They don’t have to collect them as in saving them, they just have to collect information about activity that would interest marketers who buy the data. “Searched for dishwashers”… “Ordered dishwasher”…
Then the marketers could process the data and go, Lovable Sidekick searched for dishwashers but didn’t buy one, maybe he’s still looking, let’s send him ads for dishwashers. Or, Lovable Sidekick bought a dishwasher, let’s not waste ad impressions showing him dishwasher ads, let’s show ads for dishwasher detergent. I mean, if I were part of that whole circus that’s how I would try to approach it.
But to do that, Google would need to collect and save save and process every URL you go to. It would need to snoop not only that you looked at the dishwasher, but that you clicked “add to basket” and then “order” and then completed the order without ever removing it from the basket. That means analysing not just the pages you visit but also the underlying requests that control the basket and order process.
There’s nothing it can do more minimally, and as far as I know it doesn’t do this.
Yes, it does seem like a lot of processing. Now google any word and see how fast it returns 200,000 results. Consider that Google delivered the same experience to millions of other people at that same time. Then consider what size drop in what size bucket the processing you described really amounts to.
It can process all of that easily on its servers. But there should then be evidence of this very large quantity of data being exported out of Chrome and uploaded to Google, which I don’t believe there is.
There are some other difficulties, too: no two shopping platforms encode “user completed the order” in the exact same way, so performing that analysis is actually quite hard and not nearly 100% accurate, even if you can get the complete browsing data.
They don’t have to collect them as in saving them, they just have to collect information about activity that would interest marketers who buy the data. “Searched for dishwashers”… “Ordered dishwasher”…
Then the marketers could process the data and go, Lovable Sidekick searched for dishwashers but didn’t buy one, maybe he’s still looking, let’s send him ads for dishwashers. Or, Lovable Sidekick bought a dishwasher, let’s not waste ad impressions showing him dishwasher ads, let’s show ads for dishwasher detergent. I mean, if I were part of that whole circus that’s how I would try to approach it.
But to do that, Google would need to collect and save save and process every URL you go to. It would need to snoop not only that you looked at the dishwasher, but that you clicked “add to basket” and then “order” and then completed the order without ever removing it from the basket. That means analysing not just the pages you visit but also the underlying requests that control the basket and order process.
There’s nothing it can do more minimally, and as far as I know it doesn’t do this.
Yes, it does seem like a lot of processing. Now google any word and see how fast it returns 200,000 results. Consider that Google delivered the same experience to millions of other people at that same time. Then consider what size drop in what size bucket the processing you described really amounts to.
It can process all of that easily on its servers. But there should then be evidence of this very large quantity of data being exported out of Chrome and uploaded to Google, which I don’t believe there is.
There are some other difficulties, too: no two shopping platforms encode “user completed the order” in the exact same way, so performing that analysis is actually quite hard and not nearly 100% accurate, even if you can get the complete browsing data.