Latest / The Yield Doctor with James Deaker / Three Ways Publishers Go Programmatic (And Why Most Get It Wrong)

Transcript

Three Ways Publishers Go Programmatic (And Why Most Get It Wrong) · The Yield Doctor with James Deaker

• 0:00
  Most publishers don't struggle with
• 0:03
  programmatic because of bad technology.
• 0:06
  They struggle with it because they never
• 0:08
  decided what programmatic was for. When
• 0:11
  someone tells me, "We're thinking about
• 0:13
  going programmatic," my first reaction
• 0:15
  isn't to think about SSPs or floors or
• 0:18
  auctions. It's to ask them, "What is the
• 0:22
  role that you actually want programmatic
• 0:24
  to play in your business?" Because going
• 0:27
  programmatic isn't just a switch that
• 0:30
  you flip. It's an operating model. And
• 0:33
  if you don't choose one deliberately,
• 0:36
  you are usually going to end up with
• 0:38
  something accidental.
• 0:40
  Going programmatic is one of the biggest
• 0:42
  decisions that a digital publisher can
• 0:44
  make. And if you don't think through the
• 0:47
  role of programmatic explicitly upfront,
• 0:51
  then what is usually going to happen is
• 0:53
  you drift into a model without realizing
• 0:55
  it. one that can create internal
• 0:58
  tension, confused price signals, uh an
• 1:02
  unhappy sales team, and usually very
• 1:05
  mixed revenue results. There are also
• 1:07
  really real costs associated with
• 1:10
  getting this wrong. You can't just turn
• 1:12
  on the pipes and expect that the money
• 1:14
  is going to flow. Ad quality can suffer,
• 1:17
  the sales team can get undermined, and
• 1:20
  advertisers can get confused about price
• 1:23
  and about access. So before you turn
• 1:26
  anything on, you need to be honest about
• 1:28
  what do you really believe. Do you
• 1:30
  believe that your inventory is best
• 1:33
  protected by direct relationships?
• 1:36
  Or do you believe that opening up more
• 1:38
  demand will reveal your true market
• 1:41
  value? Or do you believe that the right
• 1:44
  answer sits somewhere in between? And in
• 1:46
  this video, I want to break down this
• 1:48
  decision. I'm going to walk through the
• 1:50
  different ways in which a publisher can
• 1:52
  choose to go programmatic not as tactics
• 1:56
  but instead as operating models so that
• 1:58
  you can recognize the one that you are
• 2:00
  running today and whether it's the one
• 2:03
  that you intend to run. So let's start
• 2:06
  with model one and it's the model that
• 2:09
  dominated digital media for basically
• 2:13
  everything up to 2015.
• 2:15
  And I want to start here because the
• 2:17
  model is internally consistent. The
• 2:19
  logic is clean and in the right context
• 2:23
  it still exists today especially in
• 2:26
  niche and in premium and in dei
• 2:28
  developing environments like digital out
• 2:31
  of home or audio and podcast. And at the
• 2:34
  core of this model is a simple belief
• 2:37
  that inventory creates the most value
• 2:41
  when it's sold through direct
• 2:42
  relationships.
• 2:44
  Programmatic in this world isn't the
• 2:47
  business, it's insurance. Sales teams
• 2:51
  have strong relationships with
• 2:52
  advertisers. They control the access.
• 2:55
  They protect the pricing. And
• 2:57
  programmatic largely exists to catch
• 3:00
  what doesn't get sold directly. And
• 3:03
  that's the phrase that everybody was
• 3:05
  using in conferences in that period 2010
• 3:09
  through 2015. They'd use phrases like
• 3:13
  don't trade brand dollars for
• 3:16
  programmatic pennies. And this wasn't
• 3:18
  because of fear of technology. It was a
• 3:20
  deliberate choice about how value was
• 3:23
  created and protected. So let's look at
• 3:27
  the model. My apologies that this video
• 3:30
  is going to involve a lot of tables for
• 3:33
  for those of you who are going to be
• 3:34
  looking at the visuals rather than just
• 3:36
  listening to the audio. But the reason
• 3:38
  for that is I wanted to ensure that
• 3:40
  there were consistent dimensions for
• 3:42
  comparing across the different models.
• 3:45
  So let's look at the table. What you are
• 3:48
  going to see here is that the main u
• 3:52
  belief is that inventory value is best
• 3:55
  protected through direct relationships
• 3:59
  and not through the open market
• 4:01
  competition.
• 4:02
  The primary objective of programmatic in
• 4:05
  this model is simply to fill the unsold
• 4:09
  inventory. It's the idea that something
• 4:12
  is better than nothing in this model.
• 4:16
  What is the role of the direct sales
• 4:18
  team? The direct sales team is
• 4:20
  responsible for selling and protecting
• 4:23
  all premium inventory is what I wrote
• 4:26
  here. But in pract essentially all
• 4:28
  inventory. It's the main way in which
• 4:30
  the business makes money. The role of
• 4:33
  programmatic therefore is a secondary
• 4:36
  revenue channel. It's all of the
• 4:39
  leftover demand. Now, there are
• 4:42
  challenges with this model that I'm
• 4:44
  going to get into here shortly. And one
• 4:46
  of them is the breadth of market
• 4:48
  exposure. Because you're dealing with a
• 4:51
  direct sales team, there are only so
• 4:53
  many advertisers or agencies that they
• 4:56
  can call on. And so, even the biggest
• 4:59
  sales teams or most sales teams are
• 5:01
  probably only contacting dozens of
• 5:03
  advertisers.
• 5:05
  Even the largest premium publishers
• 5:08
  might be dealing with hundreds or
• 5:10
  possibly thousands of advertisers. They
• 5:13
  certainly aren't touching hundreds of
• 5:15
  thousands of advertisers which can occur
• 5:18
  in a programmatic or auctionbased
• 5:20
  environment.
• 5:22
  So what else do we can need to consider
• 5:25
  when we're looking at the model? The
• 5:28
  other main thing that we need to think
• 5:29
  about is what is the quality of the
• 5:31
  pricing signal that you receive? And in
• 5:34
  this model, the signal is low because
• 5:37
  you are only getting limited market
• 5:40
  feedback on the value of your inventory.
• 5:43
  You're getting told by some major
• 5:46
  advertisers through your own sales team
• 5:49
  whether your inventory is any good,
• 5:51
  whether it's performing. You don't know
• 5:54
  whether if there is underperformance,
• 5:56
  whether it's due to a downturn in the
• 5:58
  market, whether your sales team is
• 6:01
  underperforming
• 6:02
  or whether there are other dynamics
• 6:05
  going on. Maybe your inventory is
• 6:06
  starting to fall out of favor. So you
• 6:09
  have challenges around the quality of
• 6:11
  the pricing signal.
• 6:14
  How do we think about ad serving and
• 6:17
  clearance in this model? The traditional
• 6:20
  way it's run is what's called a
• 6:21
  waterfall. Basically, everything sold
• 6:23
  through your direct sales team gets
• 6:26
  priority. What they don't sell, as soon
• 6:29
  as something looks like it's going to be
• 6:30
  unsold, would otherwise served as a
• 6:32
  house ad, it goes out to a programmatic
• 6:35
  channel. But the idea of this is
• 6:38
  obviously that the programmatic is
• 6:40
  getting what's left over. So, how
• 6:44
  important is the pricing power of
• 6:46
  programmatic in this world? It's low uh
• 6:50
  and it's low by design. And so what you
• 6:52
  are often going to find in this type of
• 6:54
  environment is that the managed sales
• 6:57
  team or the brand advertisers
• 7:00
  continue to outperform the programmatic
• 7:03
  by a significant factor. 3 5 10x is not
• 7:08
  uncommon in terms of the prices that you
• 7:11
  see from your primary channel compared
• 7:13
  to your programmatic channel. And again
• 7:16
  that's a function of how you've set up
• 7:18
  this model.
• 7:20
  What access does programmatic get to the
• 7:22
  majority of the products? It's going to
• 7:24
  be very limited. And when I talk about
• 7:26
  access, I'm talking about what can
• 7:28
  programmatic see. Often, they might not
• 7:31
  see the URLs that they're trying to bid
• 7:33
  into. They might get less product
• 7:36
  detail. They might be uh cut off from
• 7:39
  getting access to the premium products
• 7:41
  or the sponsorship products. And they
• 7:43
  may not get the type of targeting that
• 7:45
  the managed sales team would get. They
• 7:47
  may not get audience segments. They may
• 7:50
  not get access to the high-quality data
• 7:52
  from the premium publisher.
• 7:55
  So if you embark on this model, what is
• 7:58
  the primary risk that you're accepting?
• 8:00
  The primary risk is that you may never
• 8:03
  see the full demand curve for your
• 8:05
  inventory. You're operating in a very
• 8:07
  constrained world and you are hoping
• 8:10
  that your primary advertisers that you
• 8:12
  work with are enough to satisfy your
• 8:15
  business at your current cost base.
• 8:18
  One of the things that we think about as
• 8:20
  we go through these models is what
• 8:22
  breaks first if this model is
• 8:24
  misaligned.
• 8:25
  And one of the things that people don't
• 8:27
  think about is the pricing credibility
• 8:30
  and the and the pricing discipline that
• 8:33
  erodess before revenue does. And so what
• 8:35
  I mean by that is if you are continuing
• 8:38
  to price and sell your inventory at a
• 8:40
  relatively high price point, but the
• 8:43
  market has moved on and doesn't value
• 8:45
  your inventory as much, sales is going
• 8:48
  to continue to sell at that price point
• 8:51
  and they may start to get signals back
• 8:53
  from the market that the inventory is
• 8:56
  undervalued.
• 8:57
  Well, in that world, they are going to
• 9:00
  struggle to sell, but there's very it's
• 9:02
  very much going to be a lagging
• 9:04
  indicator of the fact that the market
• 9:06
  has moved and so sales won't have
• 9:09
  credibility. Pricing discipline is going
• 9:11
  to start to erode and you're going to be
• 9:13
  playing catchup football. All right, on
• 9:16
  to the second model. So when we think
• 9:20
  about the second model uh I'm going to
• 9:23
  go to the opposite end of the spectrum
• 9:25
  deliberately and this model starts from
• 9:29
  a very different belief. The model
• 9:31
  starts from belief that the inventory
• 9:33
  creates the most value when it's exposed
• 9:36
  to as much demand as possible and the
• 9:39
  market is allowed to set the price. This
• 9:42
  was essentially the way Google
• 9:44
  approached search. It's the way that
• 9:46
  some of the biggest uh social media
• 9:49
  companies approach the display
• 9:51
  advertising. They are trying to touch
• 9:53
  hundreds of thousands of different
• 9:55
  advertisers including mid-market and
• 9:58
  very small advertisers.
• 10:00
  In this world, programmatic isn't a
• 10:02
  backup plan. It is the business. So
• 10:05
  instead of protecting inventory through
• 10:08
  relationships, these companies maximize
• 10:11
  the value through scale, competition,
• 10:14
  and continuous market feedback.
• 10:17
  Sales doesn't disappear in this model,
• 10:20
  but its role changes considerably. It
• 10:23
  becomes focused on managing the largest
• 10:26
  relationships, shaping products, and
• 10:28
  supporting the ecosystem around the
• 10:30
  auction rather than clearing the
• 10:33
  inventory on a dealbydeal basis. This
• 10:36
  model works best when products are
• 10:40
  measurable, repeatable, and perform at
• 10:43
  scale. And when it works, it's extremely
• 10:46
  powerful. Pricing is discovered in real
• 10:49
  time. demand isn't constrained by who
• 10:51
  your sales team can reach and the market
• 10:54
  tells you very quickly what your
• 10:56
  inventory is worth. So let's go through
• 11:00
  that same exercise and answer those same
• 11:03
  questions about you know how this model
• 11:06
  works. So the core belief about how the
• 11:10
  value is created is that value is
• 11:13
  maximized by exposing inventory to as
• 11:16
  much demand as possible and letting the
• 11:19
  market set the price. In this world,
• 11:22
  programmatic is the primary revenue
• 11:24
  engine. It clears the market at the
• 11:27
  highest sustainable price and often the
• 11:30
  ad server will be set up in such a way
• 11:33
  that the programmatic channel will
• 11:35
  compete directly with the manage sales
• 11:37
  team. The primary role of direct sales
• 11:40
  here is to manage the largest strategic
• 11:43
  relationships with the largest
• 11:46
  advertisers and the agencies and they're
• 11:49
  going to manage the relationship. they
• 11:51
  may or may not grant them any special
• 11:53
  preference in the auction or in the
• 11:55
  clearing mechanism.
• 11:57
  The primary role of programmatic then is
• 12:00
  it's the main channel for monetization
• 12:03
  and in terms of how much of the market
• 12:06
  gets you get exposure to it's very broad
• 12:09
  potentially tens or hundreds of
• 12:11
  thousands of advertisers depending on
• 12:14
  your geography.
• 12:16
  What is the quality of the ads signal or
• 12:19
  the price signal that you are going to
• 12:20
  get through programmatic? It's generally
• 12:23
  going to be very high. You're getting
• 12:25
  continuous realtime feedback on the
• 12:28
  demand and the willingness to pay.
• 12:30
  You're going to be able to look at how
• 12:31
  that demand changes over the course of a
• 12:34
  day, over the course of a week. You're
• 12:37
  going to look at how it varies by
• 12:39
  geography. You're going to be able to
• 12:40
  see how it varies by advertiser segment.
• 12:43
  you get a ton of high quality
• 12:45
  information about pricing that you don't
• 12:47
  get in the managed world.
• 12:50
  The ad serving and the clearing logic is
• 12:53
  generally going to be some version of
• 12:55
  open market options with limited
• 12:57
  exceptions. Uh I say open market, it may
• 13:00
  be a controlled auction that has some
• 13:03
  kind of discount factors or multipliers
• 13:06
  associated with it. There are all sorts
• 13:08
  of different variations of this, but
• 13:10
  it's generally going to be some kind of
• 13:12
  automated clearing mechanism. The
• 13:15
  pricing power of the programmatic
• 13:17
  channel is going to be high driven by
• 13:18
  the scale and the competition and
• 13:21
  hopefully the liquidity within that
• 13:22
  channel.
• 13:24
  If you think about the access that the
• 13:27
  programmatic channel is going to get to
• 13:28
  products data and targeting, they're
• 13:30
  going to get access to everything. It's
• 13:32
  going to be broad. that most products
• 13:33
  and targeting and will be available to
• 13:35
  the programmatic channel. What is the
• 13:38
  primary risk that you accept when you go
• 13:41
  down this path? One is the loss of
• 13:44
  differentiation. Once the demand growth
• 13:47
  or the performance slows, you have less
• 13:50
  levers. As I say, you are constantly
• 13:53
  getting feedback on the quality of your
• 13:55
  inventory. You may start to see that
• 13:57
  relative to your competition, it's
• 13:59
  slipping and there's not a lot you can
• 14:02
  do through sort of marketing means or
• 14:04
  sales or business development type
• 14:06
  behaviors. Everything's been exposed.
• 14:09
  The market knows the value of the
• 14:10
  inventory and you know the value of the
• 14:12
  inventory and so you have less levers at
• 14:15
  your at your disposal.
• 14:17
  All right, let's go to the third model
• 14:20
  now.
• 14:21
  So if model one and two represent the
• 14:24
  extremes, most publishers actually live
• 14:28
  somewhere in the middle. And one
• 14:30
  important thing to say up front is this
• 14:32
  that from an operating perspective, it's
• 14:36
  actually easier to live at the extremes.
• 14:40
  The rules are cleaner, the decisions are
• 14:42
  more principled, and the incentives line
• 14:46
  up much more naturally. You may not be
• 14:49
  maximizing revenue, but you're clear
• 14:51
  about what you're optimizing for. The
• 14:54
  hybrid model exists for a different
• 14:57
  reason. It's sometimes it's intentional,
• 15:00
  and sometimes it's the result of a
• 15:02
  series of perfectly reasonable decisions
• 15:06
  made over time. A company might start in
• 15:09
  model one built around a strong sales
• 15:11
  team and gradually open up to
• 15:13
  programmatic as they realize that
• 15:15
  they're leaving money on the table. Or
• 15:18
  they might aspire to model 2, a more
• 15:20
  market driven approach, but they lack
• 15:23
  the internal resources or data or
• 15:26
  technology to run that model end to end
• 15:29
  on their own. So they rely on partners,
• 15:32
  intermediaries
• 15:33
  and partial solutions. And so the result
• 15:36
  is a hybrid. This model can work uh and
• 15:40
  again it can work well but it's the most
• 15:42
  complex to operate and it's the one
• 15:45
  where many of the answers to the
• 15:47
  questions that I'm going to be walking
• 15:49
  through start with. It depends. But that
• 15:52
  is not a flaw and it doesn't mean that
• 15:54
  you uh need a much but it does mean that
• 15:57
  you need a much clearer story around
• 16:01
  yield management and tighter alignment
• 16:03
  across teams than most publishers do. So
• 16:07
  let's look at the table now.
• 16:12
  What we see is that in this world your
• 16:15
  core belief is that value is created by
• 16:19
  balancing the relationshipled sales team
• 16:22
  with some market discovery uh around
• 16:26
  price. The primary objective of
• 16:29
  programmatic is to support the both the
• 16:31
  revenue growth and the price signal
• 16:34
  discovery.
• 16:36
  The primary role of sales here is to
• 16:38
  represent the brand to sell to the
• 16:40
  largest customers and to package unique
• 16:43
  or hightouch products. And again, this
• 16:46
  happens in a lot of companies where you
• 16:48
  find that you're constantly launching
• 16:50
  new products and you need the sales team
• 16:53
  and market to represent them because
• 16:55
  programmatic simply hasn't caught up. In
• 16:59
  this world, the primary role of
• 17:01
  programmatic is to generate incremental
• 17:03
  revenue and to provide a market clearing
• 17:06
  signal.
• 17:08
  In terms of the market exposure, it's
• 17:11
  going to depend and this really does
• 17:13
  depend. I say moderate here, dozens or
• 17:16
  hundreds of clients from the direct
• 17:17
  advertisers and potentially many, many
• 17:20
  more coming in through programmatic, but
• 17:23
  they may be coming in through
• 17:24
  aggregators. So you may not have a sense
• 17:26
  for how many end advertisers are really
• 17:29
  seeing your inventory.
• 17:31
  If you think about the quality of the
• 17:33
  pricing signal here, it's mixed because
• 17:36
  you are getting what in some cases could
• 17:38
  be contradictory signals about the value
• 17:41
  of your inventory. You may have to look
• 17:44
  at what you're seeing from programmatic
• 17:46
  and it may represent different trends
• 17:48
  than what you're seeing from your own
• 17:49
  sales team. So there can be noise and it
• 17:52
  can be distorted and it couldn't be a
• 17:55
  function also of the yield management or
• 17:57
  the channel management controls that you
• 17:59
  may need to put in place.
• 18:02
  The ad serving and clearing logic, it's
• 18:04
  generally going to be a hybrid channel
• 18:06
  management across the direct. You're
• 18:09
  going to have PMPs or private
• 18:11
  marketplaces and you may have some open
• 18:14
  auction. Again, you've potentially got a
• 18:16
  lot going on here. the relative pricing
• 18:20
  power of programmatic. Uh I say here
• 18:22
  it's variable and depends on how
• 18:23
  well-managed it is but you are going to
• 18:26
  find that in some areas programmatic is
• 18:29
  outperforming your managed sales team
• 18:31
  and that's the kind of signal that makes
• 18:33
  you realize that you need to move things
• 18:35
  around. Either you need to give your own
• 18:37
  sales team some uh products that only
• 18:40
  they can sell or you need to change your
• 18:42
  ad serving priorities. Something's going
• 18:45
  on. But one of the things you're
• 18:46
  definitely looking at is how is your
• 18:48
  managed sales team performing relative
• 18:51
  to the programmatic channel. The access
• 18:54
  that the programmatic is going to have
• 18:56
  to product data and targeting is going
• 18:58
  to be selective. That is going to be a
• 19:00
  decision that your yield management or
• 19:03
  channel management team needs to make
• 19:05
  strategically as you think about where
• 19:08
  you are on this journey.
• 19:10
  The primary risk that you're accepting
• 19:12
  is partly organizational. uh you need to
• 19:16
  ensure that you are balancing tradeoffs
• 19:18
  because you are going to have incentive
• 19:20
  misalignment across your organization.
• 19:23
  You're going to have a salesperson who
• 19:26
  is constantly clamoring for better
• 19:29
  products to sell or some preference
• 19:31
  against programmatic. You're going to
• 19:32
  have some product and engineering people
• 19:34
  who think that the world should be all
• 19:36
  programmatic and that they need uh to
• 19:39
  basically just open up everything. And
• 19:41
  so you need to ensure that you have a
• 19:44
  group within your organization who is
• 19:46
  responsible for balancing those
• 19:48
  tradeoffs.
• 19:50
  When does this model start to break? Uh
• 19:52
  it's going to break if there's internal
• 19:54
  friction or slow decision making that
• 19:57
  emerges and you start to see revenue
• 19:59
  decline and everybody's going to have a
• 20:01
  perspective on whose fault that is,
• 20:03
  whether it's a result of biasing one way
• 20:06
  or another as you as you try to play
• 20:08
  this hybrid. All right. So that's that's
• 20:11
  model three. Now as we come to the end
• 20:15
  and close, couple of thoughts. One,
• 20:19
  programmatic isn't about picking the
• 20:22
  right technology uh or copying what the
• 20:24
  biggest platforms are doing. It's about
• 20:27
  being honest with yourself and with your
• 20:29
  team about what creates value for your
• 20:33
  business and what creates value for the
• 20:35
  customers that you're trying to work
• 20:36
  with. For every model that we talked
• 20:39
  about today can work and every model can
• 20:42
  fail. The problems start when the model
• 20:46
  that you are running doesn't match the
• 20:48
  model that you think that you are
• 20:49
  running or when your sales team adops
• 20:52
  and leadership are all optimizing for
• 20:55
  different versions of the truth. That's
• 20:58
  usually when programmatic gets blamed.
• 21:01
  But most of the time, the real issue is
• 21:03
  that the decisions that I've outlined
• 21:05
  here were never actually made explicit,
• 21:08
  and executive teams aren't having the
• 21:10
  conversations around how to think about
• 21:13
  these different channels. If this video
• 21:16
  helped you today recognize which model
• 21:18
  you're operating or where the friction
• 21:20
  is coming from, then it's done its job.
• 21:22
  And if you're in that middle ground
• 21:24
  trying to balance relationships with
• 21:27
  market-based signals and that's where
• 21:29
  these decisions get hard, there are real
• 21:31
  tradeoffs and these are exactly the type
• 21:33
  of problems that I love working with. So
• 21:35
  if you want a second opinion or a sanity
• 21:38
  check your existing model, reach out.
• 21:40
  I'm James Deika. I am the yield doctor.