prospecting pov

12
PROGRAMMATIC PROSPECTING CREATED BY: Eric Fung, Mark Kennedy, Nina Van Brunt, Robert Campos and Yana Skakun The prospect for a successful digital media strategy lies in clearly defining the campaign objectives – and then building the right audience to meet those goals.

Upload: nina-van-brunt

Post on 12-Apr-2017

238 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Prospecting POV

XX PROGRAMMATIC PROSPECTING

PROGRAMMATIC PROSPECTING

CREATED BY:

Eric Fung, Mark Kennedy, Nina Van Brunt, Robert Campos and Yana Skakun

The prospect for a successful digital media strategy lies in clearly

defining the campaign objectives – and then building the right

audience to meet those goals.

Page 2: Prospecting POV

XX PROGRAMMATIC PROSPECTING

Using Data More Efficiently

Navigate through a typical browser session and you may not be aware that behind the scenes, a

network of ad-tech partners are aggregating, parsing, slicing, unifying and packaging browsing

data to create custom user profiles. One user’s proclivity for specific retail sites might place them

in a “frequent shopper” segment, while another’s recent flight search identifies them as “in-market

for travel.” These ad-tech partners are known as data providers, and the custom profiles they

create are targetable audience segments.

Marketers are able to target these audience segments using an upper-funnel marketing tactic

called prospecting. Specific to the programmatic space, prospecting leverages third-party data

to define and reach custom audience segments that represent potential new customers sharing

similar attributes. Using prospecting, marketers are able to target these segments to encourage a

specific activity and create more effective and efficient digital marketing campaigns.

As the execution arm for programmatic buying, analyzing and informing thoughtful data usage,

VivaKi has a stake in the effective use of data to build new audience segments. While there are

a number of key players in the audience data space, differentiation is measured on the merits

of a player’s data-collection methodology, ability to connect across data sets and particular

DSP integration capabilities for turnkey activation. To better serve VivaKi’s agency partners, we

looked at a few select DSPs to assess the value of their audience data, determine the best ways to

maximize their potential and, finally, to determine how to use data more efficiently.

Page 3: Prospecting POV

03 PROGRAMMATIC PROSPECTING

The Value of Prospecting

When used appropriately, prospecting can add value to most campaigns.

For direct-response campaigns, it is especially effective when run in

tandem with lower-funnel tactics, such as remarketing – the practice of

reaching a user who has previously engaged with the brand – as it adds

incremental, qualified users while keeping the overall audience fresh and

raising the probability for conversions. Prospecting also helps increase

click-through-rates (CTRs), conversion rates and scale.

ADD MORE QUALIFIED USERS =

HIGHER CONVERSION RATE

When VivaKi evaluated the

purchase conversion rate of a

Telecom client, the purchase

conversion rate was 64%

higher when users were touched

by both upper- and lower-funnel

targeted impressions, rather than

just remarketing alone.

In the most general sense, prospecting is the development of qualified, new customers that have

been identified as able to achieve a specific goal, such as purchasing a product or a service. In

programmatic advertising, prospecting is further defined as using third-party data to target and

reach users with ads that encourage a specific activity.

The information used to create third-party segments is collected through a wide spectrum

of methodologies (i.e., modeling and registration-based data) and sources (i.e., browsing

behavior, point-of-sale purchase, etc.). These segments can be layered with additional broader

segmentation, such as demographic, behavioral or offline on-boarding. The information used for

prospecting is not owned by the advertiser and typically comes with usage fees.

Page 4: Prospecting POV

XX PROGRAMMATIC PROSPECTING

Historically, prospecting offered little trackable benefit for brand campaigns — traditionally

those in the CPG vertical —because users were rarely making purchases directly online.

With the advent of onboarding offline data, however, CPG advertisers are now able to

communicate with offline buyers online, closing the loop to understanding the effects of

online advertising on in-store sales.

While prospecting’s biggest appeal lies in an ability to broaden an advertiser’s pool of

potential consumers, there are instances where it might not benefit a campaign. If a client

already has an extensive user list (and therefore a robust existing remarketing pool) or a

very specific goal in mind, prospecting might not add much value. Prospecting also might

not be beneficial if the campaign budget is low and can only accommodate one strategy,

or if a client has another partner already running prospecting.

The key to implementing an effective prospecting strategy is understanding the marketing

objectives prior to a campaign’s launch and aligning the audience to meet those

objectives. In the diagram below, we’ve outlined a sampling of marketing objectives and

aligned them with proposed data segments. For instance, advertisers wishing to prompt

users whom they are not currently reaching, one strategy may be to target “in-market”

customer data sets. Another strategy may involve reaching potential customers while

they’re browsing the broader category product offerings.

Regardless of the marketing objective, it’s important for all campaigns to optimize

from the bottom up. Because prospecting is meant to support lower-funnel

conversions, campaigns should have a robust remarketing strategy before activating

prospecting budgets.

Aquire Customers

Target known category buyers or those in-market

Drive Trial

Target category buyers and brand buyers

Grow Share

Target top competitors (conquest)

Retain

Target first-party data and loyal brand buyers

Build Loyalty

Target existing brand buyers

Win Back

Target lapsed brand buyers

Page 5: Prospecting POV

05 PROGRAMMATIC PROSPECTING

A Provider Perspective

Access and integration are crucial to activating prospecting as a tactic. Due to the richness and

availability of the data, VivaKi analyzed the display channel and two main DSPs: Turn and DBM.1 To

tackle access, we created a matrix showing which data information is available within these DSPs.

1 For all of the advertisers included in the Telecommunications and Finance & Insurance

verticals, 90% of advertisers have spent on at least one of the two platforms.

Page 6: Prospecting POV

06 PROGRAMMATIC PROSPECTING

Clearly, DBM has a much larger footprint and is integrated with many more data providers than

Turn. We took this into account when analyzing and comparing performance. Although we were

not able to directly compare the overlapping 13 providers due to the difference in data sets,

such as unique (Turn) vs. non-unique (DBM) data, we did complete a qualitative review of the

providers.

Each platform has certain suggestions when it comes to selecting prospecting segments.

However, both advise testing multiple similar segments within different data providers to find

the optimal solution for a campaign. When a campaign is already live within Turn, the DSP

recommends using their proprietary audience extension tool (AET), which is located within the

UI and identifies the segments indexing the highest (relative to other segments), based on the

past two-week’s worth of data. DBM, on the other hand, warns that segment selection “is highly

dependent on the campaign objective.” Although we completely agree with that statement,

Turn’s AET is an effective way to identify above-average-performing segments.

Neither DBM’s nor Turn’s algorithms work differently between strategies, that is, between

prospecting and other strategies, such as remarketing and filtering. Indeed, both parties agreed

their algorithms “essentially work the same.” They similarly state their algorithms evaluate a

user’s value and whether or not specific users meet target audience requirements. The results

for both are directly related to the audience size as well as the structure, campaign set-up and

objectives and KPIs.

Page 7: Prospecting POV

07 PROGRAMMATIC PROSPECTING

The Measure of Success

Strategically, prospecting is a different data set than remarketing, which makes use of a marketer’s

first-party data, and should not be held to the same success metrics. Instead of efficient

conversions, for instance, a value-additive prospecting campaign should deliver incrementally

at an appropriate price. To evaluate the dynamics of prospecting incrementally, the measures of

assessment should be: media efficiency, quality and scale. A prospecting campaign that is able to

deliver on the intersection of the three metrics is ideal.

Media Efficiency Scale

Quality

In the case of media efficiency, or price, CPMs can vary from $.25 to $3, and advertisers should

determine a break-even point where additional data cost is not worth the incremental conversions.

A flat $.50 data fee may be worth the additional cost on a $6 CPM campaign but becomes less

worthwhile for a $1.50 CPM media plan. Additionally, some DSPs will offer a percentage of media

pricing. This option is especially attractive for low-CPM media plans, which generally want more

budget allocated for media. (Within the VivaKi data landscape, The Trade Desk is the only DSP to

offer percent of media pricing within its Data Alliance.)

Quality should be evaluated through media metrics like conversion rate or CTR, depending on

the campaign KPIs, as well as volume of uniques driven into the remarketing pool. At face value,

the cost-per-acquisition (CPA) for a prospecting strategy will never be comparable to those of a

remarketing strategy. In many instances, the qualified users that were found via prospecting will

go into the remarketing pool either by visiting the advertiser’s website later or clicking through to

the advertiser’s landing page. From there, the user will fall into the remarketing pool and begin to

be exposed to remarketing impressions. In both cases, if the user eventually converts, remarketing

receives credit in a last-touch attribution model.

Page 8: Prospecting POV

XX PROGRAMMATIC PROSPECTING

Taking things a step further and evaluating user-level data, prospecting efficacy

should be measured by how many users were funneled into the remarketing pool

and then eventually converted. This can be achieved by tracking the user funnel and

calculating conversion rates for three types of users: prospecting-only, prospecting-

then-remarketing and remarketing-only users. Evaluating past campaigns, users

who are shown both prospecting and remarketing ads have comparable or better

conversion rates than users who are only shown a remarketing ad.

Prospecting user pools are, unfortunately finite, and the ability to scale is an important

consideration when understanding how to select a data segment. The primary values to

evaluate scale are segment uniques, which are readily available in the Turn and DBM DSPs.

For this study, we combined these elements to create one robust metric to account for price and

quality. This new metric, which we identified as Cost Adjusted Conversion Rate (CACR), calculates

how conversion rates will adjust when price is included. A high CACR score (above 100) indicates

that the segment meets both relevancy and value considerations. In the example below, the two

segments have similar conversion rate indices, but the second segment has a lower price and,

therefore, a better CACR score. Extrapolating to a sample data provider dataset, the chart shows

how the stack ranking changes when CACR versus conversion rate is the new evaluation metric.

For example, AddThis jumps from third position to first, due to its relatively lower price point, while

Datonics drops to eighth because of its high CPMs.

PROSPECTING CONVERSION RATES

Conversion rates for a Travel

client were compared across

2 geos, with prospecting

to retargeting converters

having a comparable or

higher conversion rate than

Remarketing alone.

Page 9: Prospecting POV

09 PROGRAMMATIC PROSPECTING

Selecting the Right Segments

The reports that inform segment selection are all based on an underlying “continuous indexing”

feature of both DBM and Turn. The continual reporting features enable the platforms to record

which segments or, more likely, several of these data segments an impression (and its associated

click and conversion data) falls into. The data collected by these features is typically accessed

by two types of reports that aim to frame this data in order to assist data segment selection: the

audience skew report and the look-alike report.

An audience skew report is a register of the live data,

showing which data segments each impression, click

and conversion recorded by the DSP falls into for a

respective campaign, whether or not the segment was

targeted. This report usually includes the majority of

the targetable data segment universe and, as such,

approximates how a segment would have performed

(and how it would likely continue to perform after

targeting). By adding conversion rate, CTR and CACR

calculations for each segment, this report easily and

intuitively informs data segment selection on a cost-

adjusted basis. When using an audience skew report,

it is important to filter results for an appropriate

impression threshold (usually the top two quartiles of

impression volume), as some segments may have very

few impressions associated with them due to stringent

underlying data collection methodologies.

Audience Skew Report

A look-alike report predicts additional data segments

that may perform well for a campaign by taking an

existing audience (usually a remarketing pool) and

finding segments with high audience overlap. This

overlap is quantified by a “match ratio.” The match

ratio describes how much more likely it is for a cookie

from the new segment to be a member of the existing

audience than it is for a cookie chosen at random to be

a member of the existing audience. Data segments with

high match ratios will best approximate the existing

audience and, if large enough to be scalable, would be

good candidates to run a campaign against. It should

be noted that due to privacy considerations, look-alike

reports cannot display data for targeting combinations

that match fewer than 3,000 unique cookies.

Look-Alike Report

Ideally, the two reports should be used in tandem when activating Prospecting budgets: the

look-alike report off of a prior placed pixel to understand overlap and then the audience skew

report to evaluate “in-market” conditions.

Page 10: Prospecting POV

10 PROGRAMMATIC PROSPECTING

Best Practices for Prospecting

At VivaKi, we’ve developed a few best practices to work to include – but not be limited by –

when executing prospecting, regardless of DSP.

Exclude first-party lists. This will ensure you find new and unique users.

Maintain strict frequency capping.Maximize your budget in terms of delivering unique impressions.

Start broad and narrow as you go. Initiate campaigns with more segments and work your way down to find the best-performing ones,

eliminating poor-performing segments, while increasing budgets on high-indexing ones.

Employ recencies.Reach users who have freshly entered the prospecting pool.

Through the use of third-party information from outside data providers, prospecting can be

a valuable tool to discover new customers and increase a campaign’s overall effectiveness. To

achieve its full potential, it is important to remember that prospecting, as a tactic, is meant to

support lower-funnel conversions and should be paired with a robust remarketing strategy.

As with most marketing approaches, prospecting’s success lies in the ability of the campaign

creator to identify and understand a campaign’s objectives prior to launch, and then aligning the

audience to meet those goals.

Page 11: Prospecting POV

11 PROGRAMMATIC PROSPECTING

In her role as Analytics Team Lead, Yana oversees reporting and optimization strategies for her

clients, which include accounts in the Insurance, CPG and Financial verticals. Since joining the

Analytics team in 2012, Yana has been a part of several projects to streamline insights, including

integration of ad-server data. She is currently working to operationalize an automated daily

performance tracker. Before making her foray into digital analytics, Yana worked for the NY

Federal Reserve in analyzing the lifecycle of sanctions and anti-money laundering investigations.

Yana graduated from Duke University in Durham, NC, with a bachelor’s degree in in Economics.

Yana Skakun

Nina Van Brunt stepped into her Analyst role with full force, learning the digital

space, training pod members and leading her pod in client communication. She

is the Video Channel Subject Matter Expert, serving as the liaison between the

VivaKi Video Team and the Analytics Team, facilitating communication, education

and optimal workflow between the two. She also spearheaded the effort to

integrate specific video DSP data into VivaKi’s proprietary SkySkraper database.

Nina graduated from Boston University with a BS in Film.

Joining AOD Analytics in March of 2012, Eric has worked on accounts across a wide range of

industry verticals, including Telecom, Consumer Electronics, Travel and Hospitality, Financial Services,

Business Services, CPG and Retail. In his role as a Senior Analyst, Eric contributes to both account

work and special projects to further AOD thought leadership and streamline business operations.

Prior to joining AOD, Eric worked in the private equity industry, managing private placements of new

funds and sales of limited partnership holdings on the secondary market. Eric received his bachelor’s

degree in Commerce, Organizations and Entrepreneurship from Brown University.

Eric Fung

Nina Van Brunt

Mark joined the Analytics team two years ago to work on key accounts in the CPG and

Financial verticals. His experience with direct-response, awareness and branding campaigns

has provided for a solid background in the digital media ecosystem. Mark is currently

working on creating a targeting strategy utilizing a DMP to garner deeper consumer trends.

Prior to joining VivaKi and the programmatic landscape, Mark worked as an analyst at a

privately held retail company. Mark received his bachelor’s degree in International Business

Management from Boston University.

Mark Kennedy

About the Authors

Page 12: Prospecting POV

XX PROGRAMMATIC PROSPECTING

VivaKi’s Analytics Team exists to not only evaluate what they

find in programmatic campaign data, but why. This dedicated team of expertly

trained analysts mines the wealth of data that’s been collected by VivaKi’s data

solution for clients across all programmatic channels to discover actionable

insights about audiences and inform optimization strategies. Their efforts result in

the development of best practices that advance the intelligence of VivaKi and, in

turn, agencies and their clients.

About VivaKi

Analytics