I used to run trainings on personal cybersecurity and explaining to people how much their data is worth. I've been paid to study this.

So, specific data about what you're worth to a company is proprietary. I can't find a link to a PWC or McKinsey report, but IoT device data typically sells for a range that's an estimate of cost per user per year. On the upper end, I've seen estimates of up to $50 per user per year. Low end is $1. So if the assumed lifetime of the TV is 4 years and a "household" is 2 adults and 2 kids, you end up at ($50x2 and $25 x 2)= $150 x 4 years = $600. So if Samsung sudsidizes the cost of a smart TV by $400, they're coming out ahead $200 on average, just on the subsidy. That's the kind of math going on for TV sales. Again, that's proprietary data, so short of trying to track down reports I saw years ago, all I can explain is that data monetization is a well-known cornerstone of business. Here's a quote for you about companies needing to know the value:

The exact same dataset, when sold to a financial services company, was being used to make multimillion-dollar decisions, so the data aggregator could charge $100,000. https://mitsloan.mit.edu/ideas-made-to-matter/what-data-wrapping-and-how-does-it-make-products-better

That's for companies operating legally and in the clear. What's crazy is that our data is treated sort of like student loan debt with them, because it's seen by them as debt we owe to the company and paid back over the life of the device. For criminals, it's pennies-on-the-dollar fire sales because nothing is guaranteed to work. So the data needed to steal your identity as a single line on a spreadsheet might only be $20 a person because the list of 10,000 records might only contain 200 winners. So you buy a $200,000 spreadsheet and hope to commit at least $1,000 per successful hit to come out ahead. which is a fairly low bar for fraud. Then the whole list is burned and you start over.