Recently, a performance marketer set up an A/B test comparing Google's Performance Max (PMax) with a Standard Shopping campaign. The results were baffling: PMax spent almost nothing throughout the day, only to aggressively burn through the majority of its daily budget between 11:00 PM and midnight—resulting in zero conversions.

While this might look like a system glitch, it is actually a classic behavioral pattern driven by Google's algorithmic logic. In this article, we will break down why this "budget dumping" happens and how you can prevent it from draining your ad spend.

Understanding the "Budget Pacing Catch-up" Phenomenon

In performance marketing, we refer to this behavior as Budget Pacing Catch-up. Simply put, Google's PMax algorithm operates under a core directive: spend your allocated daily budget by the end of the day.

Early in the day, if the algorithm detects high competition or predicts a low conversion rate (CVR), it bids conservatively to preserve budget for better opportunities. However, as the clock winds down to 10:00 PM or 11:00 PM, the system realizes a significant portion of the daily budget remains unspent. At this point, the algorithm's priority shifts from "finding high-ROI users" to "spending the remaining budget."

To burn through the budget quickly, Google often pushes ads into low-cost, high-inventory, non-search placements, such as the Google Display Network and lower-quality YouTube placements. This explains why campaigns experience a sudden spike in clicks with absolutely zero conversions—the traffic is simply low-quality filler.

Think of it like a student frantically rushing to finish their summer homework on the very last night. The goal is no longer quality; it is simply filling in the blanks before the deadline.

Why This Is Worse in A/B Tests and Small-Budget Campaigns

This late-night budget dump is particularly prominent during A/B tests or when launching new campaigns with limited budgets. There are two primary reasons for this:

  • Data Scarcity: When traffic is split or budgets are small, PMax lacks the historical data needed to make confident bidding decisions early in the day, causing it to hesitate and hoard budget until the final hours.
  • Attribution Cannibalization: If you run a strong Standard Shopping campaign alongside PMax, the Standard Shopping campaign may capture the majority of high-intent search queries. Left with no room to scale on Search, PMax is forced to spend its remaining budget on the Display Network late at night.

4 Actionable Strategies to Stop Late-Night Budget Bleeding

If your PMax campaigns are draining your budget in the final hours of the day, implement these four optimization strategies:

1. Set a Strict Target ROAS (tROAS)

Do not give the algorithm budget without constraints. If you use the "Maximize Conversion Value" bidding strategy without a target ROAS, Google will accept traffic at any cost just to spend your budget. Setting a realistic tROAS acts as a guardrail. Even if the algorithm wants to dump budget late at night, it will refrain from bidding if the predicted return falls below your threshold.

2. Implement Ad Scheduling

While PMax is highly automated, you still control when your ads run. If your data shows that conversions rarely occur during the late-night hours, use ad scheduling to exclude the 11:00 PM to 12:00 AM window. If the system cannot learn to spend efficiently during these hours, physically restrict its ability to serve ads.

3. Avoid Aggressive Testing During the Learning Phase

PMax is a "black box" that relies heavily on data. When data is scarce—especially in new campaigns—performance can be highly volatile. We recommend giving PMax an independent budget and a longer observation window of at least 2 to 4 weeks to allow the algorithm to stabilize.

4. Use Google Ads Scripts for Real-Time Monitoring

For advanced advertisers, Google Ads Scripts can monitor hourly spend. You can write a script that detects sudden spikes in hourly spend without corresponding conversions and automatically pauses the campaign or sends an alert to your team.

Conclusion

Performance Max is a powerful tool, but it is not a set-it-and-forget-it solution. To prevent it from wasting your budget on low-quality traffic, you must establish clear boundaries using tROAS constraints, negative keywords (which can be added via Google support), and precise audience signals.