> If this product is capable of collecting that sort of data in the first place
It's based solely on the http referer, which your browser sends.
It's based solely on the http referer, which your browser sends.
Note that referrers are self reported, so there's an incentive for such a service to avoid relying on them. Each web site can set its own desired referrer policy, and users can further configure when, if, and how their browser sends referrers. (For example, Firefox has referer.XOriginPolicy and referer.spoofSource among other preferences.)
> Each web site can set its own desired referrer policy
Yes, and it won't be able to tell which site you come from if it's set to no-referer.
I never claimed otherwise? There are two separate metrics in question here. One is the "visit" metric which the post specifies is based on the referrer. The other is quantifying which other sites sent traffic your way. The post conspicuously fails to define how that second metric is computed.
For that second metric, I noted that there is an obvious incentive to avoid using the referrer and that the current status quo is to rely on other methods. (In particular, fingerprinting combined with centralized analytics services that have vision into a vast amount of traffic. Note that Cloudflare easily qualifies here given the sheer number of assets and websites they host.)
If the entirety of this analytics service is based solely on referrers coupled with IP addresses then the author should unambiguously state that up front. Otherwise I'm going to assume the worst for the reasons given above.