Analytics javascript can assign you an Unique Identifier string that can be passed along in a form where you fill out your phone number. They then tie your "session" or "user" (depending on how they have it set up) to a cookie they have placed on your machine.
They then leverage a partnership with a large ad network, to see where they found that cookie, as well as other cookies, from other sites in their network. They then correlate where they have seen essentially your user footprint (collection of cookies and other identifiers from your machine) stored throughout their network, to put you in some kind of demographic or intent-based bucket.
When you include data from mobile apps that are tracking your location, this can get very scary. There is a small leap required in tying a mobile ID (your phone's unique identifier) to web traffic, but it happens. Obviously all depending on the level of data you "share."
They then run every user in these buckets through a machine learning algo, that tries to predict things. Income, gender, spending habits, life events, etc. They can determine where you live easily, zip codes, compare that to census data, etc. It's all a part of predictive modeling.
That is the tracking portion.
The other portion is just tying a simple customer service rating to your phone number.
So imagine you piss of several customer service reps, live in a low income part of town, and rarely spend money. You might be on hold for quite some time...
They then leverage a partnership with a large ad network, to see where they found that cookie, as well as other cookies, from other sites in their network. They then correlate where they have seen essentially your user footprint (collection of cookies and other identifiers from your machine) stored throughout their network, to put you in some kind of demographic or intent-based bucket.
When you include data from mobile apps that are tracking your location, this can get very scary. There is a small leap required in tying a mobile ID (your phone's unique identifier) to web traffic, but it happens. Obviously all depending on the level of data you "share."
They then run every user in these buckets through a machine learning algo, that tries to predict things. Income, gender, spending habits, life events, etc. They can determine where you live easily, zip codes, compare that to census data, etc. It's all a part of predictive modeling.
That is the tracking portion.
The other portion is just tying a simple customer service rating to your phone number.
So imagine you piss of several customer service reps, live in a low income part of town, and rarely spend money. You might be on hold for quite some time...