“Pause every keyword after 30 clicks without an order” is easy to automate, but it is not equally valid for every product. The correct threshold depends on expected conversion rate, CPC, selling price, margin, and the campaign’s purpose. Amazon defines conversion rate as conversions divided by the relevant audience. For PPC, brands commonly evaluate orders divided by clicks. If the expected conversion rate is 10%, one order is expected for every 10 clicks on average, but real results do not arrive on schedule.
1. Calculate the probability of zero orders
If each click has an independent conversion probability of p, the probability of receiving zero orders after n clicks is: Probability of zero orders = (1 - p)^n At a 10% expected conversion rate, zero orders after 30 clicks has a probability of:
That result is unlikely enough to justify a close review. At a 3% expected conversion rate, however:
- 90^30 = 4.2%
- 97^30 = 40.1%
Zero orders after 30 clicks is not unusual for a term expected to convert at 3%. A fixed rule would frequently pause such targets before the data becomes conclusive. To find the clicks needed for a 95% confidence threshold, use: Clicks = ln(0.05) / ln(1 - expected conversion rate) Rounded up, the threshold is about 29 clicks at 10% CVR, 59 clicks at 5% CVR, and 99 clicks at 3% CVR.
2. Add an economic stop before the statistical stop
Waiting for statistical confidence can cost more than the target is worth. Set a second limit based on allowable ad spend per order: Allowable spend per order = Selling price x Target ACOS Economic click limit = Allowable spend per order / Average CPC For a $40 product with a 25% target ACOS, allowable spend is $10. At a $1.25 CPC, the economic limit is eight clicks without an order. The brand may choose to reduce the bid or pause at that point even though the statistical evidence is weak, because the next conversion would need to recover excessive spend. For launch or ranking campaigns, the allowable spend may be higher. Define that exception before spending, not after the keyword exceeds the normal target.
3. Use a two-threshold decision rule
- Calculate a statistical threshold from the expected conversion rate.
- Calculate an economic threshold from target ACOS and CPC.
- Use the lower threshold as the review point.
- Review relevance, placement, search-term intent, and listing conversion before pausing.
4. Amazon Brand Management Blog Collection 15
A relevant exact-match term with strong click-through rate may deserve a lower bid instead of a full pause. A loosely related search term can be negated earlier because relevance is already weak. Product targets should be judged against the targeted ASIN’s price, rating, review count, and compatibility. Also account for attribution delay and low sample sizes. Do not make repeated daily changes to the same target. Use a consistent review cadence and preserve enough history to distinguish a poor target from normal variance. The best pause rule combines probability with economics. Statistical evidence tells you how surprising the result is. Unit economics tells you how long the brand can afford to wait.
5. Apply thresholds at the right level
Use search-term data for customer queries and targeting data for the keyword or product target that triggered them. A broad keyword can look acceptable while one matched query consumes most of the waste. Conversely, a target can appear inefficient because several weak queries sit beneath it. Negate the poor query before pausing the entire target when the remaining traffic is relevant. Keep branded, non-branded, competitor, and product-target campaigns under separate thresholds because their conversion rates and strategic roles are different.
