Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
THE NEW PLAYBOOK FOR SELF-SERVICE ADVERTISING:FIVE STRATEGIES FOR DSP MEDIA BUYING IN THE AGE OF AUTOMATION
2The New Playbook For Self-Service Media
Adopting a self-service media-buying platform
used to mean building deep competencies in highly
esoteric skills and disciplines. It used to mean serious
resources – people, hours, budget, time, effort, and
hassle. And it used to mean endless uncertainty – are my
campaigns performing as well as they could be? Am I
using the right tactics? Am I deploying the right data? Am
I bidding at the right levels? How do I truly make sense of
the billions of messaging opportunities and data signals in
the ecosystem? It’s 2 a.m. It’s Sunday. Are my campaigns
delivering? Are they performing?
That’s what adopting a self-service DSP used to mean.
In this guide, we’ll present some cutting-edge strategies
and methodologies being used by early adopters of the
next generation of self-serve media buying technology.
You’ll see that with the right tech and the right approach,
you can leave the uncertainty behind, put results first, and
still benefit from the control and agility afforded by self-
service. This is the new self-service playbook.
3The New Playbook For Self-Service Media
1
In this example, one Rocket Fuel advertiser, a nonprofit seeking sign-ups, started out their campaign with tactics that specifically targeted demographic groups.
A couple weeks in, the advertiser increased spend on unconstrained line items, and found that by allowing performance models to zero in on high-propensity opportunities at the impression level, results greatly improved. The advertiser continued to increase spend on the unconstrained model throughout the remainder of the campaign.
FOCUS ON OBJECTIVES – NOT TACTICS
PRESCRIPTIVE VS. OPTIMISED
€21.44
€42.21
Oct
0
10M
20M
30M
40M
50M
60M
Nov Dec
Demographic-constrained line items
Cost-Per-Signup – Compared
Impressions by Week
Unconstrained line items
Increased spend on unconstrained media relative to demographic targeting
4The New Playbook For Self-Service Media
Humans and hubris
Let’s start with a humble admission: Marketing, and
digital marketing, in particular, is no longer a human-scale
challenge. The emergence of programmatic media trading
has created opportunities for marketers large and small to
make their advertising more addressable, more precise, and
more efficient. In doing so, it has also made media buying
more complex than ever before. In 2015, there are tens
of billions of unique opportunities to reach consumers, on
millions of web properties, across a range of devices, with
trillions of ways to slice and dice an audience every day.
Many ad tech companies have built tools and platforms that
help humans parse this ever-more-fragmented landscape.
But even with armies of highly educated analysts, these
tools can’t live up to the true potential of real-time marketing,
because this first generation of platforms set out to equip
humans -- not algorithms -- to make most of the necessary
adjustments. Unfortunately, humans are severely limited in
how many adjustments we can make for a campaign, and
across how many attributes.
The belief that humans can adequately manage and
optimise a campaign persists because the ecosystem
matured long before there were sufficient tools to harness it.
And though humans can achieve outcomes that are better
than random, we cannot achieve true optimisation.
But today’s marketers and agencies – whether they know it
yet or not – have access to sophisticated technologies that
let machines do the heavy lifting without sacrificing control
or transparency.
Embracing automation
The key to succeeding with the next-generation demand side
platform (DSP) is focusing on strategy and goals, rather than line
items and tactics. Start by telling the system what your objectives
are and the KPIs you’ll use to measure them. Perhaps it’s
purchases, or clicks, or lead-form fills. Regardless, the most
crucial steps in taking advantage of the best optimisation
technologies are setting your goals and configuring your
feedback loop, typically in the form of pixels or beacons.
From there, consider what we’ll call inherent constraints, or
targeting that must be applied due to the nature of the offering
and the campaign. This might be geotargeting (In which
countries, states, and cities is the offering in question available?),
device types (Is the offering an app download available only in
iOS?), or a first-party audience segment (Is the objective of the
campaign to upsell or cross-sell existing customers?).
Beyond these inherent constraints, resist the urge to
turn assumptions into prescriptive tactics. We can define
assumptions, broadly, as any tactical direction that is not
an inherent constraint. Contextual targeting involves an
assumption. Behavioural targeting involves an
assumption. Dayparting involves an assumption. You or
your client may have invested heavily in research to
define just who your target audience is and to what they
are drawn. That’s great, and it should inform your
messaging strategy and your tactics in less-optimisable
media. But in digital ad buying, particularly for direct
response efforts, these are assumptions -- assumptions
that get in the way of truly effective marketing. Really.
This is because, with a next-generation DSP built for machine
learning, you have the luxury of testing all assumptions
automatically. The best way to start is typically from a place
of total neutrality. Define your objectives and your inherent
constraints, and let high-horsepower optimisation technology
take the handoff. Rather than letting a human paint in broad
strokes and hoping for the best, let the system evaluate
thousands of signals for each potential moment of influence
with a consumer while learning and adjusting on the fly.
Impression-level optimisation beats broad-brush optimisation
for a simple reason: At the impression level, a smart DSP
understands that what might look like a disqualifier (e.g. a
user identified as a man for a women’s magazine subscription
campaign) should actually be taken into context with all other
available signals. Maybe the man is in-market for a birthday
gift for his wife. Impression-level optimisation discerns the true
value of an impression for a campaign by considering the true
value of that moment with that individual. It recognises that
individuals are, well, individuals who are multi-dimensional, not
members of segments. It amounts to an information advantage
that prevents you from wasting dollars on poor messaging
opportunities and from neglecting moments that are worthwhile.
Embracing automation does not mean forfeiting control and
agility. It frees you up to focus that expertise on more strategic
tasks, while also driving the best performance possible.
5The New Playbook For Self-Service Media
Smartphone release
Goal: Drive purchase intent for new Android smartphone
Methodology: Advertiser used control and exposed groups with in-flight survey optimisation.
Outcome: 207% lift in purchase intent among 1.5 million consumers
BRING A PERFORMANCE MINDSET TO BRAND INITIATIVES
REDEFINE PERFORMANCE MEDIA
2
Denny’s
Goal: Deliver in-target impressions for their demographic
Methodology: Optimise for comScore-validated audience reach
Outcome: Delivered in-target at a rate 60% better than comScore’s benchmarks
207% Lift
Control Group (no ad exposure)
Consumer Purchase IntentPerforms 60% Better
Exposed Group (served product ad)
1.5 Million Consumers
6The New Playbook For Self-Service Media
Traditional Targeting Includes Wasted Impressions
Last year, Nielsen published a series of industry benchmarks showing that conventional targeting methodologies produced a profound volume of wasted impressions. For example, advertisers trying to reach the male 18-34 demographic were successful in doing so only 46% of the time. For females 35-54, Nielsen found a 28% success rate.
Programmatic came of age with direct response – driving
conversions, low-funnel clicks, etc. -- mostly because it offered
three things that are indispensable to direct marketers:
Addressability, measurability, and scale. Direct response and
programmatic seemed to be made for one another.
But somewhere, something obvious got lost in the
conversation: Brand marketers have objectives, too. And
now those objectives are supported by programmatic tools
that drive measurable, meaningful impact.
Reaching your audience:
In many upper-funnel marketing initiatives, reaching a
broad demographic group is an objective in itself. At face
value, this is easy to accomplish: buy some audience
segments and monitor delivery. But many are coming
around to the realisation that third-party demographic
data fails, often horribly, to accurately reach the intended
audience.
Getting to a better place on demographic targeting starts
with, once again, a focus on objectives and measurement. If
“optimising to a target audience” sounds oddly dissonant, it’s
because ad tech has for so long associated optimisation with
direct response goals. But when we treat reaching a target
audience as an outcome rather than a tactic, we can increase
our precision by leaps and bounds.
The right DSP will allow you to set an audience targeting
goal, perhaps even expressed as an in-target effective CPM
or cost-per-point, and then empower an optimisation engine
to study correlations between in-flight feedback from a third
party like Nielsen and thousands of available signals. Done
right, the result is astounding audience accuracy with verified
third-party reporting right in the UI.
Driving lift
Beyond reaching a demographic audience with precision,
brand marketers want to move the needle, too. While many
have long clung to clicks as the KPI of choice in programmatic
media, it’s been well documented that clicks do not tell the full
story. In fact, they often tell the wrong story.
Fortunately, the tools are now out there to combine powerful
automated media buying with metrics that brand marketers
actually care about. Specifically, attitudinal lift for measures
such as awareness, message retention, consideration,
purchase intent, or competitive preference.
In-banner brand studies have been part of the digital toolkit
for years. But by integrating them into a system that automates
real-time on a breadth of attributes based on survey samples,
we go beyond merely measuring these results.
By setting up two budgets -- one for a group exposed to
the advertiser’s message and one for a smaller control
group that is not -- the system can read lift signals across a
breadth of attributes and make automatic adjustments in-
flight. These adjustments allow self-service buyers to drive
meaningful, measurable lift.
7The New Playbook For Self-Service Media
MAKE YOUR DATA WORK FOR YOU
OPTIMIS ING WITHIN, BEYOND A SEGMENT
3
(Within a segment)
(Beyond a segment)
AUDIENCE A
AUDIENCE B
AUDIENCE C
AUDIENCE A
AUDIENCE B
AUDIENCE C
8The New Playbook For Self-Service Media
The migration of first-party customer data to the realm
of digital advertising represents one of the most exciting
developments in the field -- and in business. Data
management platforms (DMPs) and onboarding vendors
help centralize, taxonomise and enhance such data and
turn offline customer databases into online segments.
However, for all the advancements in making these
consumers identifiable and addressable, taking action
on the data in media execution leaves many marketers
scratching their heads.
Within the segment
It’s tempting to treat first-party customer segments like a
silver bullet for campaigns involving existing customers
(upsell, cross-sell, renewal, etc.): Set your DSP to target
these high-value consumers and reap the benefits. But in
reality, targeting on its own leaves value on the table. Even
within such a hyper-curated audience, there are plenty of
opportunities to get smarter, more responsive, and more
successful with your messaging.
Adding further targeting restrictions, such as a content
category or a third-party segment, to a first-party audience
typically constrains the volume too much to gain any real
benefit. And as we’ve established, it’s not the best way to get
performance in the first place.
By setting up a campaign to optimise within a first-party
segment—even a small one—you can skew your media
budgets to the higher-value individuals within that segment.
The machines find the right environments, times, messages,
sequence, and moment.
In other words, treat your first-party segment as a
narrower universe within which to optimise rather than a
bullseye for delivery.
Beyond the segment
Scale presents a central challenge for many marketers with
access to first-party data. Even for large enterprises, many
factors often conspire to whittle your audience down to a
fraction of its original size. But even a small first-party audience
can be transformed into a powerful asset.
Some platforms provide the opportunity to grow segments
using “look-alike” or “act-alike” optimisation. These
methodologies build statistical patterns to find similar
consumers outside of the segment based on commonalities
within it, looking at audience characteristics and actions taken
during a campaign among the original segment (browsing
patterns, ad engagement, site engagement, etc.) and finding
more consumers demonstrating congruent behaviour.
Let’s say a consumer “in the wild,” or outside the original
segment, matches on over-indexing for 15 specific
attributes with the original segment, including consumption
of sports content, purchase history in a frozen food CPG
category, and Android phone usage. You’ve identified this
consumer as someone with a lot in common with your
existing customers, which makes her a high-value prospect
to nurture toward conversion. Ideally, you wouldn’t have
to do this at all, because predictive modeling takes it on
for you at the impression level, in real time, crunching the
numbers and adapting.
9The New Playbook For Self-Service Media
ACCESS THE MARKETING LABORATORY4
10The New Playbook For Self-Service Media
DSPs have always supported reporting on campaign
delivery and performance. Some have offered analytics
tools that shed light on the best-performing audiences,
environments, and messages. This guidance has long been
seen as a highly valuable roadmap to the smartest tactics
that a campaign manager can implement mid-flight.
But as we’ve covered, an advanced platform will make
those adjustments for you faster, more granularly, and more
often that you ever could.
So what role does analytics play in a next-generation
approach to digital media buying? Oddly enough, it has less
to do with digital than traditional media. In places where
optimisation is not so readily available (think print, broadcast,
outdoor, and your messaging strategy), these learnings can
and should inform your tactics. Having machine-learning
insights, generated by objective measurement at big data
scale, at your disposal can greatly improve outcomes on
channels that do not lend themselves to adaptive learning.
INSIGHTS
You may find that many of your assumptions are confirmed.
Perhaps your assumed age and gender group align exactly
with the best-performing group in the analysis. But you
should still be glad you did not constrain to it for the whole
campaign because the system recognises that there are
good impressions that don’t meet those conditions, and
bad impressions that do.
You’ll also find many insights that do not align with your
assumptions. Embrace the non-intuitive. A high-volume
digital campaign with analytics tools that study patterns
and correlations across millions of impressions can
lead to learnings that would never emerge from a focus
group. And they can create significant lift elsewhere
when properly utilised.
In all of this, it’s easy to see the beauty of automation.
Rather than trying to teach the machine how to do its
job, the machine leaves no stone unturned to teach us
what it finds.
In this example, a campaign for a luxury automaker turned up some non-intuitive top site performers.
An airline running a bookings campaign found that the system identified some high-indexing consumer interest categories that one wouldn’t think of, like Father’s Day Shoppers and Cold & Flu.
Summer Seasonal - Fathers Day Show
Entertainment - Movies
Winter Seasonal - Cold and Flu
Spring Seasonal - Easter Eggs
Entertainment - Video Games
Job Seekers
Parenting - Teens and Tweens
Entertainment - Humour
-56.6%
-73.8%
-80.0%
-81.6%
97.2%
81.3%
64.9%
56.6%
Interests
11The New Playbook For Self-Service Media
ENTER THE AGE OF OPTIMISATION WITH ROCKET FUEL DSP
5
A Look InsideSearch Optics is a digital marketing company that specialises in custom, integrated solutions with an emphasis on return on investment. The company uses an uncommon blend of leading edge technology and human interaction to turn virtual traffic into measurable results.
“The Rocket Fuel DSP allows us
to be hands-on with the media
but not so in the weeds that we’re
constantly managing bids,” says
Senior Display Media Manager
Alex Meakin. Once campaigns
started to scale, the performance
benefits became clear. “Rocket Fuel
definitely wins on performance. And
I feel confident about the quality of
impressions and engagement we’re
seeing.” Indeed, Search Optics has
seen performance consistently
improve as Rocket Fuel’s models
have become “smarter” about
campaign objectives, with
aggregate indexed campaign
performance against client goals
reaching upwards of 1,000%.
12The New Playbook For Self-Service Media
Sophistication is not thousands of knobs and levers.
Sophistication is tech that makes knobs and levers
obsolete. Rocket Fuel DSP is the first ever low-touch, high-
performance media buying platform, driven by the one of
the most powerful performance engines in the industry.
It drives unprecedented results by bringing order to the
chaos of the digital landscape and eliminating tedious
guesswork for better marketing outcomes and a lighter
operational burden.
Unmatched performance
Carry out high-performing digital campaigns across all
channels (desktop and mobile display, desktop and mobile
video, social) and throughout the marketing funnel. Rocket
Fuel’s Moment Scoring™ technology learns the critical—
and often hard to recognise—predictors of what makes
one ad more appropriate than another in a particular
moment based on a marketer’s goal.
Tools that get out of your way
Assimilate what you know into campaign setups and
let the models handle every tiny detail you couldn’t
possibly anticipate. Manage your creatives,
campaigns, line items, and tactics with ease and
launch brand new campaigns in a matter of minutes.
Insights for action
Rocket Fuel DSP’s analytics tools translate machine
learning into clear, actionable reports. These custom
reports, in millions of potential configurations, show how
Moment Scoring is identifying the attributes and variables
for each ad opportunity that improve engagement and
lower CPA. For example, a report called ‘Model Learnings’
is a visual representation of how each attribute is driving
down CPA, real-time insights that help you discover all new
things about your customers.
Average Campaign Performance Against Goal
Oct Nov Dec Jan Feb Mar Apr May
0.00%
200.00%
400.00%
600.00%
800.00%
1000.00%
1200.00%
Weighted by spend
Let’s get started
The next generation DSP replaces the complexity
of knobs and levers with the simple sophistication
of machine learning at big data scale. It taps the full
marketing potential of your organisation’s data
while supporting the business acumen that put you
at the helm of your marketing initiatives.
Talk to your Rocket Fuel account manager or contact [email protected] to learn more about Rocket Fuel DSP today.
About Rocket Fuel
Rocket Fuel operates in more than 20 offices worldwide and trades on the
NASDAQ Global Select Market under the ticker symbol “FUEL.” For more
information, please visit www.rocketfuel.com/UK or call 020 3651 1300. Fuel
can unleash Moment Scoring’s full potential to each valuable moments, on
whichever device and channel it may be and for whichever marketing goal it
may best serve.