2 ms·
Means (and typically last-click based means at that) are surprisingly poor summaries of distributions. Even if the average for Twitter users is low, there are
by disgruntledphd2 10y ago
Means (and typically last-click based means at that) are surprisingly poor summaries of distributions.
Even if the average for Twitter users is low, there are probably pockets of value for particular services. These are potentially not the kinds of customers these affiliate networks are interested in.
Your Android point is rather different, given the wide disparity of Android devices.
If 80% of Android devices are worth less than $10 per year, while 20% are worth $100 per year, and all iPhones are worth $100 per year, you'll have a much lower mean value for Android while also having a subset of very profitable users.
As you may have noticed, I'm not a fans of means in the advertising world.
On your point on Twitter's value, I see a lot of missed opportunity.
It felt like they slowed down really early in terms of what they did (and specifically their success with their product changes). Their growth stalled at quite a low point (relative to other social networks/services), which made it hard for them to attract good people, which further compounded the effect.
But then, I never used Twitter (and maybe that's the problem).
- downandout 10y ago>Your Android point is rather different, given the wide disparity of Android devices. I disagree that it's different. Of course there are pockets of both value and worthlessness everywhere, but overall the LTV of an Android user to developers and advertisers is a small fraction of that of an ioS user, and a Twitter user is worth (to advertisers) an even smaller fraction of the value of a Facebook user. At scale, averages are what count. >As you may have noticed, I'm not a fans of means in the advertising world. I did notice that, and I cannot disagree with you there. But currently, for many advertisers, last-click based means is the most efficient available attribution model, even though I agree that it's far from perfect.
- disgruntledphd2 10y agoI don't think its efficient, just easy. It makes very little sense though. Do you attribute which beer you drink to the sign outside the bar? I think that its difficult to assess your premise under a last-click model, to be honest. When you look at the history of advertising, its very difficult to believe that a direct response (measurable by click) is actually the whole thing. I'm pretty sure I've never clicked on a Coca-Cola ad, but I have bought a lot of their product over the years. I mean, conditional on your experience with the value of Twitter users (for a particular set of clients/services) and a last-click model, I don't disagree with you, I just think that such a view of advertising leaves an awfully large amount of stuff out. Full disclosure: mobile attribution is hard, there is not standard or even dominant way of tracking views there (i believe) and so for some mobile-heavy/only advertisers better attribution methods are just a pipe dream at this point.