I'm saying that I believe it's silly to suggest Google is monitoring *sales* and then enforcing a constant weekly number of *sales* for a website.
I believe Google collects as much data as they can from the original search to the final user action just like we all do. Google's own search data can be combined with all their other data collection methods (Chrome, Analytics, YouTube, APIs) to determine what that user does when they leave Google search. And if a sale is made, Google would log that if they could. With most common ecommerce platforms being consistent in their sale success pages, I don't think it would be hard for Google to do. There's also Gmail where order confirmations may be sent to as well. Combine order confirmations with tracking numbers sent, Google could determine how responsive some sellers are when orders do arrive, much like Amazon does. Whether Google is using this data to intentionally manage traffic is questionable, and we probably will never know.
Regarding your #2 point. A good number of us were/are in Adwords. When zombies hit, I was the highest bidder with good quality scores on a number of buyer intent keywords. An increased prominence of ads being displayed surely would have raised the CPC, but total sales would not have dropped for the day because the more targeted ads would still be displayed. Getting back to the baseline, which is the average daily sales originating from Google pre-September for YEARS, I could always count on an average number of sales coming from Google. This floor on sales, which I had experienced for YEARS was shattered. Beginning in September daily sales on zombie days were few to none and on good days I was lucky to return to what I had seen for YEARS up to August despite logging many more combined visitors from organic and paid Google placement. If Google increased the number of paid shopping/ad widgets, yes my CPC would have risen as the result of absorbing mismatched clicks, but what about the appropriately matched targeted traffic and the sales they produced? #2 does not carry any weight with what I see. Increased clicks and a slight increase in traffic on zombie days resulted in a day with few or no sales from Google AT ALL.
From what is known about RankBrain, Google has said it is the third most important ranking factor for about 15% of the queries Google has not seen before. I've been in business for a while, and the buyer intent keywords/keyphrases are well known to Google. Google even suggests them in Adwords and has for a long time. If RankBrain is being used on searches Google has not seen before, as Google has said, I doubt it would apply to my site. Therefore, point #3 about RankBrain learning does not seem fitting. Is it possible that Google released RankBrain to manage more queries? Absolutely. But if this is how RankBrain handles queries, then the AI has split personalities by sending good traffic one day and crap traffic for the next 2+ days. So a broken RankBrain would apply, if it is responsible.
I sum up everything I see as Google doing it intentionally, which has led to an economic loss by me. Others are losing too. But where are the winners? I don't see many people talking about doing better and in fact these zombies are the main topic of discussion in most Google SEO threads here and elsewhere. With no winners posting, is this loss getting ready to spread throughout the rest of the digital economy?