The generated response cannot be sent yet: 5 checkpoints

A response has been generated—and it already looks like a ready-to-publish post. All that remains is to click the button, and the question is closed. This is precisely the step at which teams most often lose control over quality.

A shopper asked about the package contents. The response came across as confident and appropriate in tone. However, the product listing includes only the main specifications: composition, color, and size. The answer to the question about what is included cannot be confirmed.

Such a response should not be sent. Not because of its style, but because it relies on an unverified fact.

Response automation becomes manageable not when text appears quickly, but when the team understands which inquiries can be closed using an agreed-upon framework, which require additional facts, and which should be handed over to a person.

This is a practical quality-control methodology for an online store manager, a marketplace specialist, and a customer service professional. It can be used regardless of which tool prepares the drafts.

Quality control starts with identifying the type of inquiry

The first question after generation is: what type of inquiry are you dealing with? The acceptable response format, processing speed, and responsible employee all depend on the answer.

As a practical approach, identify three groups.

A routine inquiry requires a response based on a predefined procedure. For example, a shopper asks where to find the product instructions. A response can be prepared using a template, provided the instructions are actually available in the listing or at the specified location.

A clarifying inquiry contains a question that requires a fact. For example: “Does the package include a second product?” or “When will my order be delivered?” The answer depends on the listing, order, delivery terms, or an employee’s note. If no source can be found, the wording must not be turned into a guess.

An inquiry that requires a person’s decision involves a complaint, return, disputed condition, conflict, or a decision that cannot be made using a general template. Here, even a well-drafted response remains a preliminary draft, not a final answer.

For example, a shopper writes: “A different product arrived. Refund my money.” A response based on general guidelines may include a polite formula, but it cannot decide whether to acknowledge an error, initiate a return, or what option to offer the shopper.

The approach is simple: decide in advance which types of messages the team will handle using templates, which require data verification, and which will always end with a response from the responsible employee.

Find the missing fact before drafting the final wording

The second control question is: what is the answer based on?

It may rely on one or more verified sources:

1. Product specifications and description.
2. Package contents.
3. Order status and details.
4. Delivery terms and available pickup options.
5. Return or exchange policies approved by your team.
6. A note from the employee who made the decision on the inquiry.

A missing fact is not simply an absence of data in the message. Sometimes a shopper asks a direct question, but the system contains no answer. Sometimes the necessary information is in another source, but it has not yet been cross-checked against the listing or order.

Verify claims literally. “The package includes two cups” must be supported by package-content data. “We will deliver tomorrow” must be supported by the order status and confirmed delivery terms. “You can request a return” must be supported by the policies the store actually applies.

If the fact is found, record it in the inquiry’s working record. If it is not found, do not fill the gap with more confident wording. It is better to acknowledge that the team is checking the information than to give the shopper an unverified detail.

Check the wording: the fact may be correct, but the response may still be ineffective

Even a verified fact can be turned into a poor response. Review not only the content, but also how well it answers the question.

First: does the text answer the inquiry? Sometimes the drafted response merely paraphrases a review without answering the shopper’s question.

Second: does the statement expand on the known fact? If it is confirmed that the package includes a thermal mug, there is no need to add that it is “completely leakproof” or “suitable for all types of coffee” unless those details are also confirmed.

Third: does the response make a promise the team cannot control? The wording “we will reimburse the cost immediately” assumes a procedure and employee authority. If no such decision has been made yet, the response should not promise an outcome before it has been approved.

Fourth: is the tone appropriate? Politeness should not obscure the substance of the response. The shopper needs to know what will happen, who will review the inquiry, and where they can follow up if clarification is needed.

Fifth: are the known fact and the assumption combined in a single sentence? If part of the response is confirmed and another part requires verification, separate them. This will allow the team to send the confirmed portion and refer the rest to the responsible employee.

Compare the finished text with the original inquiry. Make sure it has not omitted an important detail, narrowed the question, or created a new one.

Who approves the response: assign the procedure in advance

The fourth step is to determine whether the response requires the responsible employee’s approval.

The mere generation of text does not establish authority. Some inquiries require only an agreed-upon template. Others need an employee to verify the facts. Disputed situations require a decision that cannot be delegated to the wording.

A convenient framework for the team:

1. Routine inquiry with a verified fact—send using the agreed-upon procedure.
2. Clarifying inquiry—the responsible employee finds the fact or identifies it in a verified source.
3. Complaint or unusual situation—the responsible employee makes the decision before sending a response.
4. Insufficient data—send a message stating that the information is being checked, without inventing an answer.
5. No responsible employee available during business hours—the inquiry remains in the queue rather than being sent with unverified wording.

Separately document what happens to an unanswered inquiry: who can see it, how it returns to the workflow, and what counts as sufficient confirmation. It is best to establish these rules before the first stream of messages begins.

The team’s methodology is not an active Pricerio algorithm

The five-check framework is a team-managed process. It does not mean that Pricerio independently classifies every inquiry, verifies an employee’s authority, maintains an audit log, or sends a response after manual approval.

What has been confirmed is that Pricerio can use AI to draft responses to reviews, questions, and customer messages. The cost of this capability is not specified in the provided plan description.

Source: Pricerio’s official website — https://pricerio.ru

Therefore, the framework described here should be treated as a process requirement: which decisions the service prepares, where human involvement is needed, which checks a response must pass, and who is responsible for sending it. If your account includes additional control settings, their scope and operation should be verified separately.

Do not base your methodology on an unsubstantiated connection between response speed and product rankings. The research did not confirm that unanswered inquiries directly affect rankings or ratings. Rules and response deadlines for Wildberries, Ozon, and Yandex Market should also be checked separately against the marketplaces’ current documentation.

This does not eliminate quality control. Even if sending a response has no bearing on product promotion, an error involving a promise, package contents, or return terms remains an error in the eyes of the customer.

A template should demonstrate applicability, not merely sound confident

Simple test: can you fit a specific inquiry into the approved template and confirm every material statement?

If the answer is “yes,” the template can be used as part of the established procedure.

If the answer is “no, but only one fact is missing,” route the inquiry for clarification. The responsible employee needs a short list: what is known, what is missing, and what cannot be promised until the information is verified.

If the answer is “no, and the resolution depends on the circumstances,” route the inquiry to a person.

Example. A customer writes: “The second box is missing from my package.” If the package contents have been confirmed, the team can acknowledge the discrepancy and initiate the applicable procedure. If it is unclear what the order included, a photo of the packaging, a photo of the contents, or an operator’s record is needed.

Another example: “Can this product be returned?” The answer depends on the store’s current terms and the product category. The possibility of a return cannot be confirmed merely because the question appears routine.

A third example: a complaint about a delivery delay. A response can be drafted, but any promise of a specific date must be checked against the order status. If the date has not been confirmed, tell the customer that the team is checking the delivery and assign an owner to the inquiry.

These examples illustrate the limits of a template: it helps formulate a processed response, but it does not replace factual verification.

Manager’s checklist

Run through the workflow with several real inquiries from different categories. Complete the checklist for each one:

1. Has the inquiry type been determined?
2. Has the inquiry been classified as routine, requiring clarification, or requiring a person to make a decision?
3. Is the source of each material fact identified?
4. Do you have the necessary information about the product, order, package contents, delivery, or return?
5. Does the wording avoid assumptions and promises beyond what is known?
6. Is it clear who will verify the missing fact?
7. Has the person responsible for making a decision on a disputed inquiry been identified?
8. Can the response be sent without additional approval?
9. What will happen if confirmation is not received?
10. How will the team identify an unresolved inquiry and return it to the workflow?

If even one question remains unanswered, the procedure is not yet ready for automated sending. This is not a wording problem. It is a process-management gap.

Evaluate decisions, not just the number of responses

When the team connects response generation with sending, it is useful to look at more than the number of inquiries closed. The following questions are more informative: How many responses were sent without the missing fact? How many cases required human involvement? How many drafts were returned for review? How many inquiries had no assigned owner?

These indicators are presented as an example of a control system, not as ready-made Pricerio metrics. You can track them in your operating system independently of the response-drafting service.

Start with one approved workflow. Choose one routine question, one question requiring clarification, and one complaint. Run them through the five checks, document the decisions, and identify where the team is spending time. Then add the next category.

This is how AI prepares drafts while the business remains responsible for the content, review, and sending. This separation is what makes response automation manageable.