Database Reference
In-Depth Information
me, of who I am at work, of who I am at home, and what I do with my hobbies.
So labeling people in this kind of categorical way is meaningless. But that is the
way the industry has worked for the last hundred years.
And now we find that the industry is shifting to this very fine-grained, spe-
cific data that people can see about what consumers actually do, how they
express themselves in their actions, which means they can stop using some
arbitrary label that is typically wrong. I think this shift that the industry is fac-
ing right now—and we are part of that -will take a while before that really
gets embraced. With it come notions of communicating the value proposition
of advertising.
From a consumer perspective, most of them are not terribly happy about
seeing a whole bunch of banner ads. The fact that roughly ninety-five percent
of the Internet infrastructure is paid for by advertising is very far in the back,
back, background of people's thoughts. You will not find a free blog-hosting
system anymore once you ban all advertising. So there is a discussion that
needs to occur about the tradeoff of finding the right balance and level of
advertising. The discussion should center around giving people the large-scale
voice and choice as to where they want to stand on the continuum of paying
for things and not being bothered and tracked, or having systems that are not
paid-for content, and therefore having to deal with the fact that you will see
advertising. So that is one of the current big issues.
Another challenge facing the industry right now is about how to integrate
mobile, because more and more people are spending time on mobile. The
conversation is based around how to properly deal with even more sensitive
information about where exactly people spend their lives. Dealing with sensi-
tive information is a very interesting challenge, as well as how to, on the other
side, actually connect it to metrics so that we can make sure we are not wast-
ing anybody's time. Advertising should be able to be relevant to the person. In
order to be able to deliver that, I need to tie your identity back to when you
take an action. I can only optimize it if I can relate the same identity back to
the person purchasing.
However, there are a lot of issues around attribution and last click. These
issues really create a lot of incentive mismatch, which is a big problem for the
industry. If you say the credit goes to the last impression just before purchase,
then my incentive is to show you 50,000 ads all over the board at minimal
cost—I am not going to be very specific about this—whereas if you try to
do a more causal analysis to see which impressions really seem to have in
aggregate—obviously, you cannot do this for one person—tipped the needle
and changed that person's behavior, it is a much harder problem. There are
analytical solutions available to do it, but they are really difficult to make reli-
able. So that is the attribution issue of which metrics is another big part of it.
