Advances in public relations measurement

© Geddes Analytics™ 2012. Not for distribution without permission.

Advancing Public Relations MeasurementSeptember 19, 2012

David Geddes, Ph.D.

Managing DirectorGeddes Analytics LLC

Chair, Institute for PR Measurement Commission

2

© Geddes Analytics. Not for distribution without permission.

Agenda

Part I: PR measurement today▫Standards

▫A standard framework

Part II: PR measurement tomorrow▫Predictive analytics

▫Building organizational value

3

© Geddes Analytics. Not for distribution without permission.

A standard framework1 Begin with theory

4

© Geddes Analytics. Not for distribution without permission.

Message sender

Send

Originate

Encode

Transmission mediumor channel

1

2

3

Message recipient

Receive

Understand

Decode

3

2

1

Noise

N N

N

5

© Geddes Analytics. Not for distribution without permission.

A standard framework2 Translate to practice

Activities

Outputs

Reception

Outtakes orPR outcomes

6

Send

Originate

Encode

Transmission

channelReceive

Understand

Decode

Recip

ien

tS

en

der

Act Business results

7

© Geddes Analytics. Not for distribution without permission.

Activities

•What did you do?

•Control by PR

•Visible

•Operational efficiency▫Staff time

▫Budget

•Use: PR group, CFO

Outputs

Reception

Outtakes

Comms results

Activities

Business results

8

© Geddes Analytics. Not for distribution without permission.

Outputs

•Message availabilityto target audiences▫Media analytics

▫Social media analytics

▫Other (events, etc.)

•Correlate with activities

•Use: Within PR group

Reception

Outtakes

Comms results

Activities

Business results

Outputs

9

© Geddes Analytics. Not for distribution without permission.

Reception

•Handle PR outputs

•Manipulate outputs

• Involvement with outputs

•Correlate with activities

• Is this engagement?

•Use: Within PR group

Outputs

Outtakes

Comms results

Activities

Business results

Reception

10

© Geddes Analytics. Not for distribution without permission.

Outtakes

•Cognitive change▫Awareness

▫Understanding

▫Perceptions

▫Advocacy

•Use: CMO, marketing and communications executives

Outputs

Reception

Comms results

Activities

Business results

Outtakes

11

© Geddes Analytics. Not for distribution without permission.

Communicationsresults

•Specific, desired behaviors

•Precursor to business value

•Link to outcomes, etc.

•Use: CMO, marketing and communications executives

Outputs

Reception

Outtakes

Activities

Business results

Comms results

12

© Geddes Analytics. Not for distribution without permission.

Business results

•Tactical, consumer results

•Strategic business results

•ROI

•Statistical methods

•Use: CEO, CMO, business unit leaders

Outputs

Reception

Outtakes

Activities

Business results

Comms results

13

© Geddes Analytics. Not for distribution without permission.

A standard framework3 Influence and engagement

14

© Geddes Analytics. Not for distribution without permission.

Engagement 1

•Where does it fit?▫Reception

▫Handling

•What is it?▫Precursor to

cognitive change

▫Upstream

Outputs

Reception

Outtakes

Comms results

Activities

Business results

15

© Geddes Analytics. Not for distribution without permission.

Engagement 2

•Where does it fit?▫Relevance▫Relationships▫Advocacy

•What is it?▫Result of cognition

▫Downstream▫Precursor to results

Outputs

Reception

Outtakes

Comms results

Activities

Business results

16

© Geddes Analytics. Not for distribution without permission.

Influence 1

•Where does it fit?▫Reception

▫Information availability

•What is it?▫Precursor to

cognitive change

▫Upstream

Outputs

Reception

Outtakes

Comms results

Activities

Business results

17

© Geddes Analytics. Not for distribution without permission.

Influence 2

•Where does it fit?▫Outtake

•What is it?▫Result of cognition▫A new way of

thinking▫Downstream▫Precursor to results

Outputs

Reception

Outtakes

Comms results

Activities

Business results

© Geddes Analytics™ 2012. Not for distribution without permission.

Predictive analytics

19

© Geddes Analytics. Not for distribution without permission.

The situation today

•Companies monitor traditional and social media…

▫But don’t know what media messages actually change opinions and drives behaviors

20

© Geddes Analytics. Not for distribution without permission.

The situation today

•Companies probe the marketplace via surveys and media listening platforms, but …

▫Cannot look ahead

▫Risk getting blindsided

21

© Geddes Analytics. Not for distribution without permission.

The situation today

•Monthly measurement reports give a look backwards, but …

▫Cannot evaluate possible scenarios

22

© Geddes Analytics. Not for distribution without permission.

Today’s challenge

▫Analyze past data

▫Understand what really drives change

▫Link social / traditional media to business outcomes

▫Build systems to give a look ahead

▫Facilitates responsive management

23

© Geddes Analytics. Not for distribution without permission.

Case studies

Toyota recall crisis

Consumer sentiment about the economy

Teenage smoking behavior

24

© Geddes Analytics. Not for distribution without permission.

1. Toyota recall crisis•Key questions

1. What shapes opinion about the Toyota brand?

2. Can we predict brand reputation from media?

•Data▫Predictor variables

News, online, blogs, forums … no broadcast

▫Dependent variable Brand reputation

25

All mediaR2 = 0.84

Survey− Model

Positive documents

Negative documents

Positive opinion

Neutral opinion

Negative opinion

26

© Geddes Analytics. Not for distribution without permission.

2. Consumer economic sentiment•Key questions

1. What shapes consumer opinions about the economy: media or marketplace experience?

2. Can we predict consumer sentiment from media alone?

•Data▫Predictor variables:

AP wire and Washington Post

▫Dependent variable: University of Michigan Consumer Sentiment Index

▫1977 to 2007

27

© Geddes Analytics. Not for distribution without permission.

Key results▫ We can predict based

on media alone

▫ Models based work over long time period

▫ Media play primary role in shaping opinion about their economic well-being

▫ Shelf life of information near zero

Training data

Test data

Training and test data

28

© Geddes Analytics. Not for distribution without permission.

3. Teenage smoking

Key questions

• Can we predict teenage smoking behavior from media alone?

• Does public policy affect smoking behaviors?

• Do anti-smoking campaigns work?

Data• Predictor variables:

▫ Newspapers, Google newsgroups

▫ 10 messages

▫ Cost per pack

• Dependent variable:

▫ Monitoring the Future nationwide surveys(http://monitoringthefuture.org/)

• 1991 to 2002

29

© Geddes Analytics. Not for distribution without permission.

Key results▫We can predict behaviors based on media alone

▫Cost improves prediction

▫Attitude that cigarettes are hard to get improves prediction

▫Anti-smoking campaigns and policy changes work

− Predicted smoking☐ Reported smoking

30

© Geddes Analytics. Not for distribution without permission.

The InfoTrend™ Model

31

© Geddes Analytics. Not for distribution without permission.

Powerful yet simple: basic structure

NeutralPositive Negative

K3

K4

K2

K1

Negative message force

Positive message forceStep I

32

© Geddes Analytics. Not for distribution without permission.

Minimal ideodynamic model

Do not try this at home

33

© Geddes Analytics. Not for distribution without permission.

Data requirements

Predictor variables

•Multi-channel

•Sentiment scores

•Messages tagged

Outcome variables

•Survey data

•Behavioral data

•Business data

•Must show change

34

© Geddes Analytics. Not for distribution without permission.

Concluding remarks

35

© Geddes Analytics. Not for distribution without permission.

What have we learned?

•Media Outcomes▫Sentiment

▫Messages

•But … random variability▫What does this mean for communications>

36

© Geddes Analytics. Not for distribution without permission.

Prediction and forecasting

© New York Times

•Early enough to take action

• Probable vs. improbable

•Scenario planning and simulation

•Strategic responsiveness• Michael Raynor. 2007. The Strategy Paradox: Why Committing to Success

Leads to Failure. New York: Doubleday.

37

Prediction and forecasting © Geddes Analytics. Not for distribution without permission.

© Geddes Analytics™ 2012. Not for distribution without permission.