You're right to question if there's a "hidden learning period" or "throttling."
Here's a concise breakdown of what's likely happening:
Yes, there's absolutely a learning period for Performance Max, and it's longer and more sensitive than other campaign types.
PMax is designed to find the absolute best-performing combinations of assets, audiences, and placements across all of Google's inventory (Search, Shopping, Display, YouTube, Gmail, Discover). When you launch it, especially replacing Smart Shopping, it needs time to learn, and any significant change can reset or extend that learning phase.
Why your spending likely stopped (the "throttling"):
Extended/Reset Learning Phase:
The initial strong ROAS and impressions (the first 2 weeks): This is often a "honeymoon" period where PMax is aggressively testing to gather data. It might be spending on broader, cheaper placements to quickly collect signals.
The sudden drop: PMax has likely collected some initial data during those first two weeks. Now, it's moved into a deeper "learning" or "refinement" phase where it's trying to optimize more precisely for your ROAS target. If it can't find enough high-quality opportunities to meet that target, it will drastically pull back spending rather than spend on low-performing impressions. It's essentially saying, "I can't find enough good conversions at your desired ROAS, so I'm not going to spend your budget unnecessarily."
ROAS Target Too Aggressive: If your Target ROAS is set too high (even if it was met initially), PMax might find it difficult to scale while hitting that target. It prioritizes the ROAS goal over spending the full budget.
Conversion Tracking Issues: If your conversion tracking has any glitches or inconsistencies (e.g., conversions aren't firing reliably), PMax loses its crucial optimization signal and will stop spending.
Asset Group Performance: Google rates your assets (images, videos, headlines, descriptions). If too many of your assets or asset groups are performing poorly, PMax might limit serving.
Market Dynamics: Competition in the auction can change. Other advertisers might have increased bids or launched new campaigns, making it harder for your PMax to compete at your current ROAS target.
What to do now:
Be Patient (mostly): The full learning phase for PMax can be 4-6 weeks, especially if you don't have a massive amount of conversion data. Avoid major changes during this time.
Check "Insights" and "Explanations": In your Google Ads account, check the "Insights" tab for your PMax campaign. Also, look for "Explanations" (usually a dotted blue line when you hover over performance metrics) – these can tell you why performance changed.
Review Your ROAS Target: If you have a Target ROAS, consider lowering it slightly (e.g., by 10-20%) to give the campaign more room to learn and spend. You can always raise it again once spending stabilizes and ROAS is good.
Ensure Conversion Tracking is Flawless: This is paramount. Double-check your Google Tag Manager or direct tag implementation for all relevant conversions.
Audit Asset Groups: Check the "Ad Strength" and performance ratings for your asset groups. Provide a diverse range of high-quality assets (especially videos, even if auto-generated for now) to give PMax more options.
Review Audience Signals: While not direct targeting, good audience signals help PMax learn faster. Make sure you've included all relevant customer lists and audience interests.
Check Product Feed (if e-commerce): For Shopping-focused PMax campaigns, ensure your Google Merchant Center feed is healthy, products aren't disapproved, and prices are competitive.
It's a common challenge with PMax. The system is designed to be highly automated and efficient, but that also means less direct control and a heavy reliance on its learning algorithms.