Unanswered customer questions: what "I don't know" and "we don't have it" tell you

Short answer

The answers "I don't know", "I can't say exactly" and "we don't have that" given to customers are among the cheapest and least-used sources of business information. They point to three different problems: a knowledge gap (the answer exists but the employee or bot does not know it), a product or range gap (what the customer wants really is not available) and a policy gap (there is no rule, or it is unclear, so everyone answers differently).

Collect these answers as a separate field and split them by type, because each goes to a different team: knowledge gaps to the content owner, product gaps to product and purchasing, policy gaps to leadership. Then lost demand can be measured.

Why these answers are missed

A sales report does not show "we don't have that": the customer simply does not buy and the conversation ends. Nothing is recorded in the CRM because no lead was ever created. In a support report, "I don't know, I'll check and get back to you" can look like a successful conversation — the customer got a reply. Yet nobody may ever have got back to them.

So the clearest signal — something the customer wanted but did not get — lands in no report. It can only be extracted from the conversation itself.

Three kinds of negative answer

1. 1Knowledge gap"I don't know how many years' warranty this model has." The answer exists in the company but never reached the employee, the knowledge base or the bot. Fix: content, training, bot answer.
2. 2Product or range gap"Not in this colour", "we don't sell children's sizes", "we don't open at weekends". What the customer wants really is not there. It may or may not be fixable — but the decision should be made with data.
3. 3Policy gap"I don't know, I'll ask the manager" — on returns, discounts, credit or exceptions, no rule is written. Each employee answers differently, and customers feel treated unfairly.

How to set up the field

Two fields are enough. The first is "negative answer type": knowledge gap / product gap / policy gap / none. The second is "what was asked": a closed list — product or service features, sizes, opening hours, delivery area, payment method, and so on, plus "other". State explicitly in the instruction: "the employee or bot gave the negative answer; the customer asked for something and did not get it".

How to measure lost demand

What happened after the negative answer? Look at the conversation's outcome:

• The customer bought something else — demand was substituted, the loss is small.
• The customer said "then I don't need it" and left — a lost sale.
• The customer mentioned a competitor — a lost sale and a competitor signal.
• The customer asked "when will you have it?" — future demand, a waiting-list opportunity.
• After a knowledge gap the answer was never given — a service failure.

Match these outcomes with the CRM status or how the conversation ended. The absolute number matters less than the direction: which "we don't have it" most often ends in a loss?

Knowledge gaps: the cheapest fix

Knowledge gaps are usually the fastest problems to fix. The list of the most repeated "I don't know" questions becomes the knowledge base's work plan directly: each question gets an owner for the correct answer (product, finance, logistics), the answer is written and approved, then passed to employees and the bot. A month later, the negative-answer share for that question is measured.

Systematic maintenance of a support knowledge base is a separate topic; the focus here is lost sales. Building the topic list is explained in customer inquiry analysis.

Product gaps: not every "no" should be fixed

Adding everything customers ask for is neither possible nor necessary. But the decision should be made with data. Assess the most-asked "we don't have it" answers that most often end in a loss as new demand: frequency, seriousness, segment, persistence. The method is set out in voice of customer product insights. Sometimes the fix is not a product but communication: offering an alternative instead of a flat "no".

Policy gaps: write the rule

If "I'll ask the manager" repeats often, there is no rule. That is both a customer-experience and a risk problem: one employee gives a discount, another does not, and customers tell each other. Present the most repeated policy questions to leadership as a separate list and ask for a written rule for each.

An answer standard

The negative answer itself can be good or bad. A simple standard:

• When saying "we don't have it", offer an alternative if there is one.
• When saying "I don't know", say when and how the answer will come — and then actually answer.
• When saying "I'll ask the manager", give the customer a date.
• If it will be available later, take a note to notify the customer, with their consent.

Illustrative example

This is an illustrative example. An electronics store separates conversations with negative answers for a month. The largest group is knowledge gaps: "can this laptop's memory be upgraded?". The answer is in the technical documentation, but agents do not have it. The second group is a product gap: "instalments over 24 months". The third is policy: "can I return a product whose box has been opened?" — every agent answers differently.

Decisions: 15 answers to technical questions are added to the knowledge base; the loss from the 24-month instalment gap is measured and talks with the bank begin; the returns rule is approved as a one-page written policy.

Limits

• AI sometimes misreads a negative answer: "not yet, it arrives tomorrow" is not fully negative.
• The loss is an estimate: the conversation does not show whether the customer bought elsewhere.
• Negative answers should not be the sole basis for evaluating an employee — the gap is usually in the system, not the person.

The negative-answer field in Vexvon

In Vexvon the company builds fields such as "negative answer type" and "what was asked" in its own panel; AI picks from the list in each conversation and shows the relevant message as evidence. The panel separately counts conversations whose last message is from the customer and went unanswered. Question-and-answer pairs from conversations can be exported to Excel and used to expand the knowledge base. More: Vexvon analytics and the knowledge base.

Next step

Find 40 conversations from the last two weeks containing "we don't have", "I don't know" or "I'll ask", and sort them into the three types. Start with whichever type dominates. We can build the fields on your own conversations in a demo.