Harvest Readiness
Is this search term ready for exact match?
Review a converting term before giving it an exact-match test. Compare its estimated conversion rate with break-even and see how the campaign baseline affects the result. The model’s assumptions remain visible alongside the calculation.
See how it decidesScroll the chart sideways if it does not fit.
Short answer: this checker flags a term for harvest review when its model probability of positive ad-attributed profit at today’s CPC reaches your threshold, 90% by default. Its default campaign weighting is an uncalibrated assumption, not measured luck. It does not predict whether moving the term to exact match will increase profit.
Check a whole report
Paste your search terms, get a harvest list.
Copy four columns from your Search Term Report: search term, clicks, orders and spend. All rows use the profit and campaign baseline above, so check one product and comparable discovery traffic at a time.
One term per line, tab or comma separated. A header row is skipped.
| Search term | Verdict | Chance profitable | Conv. rate | Adjusted | Next step |
|---|
Share these results with your team
A short, plain summary for Slack or email. It’s written in your browser; nothing is sent anywhere. Edit it before you copy if you like.
When has a search term earned its own exact-match keyword?
A search term costs $1.20 a click and each order earns $12 before ads, so it needs a 10% conversion rate to break even. The campaign it came from also converts at 10%. After 3 clicks and 1 order (33%), the tool blends that with the campaign average to 12%, and the chance the term is profitable is 60%. After 15 clicks and 3 orders it is 73%, after 30 clicks and 6 orders 87%, and after 40 clicks and 8 orders 92%, which clears the 90% line and flags the term for exact-match review. Use the controls below to try your own clicks, orders and cost per click.
How it’s calculated
Model and assumptions
One order from three clicks gives an observed 33% conversion rate, with little evidence about repeat performance. This model combines that observation with a campaign baseline. The weighting is adjustable and has not been calibrated against historical Amazon data.
Zero orders so far? That’s a different question: When to Negate tells you whether to cut it. Need each product’s profit before ads? Run the Catalog Profit Check first, then pick the product here.
Break-even conversion rate
Cost per click ÷ profit per order before ads. At a $1.20 click and $12 profit per order, a term must convert above 10% to make money.
Adjusted conversion rate
Combine the campaign baseline, weighted as k clicks (30 by default), with the term’s observations: (k × campaign rate + orders) ÷ (k + clicks). The assumed baseline weight is k ÷ (k + clicks): 91% at 3 clicks, 67% at 15 and 33% at 60. These are model weights, not measured percentages of luck.
Chance it’s profitable
The term’s true rate follows a Beta distribution with α = k × campaign rate + orders and β = k × (1 − campaign rate) + clicks − orders. The tool reports the chance it sits above break-even.
The verdict
Harvest now is a review flag at or above your model threshold. A suggested base bid assumes fixed bidding and no adjustments; it does not guarantee CPC. Wait is between the thresholds, with a conditional extra-click scenario. Don’t harvest is at or below 100% minus your threshold at today’s CPC.
Worked examples
Same conversion rate, different verdicts.
Each row assumes $12 profit per order, a $1.20 click (so break-even is 10%), a 10% campaign rate and the default settings.
| Clicks | Orders | Conv. rate | Adjusted | Chance profitable | Verdict |
|---|---|---|---|---|---|
| 3 | 1 | 33.3% | 12.1% | 60% | Wait; scenario: ~339 extra clicks |
| 15 | 3 | 20.0% | 13.3% | 73% | Wait; scenario: ~115 extra clicks |
| 60 | 12 | 20.0% | 16.7% | 97% | Harvest now at $1.20 |
| 80 | 3 ($1.50 click) | 3.8% | 5.5% | under 1% | Don’t harvest |
The same 20% conversion rate is a “wait” at 15 clicks and a review flag at 60. The common “sales above zero” rule treats all four terms the same way.
Limits
What it can’t tell you.
The baseline weighting and probability threshold are starting assumptions. The model has not been calibrated against later Amazon outcomes. Its recommendations support review and do not establish that harvesting causes additional profit.
Use settled data
Exclude recent dates whose attribution window has not ended, and allow for report updates. Use click-attributed purchases from the full window shown in your report.
Ad-attributed profit only
Some harvested orders would have happened organically anyway. Branded terms are the most affected.
Today’s cost per click
The chance is worked out at the term’s current cost per click. An exact-match bid that pays more per click raises the bar.
Timing, not overlap
Whether to also add the term as a negative in its discovery campaign is a separate decision, and there isn’t good evidence either way yet.
Comparable traffic and purchases
Use a baseline from comparable discovery traffic, excluding this term where possible to avoid counting its evidence twice. The model assumes stable CPC, conversion and profit per purchase, and at most one converting purchase per click. It cannot model multiple purchases per click, mixed product margins or changing auctions from these totals alone.
Extra clicks are a scenario
The curve adds fractional expected orders at the current model estimate. Actual future orders are uncertain. The displayed click count is not a forecast, a guaranteed stopping time or a recommendation to spend that budget.
For developers and AI agents
Use it from code.
The same calculation is a free JSON API. No key needed. The full spec is at /api/openapi.json.
curl "https://profitfirstppc.com/api/harvest-readiness?clicks=60&orders=12&spend=72&profit_per_order=12&baseline_cvr=0.10"To check a whole Search Term Report, POST up to 500 terms at once:
curl -X POST https://profitfirstppc.com/api/harvest-readiness \
-H "content-type: application/json" \
-d '{"defaults": {"profitPerOrder": 12, "baselineCvr": 0.10},
"terms": [{"searchTerm": "insulated water bottle", "clicks": 60, "orders": 12, "spend": 72},
{"searchTerm": "water bottle with straw", "clicks": 9, "orders": 1, "spend": 10.80}]}'Each result has a verdict (harvest_now, wait, do_not_harvest or no_evidence), a one-sentence headline, probabilityProfitable, estimatedCvr, luckShare, extraClicksNeeded and suggestedStartingBid. The other free tools have APIs too: see llms.txt.
FAQ
Harvesting questions
How many clicks before moving a search term to exact match?
There’s no fixed number. It depends on how far the term converts above break-even. In the examples above, a term converting at twice break-even needed about 60 clicks to be a 97% model probability at the observed CPC. A term converting just above break-even can need hundreds.
Should I harvest a search term after one order?
One order alone gives limited evidence. The decision also depends on CPC, profit per purchase and a comparable baseline. At the default weighting, the baseline contributes 91% of the adjusted estimate after three clicks; that does not mean 91% of the performance is known to be luck.
What is the break-even conversion rate for Amazon PPC?
Cost per click ÷ profit per order before ad spend. At a $1.20 click and $12 profit per order, a search term must convert above 10% to make money.
Should I also negate the term in the auto campaign?
That’s a separate decision from timing, and there isn’t good published evidence either way yet. This tool only answers whether the term has proven itself enough to harvest.
Tell me what you want your ads to do.
In a free 30-minute call, we’ll discuss your business, your current advertising and whether an audit, a focused project or ongoing management would help.
Learn the decision behind the numbers.
Work through the example, then apply this tool to your own reports.