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.
