Ginny Marvin, the Google Ads Liaison, confirmed some time ago that for the Google Ads machine learning models to work faster (and probably better) Google would need a minimum of 15 conversions in the last 30 days. She added that the more conversions help, when Julie Bacchini said that you really need 50 to 100 conversions if you want this method to outperform manual strategies.
I believe she was quoting a Google presentation when talking about how to use the “Golden Triangle.” Julie explained this in a lot more detail over here which was from this Google Ads experts series recording.
Here is the golden triangle:
Julie said you should “expect a big push for smart bidding + broad match + RSA” this year.
For that to work, you need more and more data and the ones that will do the best are those that have the most data.
It’s true that more conversion data helps inform the machine learning models faster. We recommend having a minimum of 15 conversions in the last 30 days, while being clear that more conversion data (including micro-conversion signals) helps. #ppcchat
— AdsLiaison (@adsliaison) January 18, 2022
Forum discussion at Twitter.