Thread · 7 tweets · 1 Sept 2022

Hey — I'm only seeing this now (somehow people don't invite me to the fun parties) but I think that answering this requires some nuance (which most studies don't seem to have). There can be value in targeting without there being value in *third-party* targeting.

Third-party data is usually abysmally low quality. So using surveillance to target isn't necessarily effective. Likewise, there can be value in *broad* targeting (age range, gender, industry, income bracket) and little in specific targeting (blue lightbulb intenders).

These things are very hard to tease apart in part because marketers aren't necessarily rational and in part because the system is so opaque that measurements are very noisy. See the ISBA study: they couldn't figure out where *15%* of the money goes at all. isba.org.uk/knowledge/digi…

One thing that is pretty clear however is that we have a choice between a system that is good for publishers and one that is good for intermediaries, but you can't actually have both. We're not in the system that's good for publishers.

In one case you keep the data with the media, push innovation to the edges where there's actual competition, place responsibility with the party that's facing users, and federate the buying of seller-defined audiences. Better privacy and better business for publishers!

In the other case you extract raw resources and push the refining to intermediaries who thrive in opacity from both people and publishers. This is bad for privacy, it also creates competition issues (which does seem to match reality). berjon.com/competition-pr…

Sadly, it's hard to have an informed debate about this because the market is too opaque and because many of the existing studies look at coarse/irrelevant proxies like cookies.