[projectvrm] Shifting the costs of ad fraud

[Don Marti < >]

Making the rounds...

  Banner “Fraud” Doesn’t Matter
  Rick Webb
  https://medium.com/@RickWebb/banner-fraud-doesn-t-matter-fc84413fe59c

  We use banners as little billboards now. We use them
  strategically as incredibly cheap (so cheap) repeat
  impressions for brand awareness. We know many people
  don’t see them, we know most people don’t see them.
  That’s okay. We use them accordingly, and the cost
  has been adjusted down to make them a perfectly great
  buy even though most people don’t see them.

  We use them to measure the efficacy of campaigns we
  run _our_ way, using _our_ metrics. 

So Webb's method for dealing with fraud is:

 * Rely on the presence of huge quantities of
   fraudulent traffic to drive down CPMs everywhere.
   Buy a mix of fraudulent and legit banner impressions
   at massive scale.

 * Assume that adtech data is bogus, and that you
   have to measure banner impact with offline tools,
   as you would for a billboard.

That solves Webb's problem.  Unfortunately, his
strategy makes things worse for publishers and users.

 * Ad fraud isn't just servers full of bots hitting
   servers full of ads. Ad fraud is now the number one
   malware payload. When Webb pays indiscriminately
   for fraudulent and legit banners, he's funding
   malware development, along with a bunch of other
   bad behavior, from copyright infringement to comment
   spam.

 * When fraud is priced in, legit sites are in direct
   price competition with fraud sites.  Costs of fraud
   end up being borne by publishers in the form of
   lower CPMs.

Important implication of this: if you can figure out
the minimum quality level that makes banner inventory
viable for Webb's "mass buy/offline measurement"
strategy, you can focus anti-fraud projects on the
_marginal banner_ -- whatever inventory is just barely
possibly non-fraudulent enough for a real advertiser
to consider buying it.

How do we spot the low-quality ad buys?  This is
where brand-unsafe placements and ad-supported piracy
could be useful.  Spot those, and you can get some
idea of where the fraud-tolerant ad money is going,
and where is best to move in order to cut it off.

-- 
Don Marti 
<
 >
                   
http://zgp.org/~dmarti/
Are you safe from 3rd-party web tracking?  http://www.aloodo.org/test/