
• Looking at frequency without measuring seriousness.
• Accepting the sales team's "customers ask every day" as a counted number.
• Never reading requests that land in "other" — new demand most often shows up there.

A rhythm with the product team

Demand signals should be discussed with the product team once a month. Before the meeting, an analyst prepares evidence cards for the three fastest-growing requests. In the meeting each card gets one of three decisions: test, keep watching, reject. A rejected request is not deleted — it keeps being tracked, because demand can strengthen later and seeing it needs a comparable history.

The demand field in Vexvon

In Vexvon the company builds fields such as "what was asked for" and "seriousness" in its own panel; AI picks from the customer's words in each conversation and keeps the sentence as evidence. You can ask the panel's AI assistant for a value's trend, a two-period comparison or a cross-tab by channel, and open the list of conversations behind each value. Frequently repeated phrases that land in "other" are shown separately — new demand often starts there. More: Vexvon analytics.

Next step

Find 30 conversations from last month where the answer was "we don't have that" and write down what the customer wanted. If two or three requests repeat, turn them into a field and start measuring seriousness. We can build the demand field on your own conversations in a demo.