will ITP and ETP kill traffic arbitrage?
12 September 2019
Today, Traffic Arbitrage is a profitable scheme.
Make a crappy site full of cheap clickbait content.
Load it up with programmatic ad slots.
Buy content recommendation widgets, or
Any time someone clicks through from a legit site, show them a bunch of ads, including high-CPM video ads.
The reason that this model has worked so well is that with programmatic advertising, impressions on the crappy site are worth something. The user from the legit site brings along the same third-party tracking cookies that their browser sends to everybody. So a legit advertiser is likely to bid to reach them.
The publisher of the legit site gets some of the revenue—they get paid per click on the chumboxes—but they don't see the revenue lost because they're now competing for programmatic bids with the lower-cost sites that they're sending traffic to. The tragedy of the chumbox is that all the legit publishers would be better off if nobody could do traffic arbitrage, but the first legit site to stop participating in traffic arbitrage loses.
Now browsers are blocking the third-party tracking that enables traffic arbitrage. First Apple Safari ITP, now Firefox ETP, soon Microsoft Edge.
A legit site that can persuade users to subcribe or sign in now has significantly more data on its audience than a crappy site where people are unlikely to share their information. Traffic arbitrage schemes are being starved of data, which should start to show up as lower RTB bids on their sites, and higher bids on legit sites.
There are uses for content recommendation widgets other than traffic arbitrage, and it will be interesting to see how the content that gets recommended on them changes as traffic arbitrage schemes get squeezed out.
This isn't the last move in the game, of course. For large Internet platform companies, privacy violation is not required, but a means to an end. The business model for platforms depends on capturing an ever-larger share of ad revenue by commodification of the content business, and violating user privacy expectations is the easiest, 1.0, way to do it. It may be possible to construct a user data scheme that both commodifies content and preserves user privacy, so the interests of legit publishers are aligned with, but not identical to, the privacy interests of their readers.