This is a study any publisher can replicate today. Compare an experimental group with setRequestNonPersonalizedAds [1] against a control group. I think you'll see similar numbers.
My non-expert [1] understanding is that by default Google's ads are both contextually targeted and personalized. If you recently visited a mattress site and they are trying to bring you back for a sale, you'll get mattress ads everywhere via remarketing, a kind of personalization. On the other hand, if you're visiting a mattress site for the first time, you'll probably also get mattress ads, but this time because of the context of the page.
If the publisher requests non-personalized ads, that turns off the personalization, but not the contextual targeting, so the answer to your question is kind of yes, kind of no: it removes personal targeting from the mix, but contextual targeting is always on.
It seemed like the lack of reference in TFA to the source of that study - specifically the lack of name-dropping of the source to serve as an ethos - made me immediately assume that it was probably a study run by the very company putting this borderline propaganda out. Glad to have it confirmed.
The source for this seems to be a Google Ad Manager "study" (a paragraph in the middle of https://www.blog.google/products/ads/next-steps-transparency...).
Was this an apples-to-apples comparison, i.e. did they serve relevant contextual ads when they turned off cookies?
Other studies disagree:
>We find that when a user’s cookie is available publisher’s revenue increases by only about 4%
-- Online Tracking and Publishers’ Revenues: An Empirical Analysis (https://weis2019.econinfosec.org/wp-content/uploads/sites/6/...)
Edit:
>After GDPR, The New York Times cut off ad exchanges in Europe — and kept growing ad revenue
-- https://digiday.com/media/gumgumtest-new-york-times-gdpr-cut...