Choose search terms for an exact-match test.

The question

Two search terms convert at 20%. One has three purchases from 15 clicks; the other has 12 from 60. Would you give them the same confidence? The rates match, but the amount of evidence does not.

Adding a term to exact match can give you a place to manage its bid and budget more deliberately. Whether that move improves profit is a separate question. Start by deciding whether its current results justify a controlled test.

Work through the numbers

Both illustrative terms cost $1.20 per click, contribute $12 per purchase before ads, and come from comparable discovery traffic with a 10% purchase rate. Their break-even rate is 10%.

Harvest Readiness combines the term’s results with that campaign baseline. The default gives the baseline the weight of 30 clicks. For the smaller sample, the adjusted rate is (30 × 10% + 3) divided by (30 + 15), or 13.3%. For the larger sample, it is (3 + 12) divided by 90, or 16.7%.

The model sees more evidence for Term B. Its output is conditional on the baseline and model assumptions; it is not a forecast of what exact match will deliver. The default 30-click weight has not been calibrated against later Amazon results.

Try predicting the effect of raising CPC to $2.40. Break-even doubles to 20%. A term that looked attractive at $1.20 now needs a higher conversion rate. Moving a term does not carry its old economics into a more expensive auction.

A published case to examine

My apparel case describes a weekly loop of promoting converting search terms to exact match and reviewing nonconverting traffic. Its Sponsored Products snapshot for November 5–December 5, 2024 showed $244,202.34 in attributed sales, $21,776.43 in spend and an 8.92% ACoS.

The snapshot documents the account result and the reported workflow. It does not isolate the effect of exact-match promotion or validate the calculator’s probability threshold. Use it as an example of a repeatable review process, not proof that every harvested term improves profit.

Read the apparel case.

Try it on your numbers

1. Open Harvest Readiness. Enter a term’s settled clicks, click-attributed purchases and spend.
2. Enter contribution before ads per purchase. Use a campaign baseline from comparable discovery traffic, excluding the candidate where possible.
3. Keep the default 30-click baseline weight visible. Compare the observed rate, adjusted rate and probability against your chosen threshold.
4. For a candidate that meets the threshold, write a proposed test bid, budget and review date. Inspect placement and other bid adjustments before applying a suggested base bid.
5. If checking several terms, use one product and comparable traffic per batch so every row has an appropriate margin and baseline.

Open Harvest Readiness

What still needs your judgment

The model assumes stable CPC, conversion and contribution, with at most one converting purchase per click. Branded purchases may have happened without advertising. Mixed product margins and changing auctions need more context than these totals provide.

Extra clicks shown by the tool are a scenario that assumes future clicks convert at the model estimate. They are not a spending instruction or a guaranteed stopping point. Whether to add a negative in the source campaign is a separate routing decision; this calculator does not settle it. The suggested starting bid assumes fixed bidding and no placement or other adjustments.

Explain it to your team

“This term meets our review threshold at its current CPC. I want to test exact match with a budget, then compare settled results.”