Could the difference between Buyer and Nonbuyer simply based upon their past browsing habits or classifications? I do believe it goes deeper than simple Info / Buying intent. But I do think that is mainly triggered when certain key phrases are used. Such as "How" etc.
Based on SEchecker's notes on user behavior. "Zombies" do behave differently than your average user. And it should definitely be so because its a different subset of visitor profile. It would actually weird if your Zombies behave all the same as the regular average visitors. Maybe I can expand my theory with my previous car example.
Site A: Good for 18 year old who loves cars
Site B: Good for 20~30 year old who loves cars
Let's say your site is Site B, you can get healthy volume of traffic who purchase pretty regularly. But once in a while Google throws different visitor profiles into the site. In an extreme example, we will say 5 year olds who love cars. 5 year olds would look at the cool images, but never buy, and most of them exit fairly quickly. None of them ever really sign up for anything.
Google throws that subset of users at both Site A and Site B periodically, and find that none of them sticks. Thus Google determines that both Site A and Site B is bad for 5 year olds who love cars. These subset group of users (5 year olds) would definitely behave differently. But both sets would get thrown with these visitors periodically. It is google's Crowdsourcing capability at force, they do not have to physically "check" the sites, they simply use the user's behaviors.
I think we would be able to decipher more if there is a way to figure out exactly the user categories of those Zombie traffic. I think from webmasters point of view, we can only figure out what they do, and how they do, our biggest task to decipher the puzzle is to figure out "who they are". I think we have already determined "why" these zombies may exist in Google's crowdsourcing algorithm. Zombies traffic do behave differently as in these traffic never convert, but they are there.
As far as why some sites like Amazon rises prominently on the SERP, my guess is that as long as a subset group of users use the service. Google eventually determines a certain subset, personalization, or localization of users is good for that particular service. No matter how illogical or incorrect it may be.
You may never have to use Amazon yourself, but if let's say 80% of the users in your area, ISP service, or hell even browser version, find a particular website useful for a query - You may be grouped with that particular subset. We can say that Google can "incorrectly" determine - if you are a Chrome user who's age 25 and search for widget, there is a 90% chance you'll like Amazon for that specific query. So even if your 10% of user that absolutely hate Amazon, Google will still serve Amazon to you.
I have read experiences of websites reporting temporary traffic gains after significant marketing pushing - periods after the pushing period. It may be that website has attracted visitor profiles not registered in Google before. Thus based upon that user input, Google throws in additional mixture of users to see their behaviors. And this may be why the traffic increase is always temporary.