You said presumably, though the article gives specific numbers. Can you elaborate on what numbers or analysis would be reasonable as to not fall into the fallacy you also call out?
A significant proportion of the userbase would be a good start. If 118,000 users unsubscribed up from 40,000 then that's an increase of 78,000 people, or 0.04% of the subscriber base.
Also, now this is total speculation, I'd hazard that a large number of these "new unsubscribers" have clicked on social media links to the cancellation page without following through with the cancellation.
That seems reasonable, but isn't this data set a sampling, since it appears to only capture the page viewing habits of people who have opted to install various browser extensions? If so, wouldn't that imply that the non-sampled numbers would be larger, and perhaps by a fair bit, assuming that the percentage of users who install these browser telemetry tools is small? I agree that if the numbers did not represent a sampling, but were in fact reflected a complete data set, then yes, they perhaps are a drop in the bucket.
Also, now this is total speculation, I'd hazard that a large number of these "new unsubscribers" have clicked on social media links to the cancellation page without following through with the cancellation.