"Vignette" ads - these ads shows only on desktop when someone clicks from one page to another (like after clicking on an internal link).
Tolomen, my personal opinion here, as I'm not experienced with these particular ads.... Though these ads technically don't affect new page load speed, IMO they might affect a
user's "navigation continuity", as they are potentially interruptive. For purposes of discussion now, I'm assuming that too many ads, particularly too many too soon, will cause some users either to leave or not to come back to a site.
(Note: Also see my comments below, at the end of this post, regarding a 2019 thread, which suggests that while my thoughts about aggressive monetization are correct, the Google "vignette ads" are not likely to be the problem. This very much depends on your individual site and situation). As far back as the beginnings of Panda, we've discussed here that the Google Adsense team will sell more ads to a site than might be good for it in search quality or in user-experience. There's also a question about whether vignette ads leave "foot-prints" that are factored into search results in 2-3 weeks, the way, say, that too many above-the-fold ads did with Panda. Unless there's a direct and clear violation of Google guidelines, though, I think the changes you mention might take longer than 2-3 weeks to be reflected in the rankings. That's just a guess.
Possibly, if you had regular returning visitors who hadn't liked the vignette ads and therefore might be turned off by your site, you might see a more immediate dropoff than you would with new visitors. Do you monitor in analytics whether you are losing returning visitors or new visitors? The distinction might provide a clue.
With vignette ads, the impact would also depend on when a in a user's session the vignette ad experience occurs. Many search problems, IMO, can be attributed to attempts to monetize a site too quickly or too often.
I do know, without question, that the Google search team looks favorably at content that is helpful or interesting to a user... and it looks unfavorably upon content designed only to funnel traffic into an ad.... This is one of many factors that makes me believe that Ideal ad placement, therefore, is towards the end of an article... ie, after the user has stayed with the content for a while, and, thus informed by it, might be looking for further resources or more detail, or purchase options.
With regard to what you see in "Updates and Search Changes", be aware that the law of averages suggests that many people might be losing traffic at any one time, but not necessarily for the same reasons that you might be losing traffic... so I'd be careful about identifying with too many of the problems you read about in that thread.
You should always use site search here, before you post, to check out what's been discussed recently, or even way in the past. Even if your question is new and unique to you, exploring past discussions will give you a perspective on what the most useful way is to frame your question.
Using site search, btw, I found an excellent discussion from 2019 in the AdSense forum, which relates closely to the vignette ads, their expected performance, etc... and you should definitely check that out here...
Auto-Ads Revisited July, 2019 https://www.webmasterworld.com/google_adsense/4953465.htm [webmasterworld.com]
Among other things, what NickMNS posts suggests that vignette ads are not automatically served to every user... that they're very high performing and do take users' interests into account... and for that reason aren't as interruptive as I suggested up above that they might be.
This also suggests that they're probably not the reason for the drops you're seeing. Again, try tracking new and returning visitors to see if these ads might be affecting return visitors.
