TL;DR: Deciding when to make changes in Google Ads is about statistical reliability, not conversion count. Based on $20M+ in managed ad spend, we use a 2x-3x click threshold: if a segment has not received two to three times the clicks needed for one expected conversion, the data is not ready to act on.

Metric	Value
Example campaign conversion rate	~5% (1 conversion per 20 clicks)
Minimum clicks before considering exclusion	40 (2x threshold)
Clicks for high-confidence exclusion	60 (3x threshold)
4am-5am at 27 clicks, 0 conversions	Too early to act
4am-5am at 46 clicks, 0 conversions	Exclusion justified (2x threshold met)
Virginia at 70 clicks, 0 conversions	Excluded
Massachusetts at ~60 clicks, 1 conversion	Kept (near expected performance)

When to Make Changes in Google Ads: Why 27 Clicks Is the Wrong Sample Size (And What Is Right)

Knowing when to make changes in Google Ads is not about spotting a segment with zero conversions. It is about knowing whether those zero conversions are statistically meaningful or just the result of not enough data.

Here is what happened when we brought in external Google Ads managers to audit one of our accounts. Two freelancers independently flagged the same segment: the 4am to 5am time slot. Both recommended excluding it from the ad schedule on the grounds that it had produced zero conversions. What they missed was that the slot had 27 clicks.

At a 5% conversion rate, 27 clicks represents just over one expected conversion cycle. In the real world, 27 clicks had barely completed a single statistical round. Cutting the slot at that point would have been noise-driven pruning dressed up as optimization.

This post is based on a video Peterson published on the Creekside Marketing YouTube channel: Optimizing Google Ads: When to Make Changes for Better Performance.

Why Zero Conversions Does Not Always Mean What You Think

Zero conversions in a segment does not indicate underperformance unless the segment has had enough clicks to statistically expect at least one. At a 5% CVR, a segment with 12 clicks has not completed a single statistical cycle. That zero is not a signal. It is an insufficient sample.

Zero looks definitive in the Google Ads interface. It sits in the conversions column with no ambiguity. What the platform does not tell you is whether that zero is the result of a broken segment or a segment that has not had enough opportunities to produce a result in either direction.

According to Creekside Marketing’s analysis across accounts managing $20M+ in paid search spend, over-optimization on thin data is one of the most common structural problems in Google Ads management. It does not look like a mistake. It looks like active account management. Every exclusion was based on real data. The problem is the data was not large enough to support the conclusion.

The result compounds quietly over time: fewer active time slots, fewer states, narrower keyword lists. Each individual cut looked defensible in isolation. Together, they leave the algorithm less room to find volume that would have been available with more patience.

How to Decide When to Make Changes in Google Ads: The Click Threshold Framework

Start with your campaign’s conversion rate and calculate expected clicks per conversion. At 5% CVR, expect one conversion per 20 clicks. Your 2x threshold is 40 clicks. Your 3x threshold is 60 clicks. Any segment below 40 clicks does not yet have enough data for an exclusion decision.

Work from campaign-level data, not account-level. Performance Max, Search, and Shopping campaigns have meaningfully different conversion dynamics, and blending them into one view distorts the baseline. Pull the last 30 days for the specific campaign you are evaluating.

Once you have the baseline:

2x threshold (40 clicks at 5% CVR): The segment has had two full expected conversion cycles with no result. The pattern is starting to look intentional. Consider an exclusion, but extend to a 60-day window to confirm it holds.

3x threshold (60 clicks at 5% CVR): The segment has had three full expected conversion cycles with no result. This is a high-confidence signal. Exclusion is defensible on the data.

If your CVR is lower (for example, 2%), the math scales accordingly: 1 conversion per 50 clicks, 2x threshold at 100 clicks, 3x at 150. Lower CVR campaigns require more time and budget before optimization signals become reliable. That is not a problem with the campaign. It is how statistics work.

The 2x Rule Versus the 3x Rule: Which Threshold to Use

Both thresholds are defensible exclusion standards, but they fit different situations. Use 2x (40 clicks) when you need to make decisions quickly and have confirmed the pattern across a 60-day window. Use 3x (60 clicks) when the campaign is still in a learning phase or when secondary metrics show mixed signals.

Use the 2x threshold when: the segment shows a clear directional negative across all metrics (high CPC, low CTR, zero conversions), you have extended the date range to 60 days and the pattern holds consistently, or the account is high-spend and decisions need to move faster.

Use the 3x threshold when: the campaign is newer and still feeding Google’s algorithm with learning data, the segment has secondary signals that are borderline (CTR near account average, CPC close to expected), or conversion cycles in your industry tend to be longer than average.

The practical reason the 3x standard is more reliable for most situations: a segment that needed 60 clicks to finally convert once is still performing at 1.6% CVR in a 5% CVR campaign. That is two conversions short of expectation, not zero. The 3x window catches those situations before you exclude a segment that was one or two conversions away from meeting your threshold.

Time Slot Exclusions: A Real Example from a 5% CVR Campaign

The same 4am-5am time slot looked entirely different across two date ranges. At 27 clicks and 0 conversions in a 30-day window, the slot had not yet completed one statistical cycle and should not be excluded. At 46 clicks and 0 conversions over 60 days, it had crossed the 2x threshold.

30-day view: 27 clicks, 0 conversions. Expected conversions at 5% CVR: 1.35. Clicks needed for 2x threshold: 40. Status: 13 clicks short of the minimum. Decision: do not exclude. This is the data the two freelancers acted on. They were wrong.

60-day view: 46 clicks, 0 conversions. Expected conversions at 5% CVR across that window: 2.3. The slot has now had more than two full cycles and still produced nothing. The pattern is no longer statistical noise. Decision: exclude.

The same segment, two different conclusions, based on whether the sample was large enough to mean anything. The 30-day view produced a false signal. The 60-day view confirmed the pattern. This is why we check both windows before making any schedule-based exclusion: the shorter window identifies candidates, the longer window confirms them.

Location Exclusions: Three State Decisions, One Framework

Virginia, South Carolina, and Massachusetts each produced a different decision in the same account using the same threshold framework. Virginia and South Carolina both reached 70-plus clicks without a single conversion, crossing the 3x threshold clearly. Massachusetts reached approximately 60 clicks with one conversion, indicating near-expected performance and no reason to exclude.

Virginia: 70 clicks, 0 conversions. At 5% CVR, the state should have produced at least three conversions by this point. Three full cycles with zero conversions is an unambiguous negative signal. Excluded.

South Carolina: Similar profile to Virginia. Clicks above the 2x threshold, zero conversions, no secondary metrics suggesting the audience exists. Excluded.

Massachusetts: Approximately 60 clicks, 1 conversion. The state was expected to generate two to three conversions at that level of spend. One conversion puts it one short of expectation at the low end. The cost per conversion is higher than the account target, but cutting the state entirely for being one conversion short would eliminate a market that is still developing. The single conversion is evidence the audience exists there. Kept.

The Massachusetts case is the one most managers get wrong. One conversion in 60 clicks at a 5% CVR account looks like underperformance when you look at the CPA column. But it looks like a segment that is one conversion away from meeting expectations when you look at the conversion count relative to the threshold. The framework prevents you from cutting that state based on CPA alone when the underlying data does not support exclusion.

For a deeper look at how we analyze location performance across multiple campaigns at once, see our guide on the AI method for finding underperforming Google Ads locations.

Why Google Ads Management Rewards Patience Over Precision

Premature exclusions do not just remove underperformers. They remove the segments Google’s algorithm is still evaluating and narrowing its search for conversions within. An account that aggressively prunes on thin data ends up with targeting constraints that compound over time, reducing the optimizer’s ability to find volume that would have been there with more runway.

Google’s algorithm is designed to get conversions. That is how the platform retains advertisers. The algorithm explores time slots, locations, and keywords to find where conversions occur, then concentrates spend there. When you exclude a segment before enough data has accumulated, you are not trimming waste. You are cutting off exploration the algorithm had not finished.

This does not mean running everything indefinitely and never making exclusions. It means the exclusions you make should be based on sample sizes large enough to distinguish a real underperformer from a segment that simply has not had enough opportunities. When something has crossed the 3x threshold with zero conversions, the pattern is clear. For everything else, the better optimization is to wait.

Google Ads can be a game of patience. That is not passivity. It is the mechanism. The accounts that perform best over 12 months are the ones that exercise restraint on exclusions in the early phases and make cuts only when the data is unambiguous.

For a broader look at how optimization mistakes constrain account performance, see our post on the most costly Google Ads misconceptions we find in every audit. And for the longer story on why performance compounds when campaigns have room to run, see why Google Ads ROI compounds over time rather than delivering instant results.

Frequently Asked Questions

When should you make changes in Google Ads for time slot exclusions? Apply the 2x-3x click rule: calculate expected clicks per conversion from your campaign’s CVR, then wait until a time slot has accumulated at least twice that number with no conversions before excluding. For a 5% CVR campaign, that means 40 clicks minimum, 60 clicks for a higher-confidence call. Always confirm across both a 30-day and 60-day window before acting.

What is the right date range for Google Ads optimization decisions? Start with 30 days for most decisions. If the segment has not crossed the click threshold in 30 days, extend to 60 days. For lower-traffic campaigns, 90-day windows may be necessary. The goal is sample size, not a fixed calendar window. A 30-day view that has not reached 2x the click threshold is not ready for a decision regardless of what it shows.

Can you have too little data to make a Google Ads exclusion? Yes. Excluding a segment before it has had two to three times the expected conversion opportunities is acting on statistical noise, not a real trend. Zero conversions in a 12-click sample at 5% CVR is not an underperformer. It is a segment with less than one full statistical cycle. Treating it as a signal leads to exclusions that restrict the account’s targeting for no valid reason.

Does this framework apply to Performance Max campaigns? Performance Max has fewer direct exclusion controls than Search or Shopping, and its algorithm is more self-correcting by design. The click threshold logic still applies to the exclusion levers available: location exclusions and campaign-level negative keywords. For PMax, the threshold question is relevant to the targeting decisions you do control, even though the campaign type operates differently overall.

What if cost per conversion is high even though the click count is low? High CPA on a thin sample is expected and not reliable as an exclusion signal on its own. If a state has 15 clicks and one high-cost conversion, the CVR might be fine and the CPA is inflated entirely by sample size. Wait for the click count to cross the 2x threshold before drawing conclusions from CPA. A single conversion on 15 clicks at any price does not tell you enough to act.

---

Ready to Audit How Your Account Is Being Optimized?

Premature optimization is one of the hardest problems to spot from inside an account because each individual exclusion looks like a decision, not a mistake. We offer a no-cost account review that identifies where campaigns have been restricted on thin data and what that has cost in volume.

Claim your free $10k profit audit

---

About the Author

Peterson Rainey is the founder of Creekside Marketing, a paid advertising agency managing $20M+ in Google Ads and Meta Ads spend annually. He works with e-commerce brands, service businesses, and local practices across the U.S. and writes about paid advertising based on real account data, not theory.