Key takeaway

Reviews do not show reputational noise, but operational risk of repeat sales.

A synthetic model links feedback topics, scores, share of criticism, and potential money trail across network points.

Average rating

4,18

for all public reviews

Negative

22%

1-2 stars and strong complaints

Critical Topics

timing, quality, price, communication, availability

Loss potential

RUB126m

scenario estimate, non-accounting fact

Which breaks trust more often

topic model

Negative Feedback Topics

Timing and waiting

31%

Quality of work

27%

Price and transparency

20%

Communication

18%

Availability and logistics

13%

Positive Feedback Topics

Staff assistance

34%

Quick solution

28%

Quality of result

24%

Recording convenience

13%

Hazard Points

branch prioritization

Facility	Evaluation	Negative	Risk index	Main topic	Action
Cluster A-03 large flow	3,42	41%	78	timing and communication	Customer Response SLA before Order Completion
Cluster B-11 high average check	3,58	36%	73	price transparency	record approval of additional works
Cluster C-07 stable base	3,91	27%	61	result quality	call back after disputed cases
Cluster D-18 low risk	4,64	8%	22	staff and assistance	use as benchmark scripts

Timing31%

Most often, the client is dissatisfied not with the expectation itself, but with the lack of a clear status.

Price20%

Negativity intensifies when the final amount looks unexpected, even if the work has been done.

Quality27%

Repeated appeals after the result directly hit the willingness to recommend the point.

Synthetic examples of wording

“First they promised one term, then they postponed it without explanation.”

“The final price became clear only at the end.”

“The problem was solved, but I had to remind myself several times.”

What to add to the management agenda

01. A separate KPI for status messages to the customer.

02. Unified pre-work price approval template.

03. Analysis of 7 red dots on the texts, and not only on the average score.