At FORECASTxSCIENCE we compared conventional forecasting tools (such as surveys) to Big Data approaches (such as Google Trends and social media listening) in their ability to help marketers predict outcomes accurately, to spot trends, and to monitor customer sentiment. We also discussed best practices to more reliably predict what’s next.
Topics covered:
- Whether surveys are still a credible tool for guiding business decisions.
- How unstructured data (social, search, WOM) combine with traditional data for better answers.
- New tools that help us analyze more data better – AI, cognition and machine learning examples.
Seth Stephens-Davidowitz, Ph.D. – Author of Everybody Lies: Big Data, New Data and What the Internet Can Tell Us About Who We Really Are
- Frequently, the value of a dataset is not its size; it is its newness. Winners in world of Big Data are entrepreneurial and fail often to find the big winner.
- People tell things to Google that they wouldn’t share with anyone else. Surveys will not be replaced, but will be massively supplemented by Google Trends and Google Searches.
- All Big Data is not the same. Think of the incentives of your data sets (i.e. anonymous Google versus public Facebook).
- An important way to use any Big Data is zooming in.
- The future is going to be combining data sources.
Rebel Data—What Outliers, Outcasts and Aberrations Have to Say About the Future
Margaret Coles – Associate Partner, Head of Data, Research, and Analytics, Goodby, Silverstein & Partners
- Advertisers are overly focused on dinosaurs that can rarely provide anything new or insightful. Tracking the “most” can miss the “different.” Different matters when finding differentiation in a big brand.
- Shared core equity: identify sameness across brands to establish core.
- Small differences: identify the smallest stable units of difference across competitors, read differences to look for patterns and leads for further investigation. Find distinct emotional equity for each brand.
Estimating Total Sales Impact: The Magic Formula
Robin Opie – Group VP, Data Science, Oracle Data Cloud
- Oracle Data Cloud focuses on using data and science to help marketers better target and measure their digital campaigns.
- Causal measurement products empower advertisers to make better investment decisions through mathematically rigorous, scientifically valid measurement that estimates true incremental value.
- Challenges to doing digital measurement include identity, platforms, data, statistics, and momentum. Foundational accuracy is essential.
- Combining both panel-only based approach with big data approach provides adequate statistical power and an accurate estimate of total campaign value.
Making Surveys More Relevant
Erik Nylen, Ph.D. – Data Scientist, Parsec Media
- Time-based advertising: ads sold based on amount of time that people choose to spend with them, cost-per-second model.
- Multiple devices and distractions create difficulty in measurement and result in inconsistent metrics.
- Surveys work when the independent variable is meaningful; exposures used to be meaningful, so surveys worked; impressions do not mean anything, so surveys do not work.
- Two requirements for success: a) capture user’s full attention; b) user must control the amount of time the ad appears. Polite interruptions allow for measurable attention.
Marketing in a Cognitive Era
Babs Rangaiah – Executive Partner, Global Marketing, iX, IBM
- In order to succeed, create experiences (ads, content, etc.) that engage consumers that must be made for mobile, must be visually social, and must be personalized and leveraging data, and must have built appropriate commerce engines for those particular brands or products.
- Blockchain: a trusted, distributed ledger. Blockchain for media benefits include transparency and trust; data, verification, and attribution; Watson AI; smart contracts.
Amplifying Survey Results with Google Trends Data
Jeffrey Oldham, Ph.D. – Software Engineer, Economics Team, Google
Google Trends is an aggregation of Google web searches over time. This presentation discussed a method to connect Google Trends to survey results in projecting national survey responses to smaller geographic regions.
Beyond Demographics: Targeting Likely Consumers through Psychographic Traits
Steven Millman – Chief Scientist, Simmons Research
- Analysis of the psychographic traits of consumers offers a powerful new way to understand predictors of consumer choice and to activate more sophisticated targets with higher ROI.
- Predictive Consumer Insights (PCI) targets likely brand users more efficiently than demos alone. The predictive power of PCI is more than 250% greater than targeting on demos alone.
The ESPN Fan Experience: Hearing Our Fans On Their Terms
Douglas Kramon – Director Customer Operations/Customer Care, ESPN, Inc.
- Fan experience: the sum of all interactions a fan has with ESPN, across all platforms.
- Communication with fans is raw and authentic; real-time; single source of the truth; unsolicited; granular (what would you do?).
- Data needs to be curated quickly.
- Opportunity exists with contact centers to make tools and technology that go beyond the standard CRM.
Finding Customer Signals – Powering Advanced Analytics with Person-Based Data
John Gim – SVP, Advanced Analytic Solutions, RAPP
David Popkin – Head of Data Strategy for Brands, LiveRamp
- With high quality identity resolution, analysts should be looking for signal-based data to put together more precise insights. (Popkin, LiveRamp)
- Cookies are not people! Nor are they a stable identity currency. (Popkin, LiveRamp)
- Identity resolution: taking customer touchpoints and bringing them all to a consistent, stable ID. (Popkin, LiveRamp)
- Data providers seek availability, transparency, and competitive advantage. (Gim, RAPP)
Behavioral Data in a Survey World (and vice versa)
Nicolas Brézet – SVP, Head of Behavioral Data Group, Ipsos Connect
Marketers are seeking a holistic understanding of consumers’ multi-platform digital behavior, touchpoints, activities, and content consumption. Single-source data can’t tell the whole picture. This presentation discusses how Ipsos combines survey data with behavioral data, finds a framework, and then makes recommendations
Do More With Your Data
Natasha Stevens – EVP, Digital Experiences & Data Integration, Consumer Experiences, GfK
- Speed: acceleration of technology and availability of data is placing extreme pressure on clients and professionals. Clients want answers quickly and cost effectively but don’t want to sacrifice data quality. How do we capture and use data that we already have to answer questions in a methodologically rigorous way?
- Attention: Smartphones have changed our relationship with data, how we interact with information, how we share information, and even how we process things in our brain.
- Behaviors: Decisions are made in complex, nuanced, non-linear ways, especially as it relates to purchases of products and services. To understand those behaviors, different techniques must be used. You cannot rely on self-reported data.
- Researchers have the opportunity and responsibility to lead and think about how to use all the available data and bring it together in a thoughtful way.
- Optimizing data in a disruptive environment is the challenge. Use the data assets you already have; it doesn’t have to be single-source data. Ask, what do we have and how can we work with it? Digital behavioral data is needed to answer questions about what people did online and is quickly obtained.
- Issues to address with data include efficiency, proper use, and quality. Solutions include searchable data (catalog), governance, and standardized analytics.
Building a Visual Brand in the Connected Kitchen of the Future
Aviva Downing – Advertising Manager, ChefSteps
Abigail Hollister – VP, Director of Client Services, Ameritest
- Sous Vide: a slow-cooking technique that results in perfectly cooked food every time. ChefSteps: a brand about cooking and connections built through social media apps. Marketing challenges were to grow awareness of Sous Vide and Joule, and an internal challenge was to grow from a content company to advertising a product. (Downing, ChefSteps)
- Memories are formed by engaging viewers emotionally, telling them the role of the brand in their life—on a rational level, telling them about how the brand/product works. (Hollister, Ameritest)
- Optimize for understanding and time. (Hollister, Ameritest)
- Outcomes: one month out, the branded search was changed. The actual brand search volume was up greater than expected. (Downing, ChefSteps)
- Use the feedback loop to create advertising specific to favorite platforms. (Hollister, Ameritest)
Can Panel Data and Attribution Get Us Closer to TV Advertising Omniscience?
Dirk Beyer, Ph.D. – VP, Data Science Research, Neustar
- Models on nonexperimental data are needed to measure effectiveness of TV advertising.
- More granular inputs for linear and addressable TV allow for more detailed analysis of TV response by audience, creative, campaign, and other attributes.
- Voracity of the analysis depends on size of media panels, frequency of exposure and conversion rates.
- Ability to connect to other media channels relies on robust Identity Graph.
Internet of Packaging
Thomas Körmendi – CEO, Kezzler
Kunal Nayar, Ph.D. – Project Team Leader, Kezzler
Moderator: Scott McDonald, Ph.D. – President & CEO, ARF
- Internet of packaging: brand protection; track and trace; and, consumer engagement.
- Uses the same ID in production, distribution, in hands of consumers, which enhances supply chain visibility and insight.
Building a Consumer Marketplace for DNA-Insights
Robin Thurston – CEO, Helix
Moderator: Paul Donato – Chief Research Officer, ARF
- Helix is a personal genomics company that has a mission to democratize DNA and create innovation on top of it.
- Privacy is a huge issue. Customer owns data, no data is released or sold by Helix.
- Health and fitness and Ancestry are the largest categories.
Slides Not Available