Below is an express diagnostic scheme that can actually be done in one short monitoring cycle.
1) We take the correct data slice
Upload statistics at least 7-14 days in advance. Set the attribution window to "one week" so that purchases that arrive with a delay are included in the calculation. A short interval helps you see overspending faster, and two weeks reduce the impact of random days.
2) We count the DRR for each article
DRR = advertising expense, revenue × 100%.
If there is an expense according to the article, but there is no revenue, the position falls into the red zone automatically. This is not a verdict on the product, it is a signal that the advertisement is working without demand confirmation or the card is not putting the squeeze on the conversion.
3) We set a “safe ceiling” DRR from marginality
Without margin, any optimization becomes a guessing game. Set a control threshold as part of your marginality. A practical guideline: A safe DRR should leave a margin for logistics, fees, and refunds.
If the margin is 30%, then the working DRR often lies in the range of 20-25%. Values above 30% require intervention, because profits begin to disappear at the article level.
4) Sorting solves 80% of the problem
Sort the positions by DRR in descending order. The top of the list is where your candidates for stopping or lowering rates are. Here it is important to look not at emotions, but at numbers: how many rubles are spent per unit of revenue.
5) We check the leftovers and the delivery plan
A high DRP on a product that is running out often means that you are buying traffic at a time when you cannot close the demand. If the stock is less than 3-5 days of sales and the replenishment has not been confirmed, you should turn off advertising immediately. This is a quick way to stop the budget drain.
6) We keep new items as a separate rule
New cards receive an adjustment period. Give them a minimum test corridor in terms of time and budget, then evaluate the dynamics: whether the conversion rate is growing, whether orders are appearing, whether the DRR is decreasing. If the expense increases, but there are no signs of learning, the test ends.
7) The decision on each “red” article
- DRR is higher than marginality and there are almost no orders → stop advertising.
- The DRR is high, orders are coming → rate reduction and control for 3-5 days.
- New position in the test → budget limitation and dynamic monitoring.
The main principle is that you need to manage at the article level, not by the average cabinet numbers. One or two products can "eat up" a monthly profit, as long as the overall metric looks tolerable. Introduce regular monitoring once a week, and the budget will start working as a growth tool, not as a loss item.
