Latest / The Yield Doctor with James Deaker / AdTech Pirate Gold: Finding value in Supply Path Optimization

Transcript

AdTech Pirate Gold: Finding value in Supply Path Optimization · The Yield Doctor with James Deaker

• 0:00
  Adtech has its own version of the
• 0:03
  missing dollar problem. An advertiser
• 0:06
  spends money, a publisher receives less
• 0:09
  money, and a long chain of intermediate
• 0:12
  trees sit in between. And everybody
• 0:14
  starts arguing about where the
• 0:16
  differences occur. Is it fees? Is it
• 0:19
  value? Is it waste? Or is it theft? But
• 0:22
  before you can answer that, you need to
• 0:24
  do three things. You need to trace the
• 0:27
  dollar. You need to classify the dollar
• 0:30
  and then you need to decide whether the
• 0:31
  dollar was well spent. And that sounds
• 0:34
  simple, but an old riddle about three
• 0:37
  pirates, an inkeeper, and a bellhop
• 0:40
  shows why people get it wrong all the
• 0:42
  time. You may have heard a version of
• 0:45
  this riddle before, but stay with me
• 0:47
  because I assure you the adtech version
• 0:50
  has a few extra characters. So, here we
• 0:53
  go.
• 0:54
  Three pirates check into an inn late at
• 0:56
  night and the inkeeper charges them 30
• 1:00
  gold coins. So, they pay 10 gold coins
• 1:03
  each and they head upstairs. A short
• 1:06
  while later, the inkeeper realizes that
• 1:09
  he's overcharged them. The room should
• 1:11
  only cost 25 gold coins. And so, he
• 1:14
  gives five gold coins to the bellhop to
• 1:17
  return to the pirates. On the way
• 1:19
  upstairs, the bellhop realizes that
• 1:22
  there are three pirates. He has five
• 1:24
  gold coins. And so he pockets two of the
• 1:27
  gold coins and gives one each back to
• 1:31
  the pirates. So let's do the math. Each
• 1:34
  pirate originally paid 10 coins, got one
• 1:38
  back, and so they each paid nine coins.
• 1:42
  3 * 9 is 27 gold coins. And the bellhop
• 1:47
  kept two coins. 27 + 2 is 29. So where
• 1:54
  did the missing pirate gold go? The
• 1:57
  answer is simple. Nothing is really
• 2:00
  missing. The riddle relies on a
• 2:02
  classical mathematical misdirection. The
• 2:06
  trap is in adding the bellhop's stolen
• 2:08
  coins to the 27 coins when the 27
• 2:13
  already includes the two. The actual
• 2:16
  accounting is 25 coins to the inkeeper,
• 2:19
  two to the bellhop, and three return to
• 2:21
  the pirates, which totals 30 gold coins.
• 2:26
  Now, if this brain teaser is an analogy
• 2:29
  for adtech, I'm sure that people have
• 2:32
  different ways of telling it, but the
• 2:34
  most obvious is that the people who are
• 2:35
  spending the money are the pirates. So,
• 2:38
  they are the advertisers. The inkeeper
• 2:40
  is the holder of the inventory. So he is
• 2:43
  the publisher or the app owner. And the
• 2:46
  bellhop is a proxy for all intermediate
• 2:49
  trees, DSPs, SSPs, measurement
• 2:53
  companies, etc.
• 2:55
  There are a lot of different ways that
• 2:56
  this story can go, but let's start with
• 2:58
  the bellhop. In the story, he takes two
• 3:01
  gold coins without telling anyone,
• 3:04
  either the pirates or the inkeeper. And
• 3:07
  so this is the equivalent of either
• 3:10
  non-transparent fees or worse outright
• 3:13
  theft.
• 3:15
  If the story had either the pirates
• 3:18
  tipping the bellhop two gold coins or
• 3:21
  the inkeeper paying the bellhop two gold
• 3:24
  coins because he carried their chests up
• 3:27
  five flight of stairs, then it wouldn't
• 3:29
  be an interesting logic problem. But
• 3:32
  generally all parties would agree that
• 3:35
  the exchange was fairly paid. So before
• 3:38
  you argue whether the bellhop took too
• 3:42
  much, first you need to understand did
• 3:45
  the bellhop take an undisclosed fee or
• 3:48
  did he receive a tip? Did he earn a
• 3:51
  commission? Did he capture margin? Or
• 3:54
  did he just steal the coins? And a lot
• 3:57
  of people who complain about the adtech
• 3:59
  value chain like to say things like it
• 4:02
  used to be that advertisers would just
• 4:04
  give their money to an agency who would
• 4:07
  give it to a publisher and all of that
• 4:09
  original dollar ended up as working
• 4:12
  media. And those same people point to
• 4:15
  conference presentations where
• 4:17
  presenters have asserted that of every
• 4:20
  dollar that an advertiser spends less
• 4:22
  than 30% goes to Worky Media. And by
• 4:26
  that they mean the display advertising
• 4:29
  that sits on the publishers's page. And
• 4:32
  the assumption here is that all of the
• 4:34
  money that isn't working media is
• 4:36
  wasted. And by working media, they mean
• 4:39
  the raw inventory, the actual roof over
• 4:42
  the pirates's head, and they assume that
• 4:45
  everything else is a scam. So if you
• 4:48
  came into digital media and you looked
• 4:50
  at Terry Kawaj's Lumiscape of all of the
• 4:54
  various components and companies in the
• 4:57
  adtech supply chain that sit between the
• 5:00
  buyer and the seller. It includes
• 5:03
  measurement verification
• 5:05
  fraud biotech sell exchanges and
• 5:09
  creative management.
• 5:12
  In the pirate story, this is the
• 5:14
  equivalent of a bellhop, a concierge, a
• 5:17
  valet, a bartender, a housekeeper, all
• 5:21
  of whom want a gold coin to help you
• 5:23
  with your stay. So, let's continue with
• 5:26
  the pirate analogy. They're assuming
• 5:29
  that the pirates each pay 10 gold coins
• 5:32
  and that the various hotel staff are
• 5:34
  taking seven from from them, one each.
• 5:38
  However, from my perspective, the key of
• 5:40
  the story isn't about the worthiness of
• 5:43
  the players or even the amount of money
• 5:45
  that's taken. It's really about the need
• 5:48
  for clarity on what's paid to whom and
• 5:51
  whether it had any value. In the pirate
• 5:54
  story, the bellhop stole the money. In
• 5:57
  the adtech, the harder question is not
• 5:59
  often, did someone steal? It's did the
• 6:03
  service justify the fee. The question is
• 6:06
  not where did the money go. The better
• 6:08
  question is what did the coin buy and
• 6:12
  did that purchase improve the outcome.
• 6:15
  And one thing that's often overlooked in
• 6:17
  the story is the pirates were prepared
• 6:19
  to pay 10 gold coins for the room. The
• 6:22
  fact that they got the room for cheaper
• 6:24
  is nice but also disconnected from their
• 6:28
  overall willingness to pay. So again, if
• 6:31
  we imagine that we're living in the
• 6:33
  Terry Kawaja Lumiscape world of Adtech,
• 6:36
  where only three gold goes to the hotel
• 6:39
  and seven goes to the combination of the
• 6:41
  valet, the concierge, housekeeping,
• 6:43
  etc., then the pirates shouldn't really
• 6:46
  care if they get the room and the
• 6:48
  services that they expected. But they
• 6:51
  should care if the extra seven coins
• 6:53
  buys them nothing. Whether it's
• 6:56
  distorted incentives, reduced quality,
• 6:58
  or hiding the economics in a way that
• 7:00
  they did not agree to, the hotel should
• 7:03
  probably care. They should probably care
• 7:06
  a lot because they're only getting three
• 7:08
  gold pieces.
• 7:10
  But if the advertisers
• 7:12
  are the pirates and the hotel is the
• 7:15
  publisher, the interesting thing is that
• 7:18
  we often see the advertisers complaining
• 7:21
  more about the inefficiency than the
• 7:24
  publishers.
• 7:25
  And I think that sometimes that's
• 7:27
  because they think that someone, in this
• 7:30
  case the bellhop, is stealing from them
• 7:33
  rather than them paying a fee or tipping
• 7:37
  them. So in reality, we continue to see
• 7:41
  increasing inefficiencies in adtech. And
• 7:44
  if we didn't, honestly, advertisers
• 7:46
  wouldn't spend and all those
• 7:48
  intermediateries would be out of
• 7:49
  business. The challenge is accounting
• 7:52
  for what actually happened to all of the
• 7:54
  media spend.
• 7:56
  To stay with our pirate story, it can be
• 7:59
  hard to add up the spend. In the same
• 8:02
  way in adtech, it is hard to work out
• 8:06
  how much in each intermediate tree is
• 8:08
  taking and for what. What is the value
• 8:11
  of the media? What is the value of the
• 8:13
  targeting? And if a dynamic creative is
• 8:16
  used, what is that worth? And so the
• 8:19
  reason that I sometimes moan is that the
• 8:21
  problem of who is paying who is actually
• 8:24
  much more complicated. And it's more
• 8:27
  complicated because of the different
• 8:29
  currencies involved. Some of the
• 8:32
  parties, some of the intermediaries
• 8:34
  take a share of the media revenue or a
• 8:38
  share of the gross revenue. Some take a
• 8:41
  flat CPM payment based on the number of
• 8:44
  impressions and some change the
• 8:47
  economics based on the amount of spend.
• 8:50
  So in the pirate story, this is the
• 8:52
  equivalent of the bellhop wanting a gold
• 8:55
  coin, the housekeeper wanting three
• 8:57
  silver coins, and the bartender wanting
• 9:01
  three pieces of eight, whatever that
• 9:03
  might be. So, as someone who acts in an
• 9:07
  advisory cap capacity helping companies
• 9:09
  make more money, I often spend a lot of
• 9:13
  my time asking who, what, and why
• 9:16
  questions about the flow of money, where
• 9:19
  is the money going and why, what are the
• 9:22
  terms of the agreement, and what is the
• 9:25
  currency that is actually being used for
• 9:27
  the calculation? Is it net CPM, gross
• 9:30
  CPM, or some kind of flat fee? Honestly,
• 9:34
  if it were all pirate gold, it would
• 9:36
  probably be much more straightforward.
• 9:39
  So, if I have advice for anyone in the
• 9:42
  adtech chain, it would be to first
• 9:45
  understand what you are paying for in
• 9:47
  terms of working media, meaning what is
• 9:50
  the clearing price in the auction. Then
• 9:53
  instead of immediately jumping to
• 9:55
  feeling like you're getting ripped off
• 9:56
  because too much money is being taken
• 9:58
  away from working media, I would
• 10:01
  encourage you to step back and spend
• 10:03
  more time and energy on the problem of
• 10:06
  who is getting paid for what and whether
• 10:09
  those components are worth it. So to
• 10:12
  summarize,
• 10:14
  one of the main things you should be
• 10:15
  thinking about if you are the publisher
• 10:17
  or the advertiser in the story is
• 10:20
  whether you accept that you need to pay
• 10:22
  the fees to the bellhop, the valet, the
• 10:25
  housekeeper, or whether you believe that
• 10:28
  one of them is essentially stealing from
• 10:30
  you. If you can see the fees, then
• 10:34
  either accept them with good grace or
• 10:37
  try to go directly around them. But a
• 10:40
  word of caution. In doing so, you may be
• 10:43
  giving up services that you valued.
• 10:46
  Targeting, optimization,
• 10:49
  measurement, fraud controls, identity,
• 10:52
  workflow reporting, frequency
• 10:54
  management, or creative decisioning.
• 10:57
  In the pirate story, going direct may
• 11:00
  remove the bellhop, but then you need to
• 11:04
  carry your own bags, find your own room,
• 11:06
  verify the inkeeper, negotiate your own
• 11:09
  rate, and keep your own receipts.
• 11:13
  If money is genuinely disappearing
• 11:16
  through waste, opacity, poor incentives,
• 11:19
  or fraud, then go find the missing gold.
• 11:22
  But do not confuse those things. And
• 11:25
  this is where a lot of adte
• 11:27
  conversations go wrong. They start with
• 11:30
  the outrage before they've done the
• 11:33
  accounting. The better discipline is
• 11:36
  simple. Trace the dollar, classify the
• 11:39
  dollar, and then decide whether the
• 11:42
  dollar was well spent. I'm James Deika.
• 11:45
  I am the yield doctor, and I work with
• 11:47
  clients on issues like these to help
• 11:50
  them make more money from their digital
• 11:52
  assets.