Professional Documents
Culture Documents
DISPLAY
TRADING
Buyers Guide
1...
Agency
Advertiser
Agency
WHAT DO
THEY ALL
DO?
2...
Advertiser
WELL....
Agency
RT
VE
AD
Ad
Network
ADVERT
BUYING
THROUGH
RTB ALLOWS
US TO USE
TECHNOLOGY
TO REACH
A SPECIFIC
AUDIENCE IN
AN AUTOMATED
WAY
Agencys
voice
Publisher
Website
ADVERT
Agency
Trading Desks
ADVERT
ADVERT
Demand Side
Platforms
Ad
Exchange
Sell Side
Platforms
Advertiser
Agency
RTB SOUNDS
A BIT
squeek!
COMPLICATED HOW DO I FIND
OUT MORE?
KEEP
READING
THIS BOOK
AND FIND
OUT!
Introduction
Phil Macauley, EU
Managing Director,
Quantcast
The real-time bidding (RTB) industry began nearly three years ago with
the September 2009 launch of Googles DoubleClick Ad Exchange, the
first marketplace powered by real-time auction technology. Since then the
industry has experienced tremendous growth. eMarketer projects that
spending on real-time exchanges will grow to $5 billion by 2015.
According to International Data Corporation, as quoted in Real-Time Bidding
Spending to Quadruple This Year, (eMarketer, October 27, 2011).
It all began with a simple question: Could display advertising borrow the key lessons
from search advertising? Being able to uniquely target every single ad impression is
immensely useful, however pricing each ad impression based on a real-time auction
is massively valuable.
From those audacious ideas, the real-time ad exchange was born a framework to
take what had been pioneered on a handful of search engines and apply it across the
entire web. Today, the infrastructure is built to run impression-level targeting based on
a real-time auction for any website.
This is unquestionably an impressive technical achievement. Yet there is a growing sense that
this revolution has not delivered on its promise. Dramatic improvements in technical infrastructure
have yielded, at best, marginal improvements in campaign performance. In fact, outside of a
few small pockets, primarily in retargeting campaigns, the hope and the hype have yielded to
disappointment and even despair. A recent headline suggests we are entering The Trough of
RTB Disillusionment. The nervous whispers of industry insiders in 2011 threaten to become
the angry roar of advertisers.
But revolutions dont proceed in straight lines. A December 2000 BusinessWeek article on the
fledgling Google search engine posed the question, Wheres the business model? The article
quoted a Google competitor as saying, There isnt really good evidence, frankly, that companies
focused purely on search, as Google has been, can support themselves with that model. As if
anticipating the question, Google had launched AdWords six weeks earlier. The industry is now
living the revolution that followed. In retrospect, all revolutions seem inevitable. Beforehand,
all revolutions seem impossible. (Michael McFaul: NSC expert on Democratisation).
allow advertisers to mix and match data from various sources, combine
with algorithms and execute through any preferred bidding technology.
Instead of focussing on how to define your objectives we feel like its
important to understand some of the hypotheses out there and the
considerations for each:
1) Cookies are far shorter-lived than thought: We all know
that cookies dont live forever. But we built a real-time display industry
structure - separating data from optimisation and lower-purchase-funnel retargeting from upperpurchase-funnel prospecting - that implicitly assumes that cookies are, at least, reasonably longlived. As it turns out, the half-life of an average third-party cookie (the type most commonly used
for data exchange, ad serving and attribution analysis) is approximately three days, and cookies
for one third of online users last for less than an hour. Data buyers today are erring in one of
two ways. Theyre either focusing only on the subset of the population with stable cookies or
they are buying lists of cookies with a good chance that many on that list will never be found
again. Stale cookie lists equate to dramatic degradation in either effectiveness or efficiency. Said
another way, real-time bidding requires real-time data.
2) Clicks are a poor metric for display advertising: Real-time display is a new medium
requiring its own unique metrics. Importing the philosophy of accountability from search is
entirely appropriate, but lazily applying the same metrics is not.
ome internal research at Quantcast saw clickers to display ads skew either
S
very young (under 18 years of age) or older (more than 50 years of age) with
lower income (less than $30,000 annually). In general, the profile of converters
is the exact opposite of the profile of clickers, a finding consistent with a host
of display advertising studies. For example, comScores Natural Born Clickers
studies have shown that only 8 percent (and declining) of online users ever click
on an ad in a given month.
3) Prospecting and retargeting are not separate activities: Most advertisers
have been encouraged to split marketing objectives into upper and lower funnel.
This practice would be entirely rational if cookies were even reasonably stable.
But if the consideration window for an advertisers product is more than a day or
two, a single customer working his or her way down the purchase funnel from
awareness to consideration to purchase will likely appear as two, three or more
different cookies because cookies are so often short-lived. When that customer
is ready to visit an advertisers website, he or she will likely appear as a different
cookie from that appearing in the awareness or consideration phase. The
advertising that drove the person to visit the website the single hardest task
in advertising will not be properly attributed because cookie churn has broken
the causal link. Such mistaken attribution causes advertisers to over-invest in
retargeting and under-invest in the important work of finding new customers.
(More on attribution later in the handbook.)
4) Data is necessary but not sufficient: Comparing performance for campaigns run with
independent components to those that integrate data with algorithmic optimisation and bidding
strategy is interesting and working with a provider to do this can give you some valuable insights
(see example below). The data leveraged in both approaches was exactly the same, but the
divergence in performance was stark. Campaigns with integrated data, algorithms and bidding
produced a two to seven times lower cost per action (CPA) than the independent approach
(see below).
Quantcast comparative campaign performance
against four different sectors
Mix & Match
Integrated
Airline
Charity
50
CPA
40
30
20
10
0
Automotive
Hotel
Programmatic buying:
What you nee d to know
Shannon Sieve,
Head of AdX Buyer
Development, UK & Benelux
DoubleClick
Ad Exchange
How It Works
One of the key technologies enabling programmatic buying is Real-Time
Bidding (RTB), a buying mechanism based on data and algorithms that allows
media buyers to evaluate and purchase ad inventory on an impression-byimpression basis. With RTB, youre able to buy access to an audience rather
than just buying space on a site.
10
Heres what happens: when a user enters a URL into their browser, a message is sent to buying
agencies asking whether theyre interested in the ad space, along with key information such as
the cookie ID, ad size, users country, and more. The media buyers algorithms make a real-time
decision about the value of that single impression and bids on it accordingly. Once the auction is
won, the ad appears in milliseconds - all while the page is loading. The transactions take place
across ad exchanges and other platforms where inventory is made available in an open way:
impressions are won via an auction with no minimums, commitments or guarantees involved.
Messaging can now be personalised through Dynamic Creative.
Images can be tweaked, ads can be localised, and different products
can be dynamically displayed depending on users demographic, the
time of year, available supply and so on. Display ad units can contain
rich media, video, micro-sites, animations, and social aspects such
as Twitter and Facebook feeds. And the scope for user engagement
is bigger than ever: in addition to games and video, ads might
incorporate use of webcams, generate individualised responses, or
enable user-to-user communication - these days, our imaginations
have to catch up with technology, not the other way around. The
result is that users can be emotionally engaged like never before,
spending more time on your ad, increasing the chance of clickthrough, and generating a strongly positive association with
your brand.
11
What to Consider
1. Trigger
Consumer considers initial set
of brands according to brand
perceptions and exposure to
recent touch points
Initial
consideration
set
Repeat
purchase
Loyalty loop
Advocacy
Awareness
3. Moment purchase
Consumer selects a brand
4. Postpurchase experience
Following the purchase, consumer builds
expectations based on their experience, which
informs and affects the next decision journey
12
Data
Today its not just about buying impressions on a contextually-relevant page, its about finding your
exact audience within that context. Buying through exchanges and other open platforms allows
you to apply data to your media buys to cherry-pick the impressions you want and customise your
message based on anything from the users browsing behaviour to demographics to local weather.
Turning raw data into usable, actionable information is one of the most critical elements in the
world of programmatic buying. Its also one of the most challenging. Start by fully leveraging your
own first party data as it gives you rich information about how the user has interacted with your
brand and is uniquely your own information. Then explore the options for third party data sources.
Finding the right partner is crucial, so adopting a test and learn mentality is critical.
The data marketplace is not yet fully developed. Pricing varies greatly and some are surprised that
data can often cost more than the CPM itself, but its money well spent if performance justifies
it.And dont forget that performance can be measured in many ways - not just clicks.
Advertisers have the right to know where their ads are showing, and transparency is one of the
fundamental benefits of programmatic buying. Transparency in where your ads are running not
only provides comfort that your brand is appearing only in appropriate places on the web, but it
also allows for continued optimisation and stronger inputs into attribution modelling.
Direct-response or branding?
Programmatic buying should not just be seen as a Direct Response (DR) function. It can and should
be used across the spectrum. Given the complexity of the user journey, the very idea that media
budgets are either DR or brand is becoming less and less relevant. Every campaign has its own
performance objectives and metrics associated with it, and buying campaigns programmatically
can benefit all campaigns by making them more accountable, transparent and measurable.
The first thing a brand usually does when moving to programmatic buying is to
consolidate all remarketing through a single platform. This makes sense, since
it solves the duplication issue of buying across multiple platforms, enables a
frequency cap across users, and - because you have data that predicts how
likely each user is to convert - allows you to pay appropriately. This is much more
efficient than placing buys against the same user-list across multiple partners.
Beyond remarketing, you can use lookalike modelling against current users to
find others who are similar, and find prospects by buying audience segments from
data providers or publishers.
13
14
Social Networking sites garner almost one in five minutes spent online (ComScore, October
2011). Facebook recently launched the Facebook Exchange, where theyre making inventory
available to real-time bidding technologies.
Traditional media moving toward Connected-IPTV is also a medium waiting to be exploited.
Its possible to imagine a time when personalised ads will be shown via connected TVs and even
on digital out-of-home media.
Attribution
As mentioned previously, most advertisers have used a last click attribution model. In other
words, they have rewarded all of the credit for the sale to the last click. This ignores the impact of
all the other activities along the path-to-purchase that influenced the customers decision.
Many new attribution models are emerging, ranging from a linear model, where equal credit is
given to each interaction, to time decay, which attaches more credit to the touch points that
occurred nearest to the time of conversion. There is no set rule for attribution, as every journey and
product is different. However, making sure you understand these journeys is vital to ensuring you
dont under- or over-value elements of your online marketing.
The core issues that have historically prevented holistic attribution modelling
include lack of complete data, lack of transparency, and data being housed in
different places.Programmatic buying solves for the lack of transparency and
lack of complete data and a single, unified platform brings all the data to one place.
15
16
When you have a single platform, your programmatic buying becomes more powerful.Consider
the possibility of taking users who visited your website via a search ad, and re-targeting them with
a customised display ad, in real-time, seconds after they left your site. Using a single platform for
all your tracking will ensure a holistic view of every interaction on the path-to-purchase, whether it
came from SEM, Natural Search, Display, Affiliates or Social.
Right technology:
Sue Hunt,
Director, Right Media
EMEA at Yahoo!
17
18
19
20
Data Exchanges were created as marketplaces where Online Data Providers could sell their data
directly to DSPs and Ad Networks.
Who Uses: Ad Networks, DSPs.
Real-time bidding (RTB) - RTB is a protocol that enables the valuation and bidding on individual
impressions in real time. The buying takes place over online media exchanges basically media
marketplaces - which connect sellers (publishers) and buyers (advertisers).
Data Management Platform (DMP) - Platforms that allow advertisers, agencies, publishers
and others to control their own first-party audience and campaign data, compare it to thirdparty audience data, and give the ability to make smarter media buying and campaign planning
decisions via behavioural targeting or extending audiences via lookalike modeling. Advertisers
and agencies generally utilise DMPs in order to buy more effectively while publishers
typically utilise DMPs in order to segment their audiences and sell more effectively.
Ad Server - A computer that delivers and tracks advertisements
independent of the web site where the ad is being displayed. Use
of an ad server helps establish trust between an advertiser and
publisher since the statistics can be maintained by an objective
third party.
Who Uses: Advertisers, Agencies, DSPs, Ad Networks, Publishers.
Selecting a partner
It may seem simple, but in selecting a technology partner, the most
important thing to keep in mind is the principle requirement you
need them to meet. The due diligence and selection process can
quickly become complicated as you look into workflow, reporting,
service models, data and audience management, as well as core
reach. These considerations are further magnified if you are looking to
operate in multiple markets, and may also involve selection of more
than one partner.
There are a few key areas to bear in mind, and prioritise the importance
of each to your business:
21
22
BEEP! BEEP!
Workflow efficiencies
How much resource do you have available, versus your expectation of
technology driving efficiencies?
If you expect to source advertising from multiple exchanges and SSPs, then a
DSP will be the most logical primary technology for your media buying, offering
one entry point to a range of supply sources. This also changes the profile of
the resource you need to recruit, from the traditional planner/buyer, to a more
mathematical analyst and optimisation expert.
Service
What service model are you looking for? Are your aims to be self-sufficient or looking for a
partner to offer media buying and budget management services?
You may already have a clear ROI metric or prefer a 3rd party to prospect on your behalf, in
which case an exchange specialist or managed service from a DSP is worth considering, with no
additional investment in resource required. If you want to understand what drives your performance,
and have brand or regional variances, then look for a partner that
has resource in your core territories to service and support
local teams on the ground.
23
Reporting
What campaign management and optimisation metrics do
you expect to see? Do you have particular business or billing
requirements that need to be addressed? Do you plan to use
custom data feeds or an API?
As well as the traditional campaign and performance metrics of
pricing type, targeting parameters, pace of delivery and budget
management, technology platforms can also be used to drive
optimisation techniques. It is also important to ensure financial and
business intelligence core metrics are available from the platform.
As the volume of data can be extensive, a custom data feed or
access to the platform API is worth considering. This provides a
deeper and broader data set, enabling greater insight into all your
activity, informing bidding techniques to drive performance.
With so many business models and platforms available, the selection
process may feel long and complicated. Keep focus on your core
objectives, and clearly state your expectations, but remember the
importance of a partnership for successful future growth.
ADVERT
24
Conclusion
Amit Kotecha,
Senior Mobile and
Networks Manager,
IAB UK
25
DIRECTORY
AdMeld
Tom Jenen
Commercial Director EMEA
tjenen@google.com
Jemm Media
Julia Smith
Consultant
jsmith@jemmgroup.com
ad pepper Media
Henrik Kjaer
MD
Hkjaer@adpepper.com
Media Math
Erich Wasserman
Co-Founder, GM EMEA
erich@mediamath.com
Rubicon Project
Commercial Director, EMEA
Oliver Whitten
owhitten@rubiconproject.com
Adconion
Maria Cadbury
ADR UK MD
maria.cadbury@adconion.com
Microsoft
Scott Burford
3P Procurement Lead
scottbu@microsoft.com
Specific Media
Katie Field
VP European Publisher Services
kfield@specificmedia.com
AppNexus
Nigel Gilbert
Director of Sales, EMEA
ngilbert@appnexus.com
Quantcast
Phil Macauley
EU Managing Director
pmacauley@quantcast.com
Unanimis
Will King
Marketing Director
will@unanimis.co.uk
Google
Tanzil Bukhari
Head of Buyer Relations EMEA
tbukhari@google.com
Quisma
Ellie Edwards-Scott
Managing Director
ellie.edwards-scott@quisma.com
ValueClick Media
Elliott Clayton
Head of Sales
eclayton@valueclickmedia.com
Infectious Media
Zoe Steventon
Managing Director
zoe@infectiousmedia.com
Yahoo!
Steve Palin
Director, Media Buying EMEA
spalin@yahoo-inc.com
BEEP! BEEP!
iab
ADVERT