ATTRIBUTION FOR TV ADVERTISING June 20, 2019 Presented By: BY THE - - PowerPoint PPT Presentation

attribution for tv advertising
SMART_READER_LITE
LIVE PREVIEW

ATTRIBUTION FOR TV ADVERTISING June 20, 2019 Presented By: BY THE - - PowerPoint PPT Presentation

ATTRIBUTION FOR TV ADVERTISING June 20, 2019 Presented By: BY THE END OF THE SESSION YOU WILL HAVE A step by step guide on measuring attribution for your TV buy. An understanding of attribution terms and concepts. A perspective on


slide-1
SLIDE 1

ATTRIBUTION FOR TV ADVERTISING

June 20, 2019

Presented By:

slide-2
SLIDE 2

BY THE END OF THE SESSION YOU WILL HAVE

  • A step by step guide on measuring attribution for your TV buy.
  • An understanding of attribution terms and concepts.
  • A perspective on the relationship between attribution and optimization.
  • Examples of how real brands measure attribution for their TV buys.
slide-3
SLIDE 3

WHAT WE WILL COVER IN THIS SESSION

INTRODUCTION

  • Defining attribution
  • Class framework

ATTRIBUTION IN DIGITAL ADVERTISING

  • Digital tracking
  • Digital attribution

ATTRIBUTION IN TV

  • Current state
  • Tracking TV
  • Attribution for TV
  • Four use cases

TWO BRAND’S STORIES

slide-4
SLIDE 4

WHAT DOES PHONE RELIEF HAVE TO DO WITH ATTRIBUTION?

slide-5
SLIDE 5

INTRODUCTION TO ATTRIBUTION

LESSON 1

slide-6
SLIDE 6

WHAT IS ATTRIBUTION?

Attribution is the process of:

1. Identifying the marketing events that contribute to a 2. User taking a desired business outcome 3. And assigning value to each of these events

slide-7
SLIDE 7

ATTRIBUTION IN DIGITAL ADVERTISING

LESSON 2

slide-8
SLIDE 8

DIGITAL MEASUREMENT

There two main ways we can measure ad exposure and business

  • utcomes in digital
  • Pixel: Code on a website that provides notice that a user was on the page.

This notice is sent to a cookie.

  • Cookie: Able to track users across sites and send tracking data to an ad server.
slide-9
SLIDE 9

IDENTITY MANAGEMENT

There are two methodologies to track and measure consumers across device

  • Deterministic: Using information such as an email address or login

to track consumers across device.

  • Probabilistic: Using billions of anonymous data points to make a

best guess as to which devices may belong the same user.

slide-10
SLIDE 10

ATTRIBUTION IN TELEVISION

LESSON 3

slide-11
SLIDE 11

MOVING TO ADVANCE MORE METHODS

  • MEDIA MIX MODELING (MMM) is a statistical analysis that estimates the impact of various marketing tactics on sales

and then forecast the impact of future sets of tactics while controlling for a number of factors.

  • MMM USE NIELSEN DATA AS AN INPUT. From the Nielsen’s People Meter, which electronically captures all viewing

from their nationally projectable sample of panelists.

  • GRP: The percentage of an audience that watched a particular spot x 100. Example – A program reaching 25M out of

100M women 18-49 reaches 25% of the audience. If the ad was delivered 4 times, it would result in 100 GRPs.

  • MMM MEASUREMENT: MMM measures bases sales, the sales that occur with no marketing activity and incremental

sales - The amount of additional unit sales that occur as a result of marketing activities.

  • MMM IS ALSO USED TO DEVELOP RESPONSE CURVES: which help brands understand points of diminishing returns

for their spend.

slide-12
SLIDE 12

MEASURING AD EXPOSURE: NIELSEN POST REPORT

4 : 4 5 : P M 4 : 5 : P M 4 : 5 5 : P M 5 : : P M 5 : 5 : P M 5 : 1 : P M 5 : 1 5 : P M 5 : 2 : P M 5 : 2 5 : P M 5 : 3 : P M 5 : 3 5 : P M 5 : 4 : P M 5 : 4 5 : P M 5 : 5 : P M 5 : 5 5 : P M 6 : : P M 6 : 5 : P M 6 : 1 : P M 6 : 1 5 : P M

X

Walking Dead, AMC

Report will show when a spot ran, on what network, in what pod, position, etc.

slide-13
SLIDE 13

MEASURING AD EXPOSURE: SET-TOP BOX DATA

Set-top box captures and stores viewing data.

slide-14
SLIDE 14

MEASURING AD EXPOSURE: AUTOMATED CONTENT RECOGNITION

ACR is becoming the standard method of measurement for TV ad exposure.

slide-15
SLIDE 15

ATTRIBUTION PLAYBOOK

q Collect exposure data via Nielsen post reports, set top box or ACR: will come via a direct relationship or via an attribution vendor.

slide-16
SLIDE 16

WHAT ARE EXAMPLES OF BUSINESS OUTCOMES?

slide-17
SLIDE 17

ATTRIBUTION PLAYBOOK

q Collect exposure data via Nielsen post reports, set top box or ACR: will come via a direct relationship or via an attribution vendor. q Make sure you know what you want to measure and have the tracking in place to do

  • so. It’s best to start with upper funnel metrics.
slide-18
SLIDE 18

ONLINE–SPIKE VISITOR ANALYSIS

Step 1: Through a web analytics system, analyze filtered traffic on an hour-by-hour basis.

slide-19
SLIDE 19

ONLINE–SPIKE VISITOR ANALYSIS

Step 2: Overlay the spots that ran during the time period.

slide-20
SLIDE 20

Step 3: Develop a mean by calculating web traffic on a minute-by-minute basis.

ONLINE–SPIKE VISITOR ANALYSIS

slide-21
SLIDE 21

Step 4: Using a 95% confidence level, determine an upper and lower bound to account for variance and uncertainty in the data.

ONLINE–SPIKE VISITOR ANALYSIS

slide-22
SLIDE 22

ONLINE–SPIKE VISITOR ANALYSIS

Step 5: Measure the number of visits during that spike and back into a cost per visitor metric.

slide-23
SLIDE 23

ATTRIBUTION PLAYBOOK

q Collect exposure data via Nielsen post reports, set top box or ACR: will come via a direct relationship or via an attribution partner. q Make sure you know what you want to measure and have the tracking in place to do so. It’s best to start with upper funnel metrics.

q Website Visitation: Track users pre and post spot to understand how running a spot contributes to web traffic. Develop a baseline and measure incrementality relative to the upper bound of that range. Measure incremental traffic before it reverts to the mean.

slide-24
SLIDE 24

ONLINE—SPIKE SALES ANALYSIS

  • Use set top box or ACR data set to get the IP addresses of those exposed to your ad.
  • Use weblog data to find the IP addresses of users who have purchased on your site over a

set time period.

  • Match those two data sets to find the users who were exposed to an ad and also

purchased.

  • Create a control group to validate your model by using random subsets of the overall
  • sample. Make sure your control group mirrors your exposed in every way.
  • A / B test for variables such as creative, network and daypart.
  • Calculate a return on ad spend (ROAS) based on the cost of the spot and number of

incremental users who purchased on your site.

slide-25
SLIDE 25

ATTRIBUTION PLAYBOOK

  • Collect exposure data via Nielsen post reports, set top box or ACR: will come via a

direct relationship or via an attribution partner.

  • Make sure you know what you want to measure and have the tracking in place to do so.

It’s best to start with upper funnel metrics.

q Website Visitation: Track users pre and post spot to understand how running a spot contributes to web traffic. Develop a baseline and measure incrementality relative to the upper bound of that range. Measure incremental traffic before it reverts to the mean. q Website Sales: Track site visitor over time to understand those who purchase. Test different networks, dayparts and creatives.

slide-26
SLIDE 26

TV’s IMPACT ON IN-STORE VISITS

  • Use any of the following to track instore visits
  • Geolocation Partners: Work with multiple apps to integrate into their

Software Development Kit (SDK)

  • Bluetooth Beacons: Small devices that track users within a retail location
  • Client App Data: 1st party app data can also be matched to identity management partner
  • Match this to an identity partner, who then can match to ad exposure
  • Develop a cost per in-store visit metric.
slide-27
SLIDE 27

ATTRIBUTION PLAYBOOK

q Collect exposure data via Nielsen post reports, set top box or ACR: will come via a direct relationship

  • r via an attribution partner.

q Make sure you know what you want to measure and have the tracking in place to do so. It’s best to start with upper funnel metrics. q Website Visitation: Track users pre and post spot to understand how running a spot contributes to web

  • traffic. Develop a baseline and measure incrementality relative to the upper bound of that range. Measure

incremental traffic before it reverts to the mean. q Website Sales: Track site visitor over time to understand those that purchase. A/B test different networks, dayparts and creative. q In-store Visitation: Using Geolocation, Beacon or Identity Management partners, track users who enter your retail location and then track that back to an ad exposure

slide-28
SLIDE 28

TVs IMPACT ON IN STORE SALES

Use any of these methods to track in-store sales

  • A CRM list
  • Companies such as Nielsen Catalina, which use purchase data and loyalty cards.

(Also IRI, Cardlytics and Shopcom)

  • Credit card companies, such as AMEX, Visa and Mastercard
  • Use an identity management partner to onboard this data and match to

ad exposure via ACR.

  • Develop a cost per in-store sales metric
slide-29
SLIDE 29

ATTRIBUTION PLAYBOOK

q Collect exposure data via Nielsen post reports, set top box or ACR: will come via a direct relationship or via an attribution partner. q Make sure you know what you want to measure and have the tracking in place to do so. It’s best to start with upper funnel metrics. q Website Visitation: Track users pre and post spot to understand how running a spot contributes to web

  • traffic. Develop a baseline and measure incrementality relative to the upper bound of that range. Measure

incremental traffic before it reverts to the mean. q Website Sales: Track site visitor over time to understand those that purchase. Use this as an opportunity to A/B test different networks, dayparts and creative. q In-store Visitation: Using Geolocation, Beacon or Identity Management partners, track users who enter your retail location and then track that back to an ad exposure. q In-store Sales: Use Identity Management partners who are integrate with Credit Card Parkers, Nielsen and

  • thers to track in store sales.
slide-30
SLIDE 30

OTHER ITEMS TO WATCH:

  • Match rates of identity management partners
  • Looming privacy issues
  • Fraud and invalid traffic
  • Deduping across multiple data sources
  • Be cognizant of the different types of TV
slide-31
SLIDE 31

THANK YOU!