Those really only cover giant tech co's though. Clearly it would be better and more accurate to get real data points, but I think the approach the article describes works well if there isn't a bunch of public data for a company already.
I got a reasonable amount of value out of skimming Blind during a recent switch from big tech to a more niche role. The amount of signal definitely drops (as does the signal-to-noise ratio) but it was still useful to a certain extent.
The outliers are pretty tough to get significant data for. Also my sense is startups tend to pay towards the 50-75th percentile of radford (read: not great) cash comp and balance with equity lotto tickets that the author values at zero. While startup equity is obviously not worth nothing, he is correct in noting it’s quite tricky to negotiate since there’s an inherent information asymmetry there as well.
