How to find low-quality search terms in Google Ads

A search term that converts once at four hundred dollars cost looks fine to the auto-bidder. It is not fine. It is a term consuming spend at four times your target cost-per-acquisition, and the auto-bidder will keep feeding it because one conversion looks like signal.

These are low-quality search terms in the cost-per-conversion sense. They are the third pattern in the wasted-spend pillar and the one most operators never look at because the standard search-terms-report sort (by cost descending) hides them.

The pattern

Low-quality terms share three properties:

1. Cost is meaningful but not extreme: $40-200 per term over thirty days. Below the threshold where it grabs attention sorting by raw cost.
2. Conversions exist but are sparse: one conversion every other month rather than one per week.
3. Cost-per-conversion is way over target: 2-5x the account's target CPA.

The auto-bidder sees the conversion and keeps assigning budget. The human sorting by cost sees a moderate-cost row and skips it. The leak persists for months.

How to find them

The standard sort hides this pattern. Switch to a different sort order.

The judgment call: phrase block or full block?

The hardest decision in low-quality term cleanup is whether to fully block a term that has occasionally converted. Two factors to weigh.

Conversion rate. If the term has converted at less than one percent over a meaningful sample (say 100+ clicks), the conversions are likely noise. Block.

Conversion value. Sometimes a low-conversion-rate term has high per-conversion value. A wholesale buyer who occasionally orders ten thousand dollars of product might convert at zero point five percent but be worth keeping. Check the conversion value column before blocking.

The default leans toward phrase blocks rather than full blocks for the borderline cases. Phrase blocks scope the leak without killing the rare valuable conversion.

A worked example

Imagine a small B2B SaaS account with a target CPA of $80 and an account-average conversion rate of 3%. Pulling the search terms report sorted by cost-per-conversion shows:

The first row is the low-quality term. It converted, so the standard sweep would have left it alone. Cost-per-conversion is $284 against target of $80. three and a half times over budget. The bidder is feeding it because of that one conversion.

The right move on row one: check the conversion value. If the value is in line with average, phrase-block "small teams" or "for small teams" to push the matcher toward more focused queries. If the conversion was a one-time fluke (no follow-up activity), full block.

What to do about Performance Max here

Performance Max makes this harder because the search categories report does not show individual queries. You cannot see the per-term cost-per-conversion the way you can for Search.

Workarounds for PMax:

• Compare campaign-level CPA to your other campaigns. If PMax is 2x your target CPA, it has low-quality term issues you cannot see directly.
• Tighten search themes to push the matcher toward better queries.
• Add brand exclusions for competitor terms that may be eating budget.
• Consider splitting high-intent queries out of PMax into a regular Search campaign for cleaner control.

How often to do this sweep

The cost-per-conversion sweep is monthly territory; weekly is too tight. The terms it surfaces are sparse-conversion, so a longer time window is needed for the data to be meaningful.

A practical cadence:

• Once a month, alongside the regular cost-descending sweep.
• Use a 60-day window for the conversion-cost sort.
• Apply changes, wait three weeks, then re-check.

Done consistently, this sweep recovers an additional 10-20% of monthly spend on top of what the standard sweep catches. It is the most overlooked piece of regular Google Ads hygiene.