sentiment: the art and science of measuring emotion
TRANSCRIPT
Karina Taylor, U.S. Bank@karinataylor
Sentiment:The Art and Science Of Measuring Emotion
• Heavy focus on marketing as an ‘art’
• Attempt to copy success without quantifying goals or using data
How Information Becomes Knowledge
• Heavy focus on marketing as an ‘art’
• Attempt to copy success without quantifying goals or using data
• Desire to be more data driven
• Focus on high-level metrics
• Lacks cadence
• Lacks specific goals
How Information Becomes Knowledge
• Heavy focus on marketing as an ‘art’
• Attempt to copy success without quantifying goals or using data
• Desire to be more data driven
• Focus on high-level metrics
• Lacks cadence
• Lacks specific goals
• Periodic reporting is standard, though often over longer periods of time
• Patterns of successful tactics are identified but not translated to case study
How Information Becomes Knowledge
• Heavy focus on marketing as an ‘art’
• Attempt to copy success without quantifying goals or using data
• Desire to be more data driven
• Focus on high-level metrics
• Lacks cadence
• Lacks specific goals
• Clear, regular cadence to measurement.
• Knows key metrics across channels and has specific goals for improvement
• Understands how to measure marketing ROI
• Benchmarking is ongoing and pervasive
• Periodic reporting is standard, though often over longer periods of time
• Patterns of successful tactics are identified but not translated to case study
How Information Becomes Knowledge
The Problems We Have
Problems We Have
Strategy• Segmentation• Targeting• Positioning• Product
Tactics• Social Channel• Content Format• Ad Format
Creative• Copy/Voice• Images• Video
Real problems that we need to test
Problems We Have
Strategy• Segmentation• Targeting• Positioning• Product
Tactics• Social Channel• Content Format• Ad Format
Creative• Copy/Voice• Images• Video
Statistics• Not Enough Data• Bad Experiment
Design• Video• Correlation vs Cause
OR
Real problems that we need to test Data problems
Problems We Have
Strategy• Segmentation• Targeting• Positioning• Product
Tactics• Social Channel• Content Format• Ad Format
Creative• Copy/Voice• Images• Video
Statistics• Not Enough Data• Bad Experiment
Design• Video• Correlation vs Cause
When you’re wrong, you don’t know it.
OR
Real problems that we need to test Data problems
Sentiment & Share of Voice
Challenges• Neutral mentions have
the most volume
• Variance is common
• One time events can affect sentiment significantly
• Share of Voice is too general to be actionable
Brand Negative Neutral Positive Total Volume
Bank of America 6% 88% 6% 253,713
Chase 7% 84% 9% 123,950
Citibank 5% 88% 7% 200,936
Fifth Third 6% 83% 12% 10,069
PNC 5% 86% 9% 42,476
U.S. Bank 3% 87% 10% 51,288
Wells Fargo 5% 87% 8% 327,957
Sentiment & Share of Voice
How do you use it?
Sentiment• Do you use it for planning, goal setting or strategy?• Can you understand your own data?• Is the data accurate enough for reporting?
Share of Voice• Does it help you monitor trends and adjust your strategy?• Is the data relevant?• Is there a place for share of voice in your reporting?
Sentiment: Don’t Work Without a Net
Using Net Sentiment
Calculate sentiment on a scale of -5 to +5, excluding mentions with neutral sentiment.
Improves readability and comparability over time.
Calculations
difference = positive - negative sum = positive + negative
net sentiment = (difference / sum)*5
Brand Negative Positive Total Volume Share of Voice NetSentiment
Bank of America 6% 6% 253,713 25% -0.1
Chase 7% 9% 123,950 12% 0.9
Citibank 5% 7% 200,936 20% 1.1
Fifth Third 6% 12% 10,069 1% 1.7
PNC 5% 9% 42,476 4% 1.3
U.S. Bank 3% 10% 51,288 5% 2.6
Wells Fargo 5% 8% 327,957 32% 1.1
Optimize Listening: AuditingAuditing Considerations
Auditing = Accuracy. You need to audit (verify and adjust) sentiment for all brand mentions.
Unadjusted, totals were: 10% positive and 3% negative (net sentiment +2.6)
Social data to the left has an adjusted net sentiment of -0.7
A difference of 3.1 points
It is getting SO much easier, but…
Sentiment Queue’s Abound• Emoji's• Facebook Reactions• Twitter GIFs• Bitmoji
Reporting & Sentiment
1 Awareness
2 Engagement
3 Acquisition
AwarenessAwareness: How many people do we reach?
Measurements include:• Impressions• Organic Reach• Promoted Reach• Ad Reach• Buzz• Share of Voice• Shifts in Sentiment
EngagementConsideration: How is the campaign being received and are we leveraging any influencers?
Measurements:• Sentiment• Clicks• Comments/Replies• Shares/Retweets• Completions [Video/Quiz/activity]
AcquisitionConversion: How many people completed the action we predetermined during the objective phase?
Measurements include:• Application Starts/Submits• New Accounts• View a video segment in its entirety or
for :30+?• Lead Form/Inquiry Completion• Referral to MLO or Wealth Manager
Challenges in Reporting Sentiment
Issues In Reporting Sentiment:
• Explaining sentiment• Accounting for variation• Differentiating from click and
impression data
Scorecards with Sentiment
Working Around The Challenges:
• Visual representation of complex topics
• Separate from more common digital metrics
• Educate within reporting
Thank You!