Jon Bates Group Product Manger Data Science and Machine Learning, gave a session on how to gear up for higher analytics maturity. Most people have nailed descriptive analytics but are now being pushed to take it to the next level. Jon wants to provide a vision of what that maturity is.
How to do it:
- Get your core services up so that insights can be actioned
- Free up your time by delivering self-service analysis to the business
- Get strategic
Predictive analytics is like planning for a Marathon. It’s a series of steps. You start simple with a 5K and then move on to a 10K, half marathon, and finally a full marathon. Here are the steps:
- Descriptive analytics (tactical) – what happened
- Diagnostic analytics – why
- Advanced diagnostic analytics – why and for whom
- Predictive analytics (strategic) – what’s next or what could happen
- Cognitive analytics- newer with machines making decisions in context
How do you get there? The Analytics Plan
It’s about freeing up your time and where to get started. It’s an analytics maturity plan.
Descriptive Analytics
You need to collect, connect, and democratize.
Challenges:
- It takes a long time to manage code
- You get an endless stream of question from the business
- You have limited time to execute
Objective
- Reduce time to configure data collection
- Reduce dependency on IT resource
Outcomes
- Foundation for taking action on insights
- More time to focus on analysis
First mistake: Not having objectives that align to your vision
Gear for success
- Tools
- Marketing cloud org id
- Dynamic Tag Management (DTM)
- Regional Data Collection (RDM)
- Visitor ID Service
- This will give you the access you need while reducing your need for IT.
- People
- people
- Device graph
- To get to people, you need to publish and share audience segment in the new co-op
- This will allow you to connect he consumer journey across devices
- Analysis workspace: democratize the process
- This will give you rapid data analysis and visualizations.
- It will also free up your time for deeper analysis
- Note: Analysis Workspace is the home for all future Adobe Analytics features
- Now you’ve got self-service up and running. This include new features like Segment IQ
Diagnostic Analytics
Take care of significance, unify, compare
Challenges
- Massive amounts of valuable analysis time is spent trying to answer “why”
Objective
- free up your time for higher value analysis and sneure insights can be executed
Outcomes
- More tie for higher value analysis and clear paths for optimization
Gear
- Anomaly detection
- Why: It’s a starting point for analysis and audience discovery. It reduces the risk and improves efficiency and uncovers opportunities.
- Anomaly detection also automates advanced statistical analysis for the non-statistician
- Marketing cloud audiences
- Gives you enriched segmentation with online and offline data (third party
- real time integration
- allows you to analyze audience reach, demographic data, and psychographic data
- Segment IQ
- Segment IQ helps you discover differences between segments with automatec analysis
- You can uncover the key characteristics of the audience segments that drive your KPI’s
- You can also identify overlap between your segments
Advanced Diagnostic Analytics
Move onto enrich, explain, and uncover
Challenges
- Audiene and customer profiles are partially understood and incomplete
Objectives
- Enrich understanding of the customer and reduce analysis time
Outcomes
- Deepen your understandg of the customer
Gear
- Customer attribute
- Understand the entire customer journey across channel
- Deliver more relevant and timely content to prospects
- Identify and target key micro-segments
- Need Analytics Premium or Premium Customer 360
- Contribution analysis
- Intelligently identifies possible cause for anomalous changes in data
- Reduces analysis from days or weeks to seconds or minutes
- Provides data science as a service for the business user
- The new contribution analysis tool in beta drove or saved $50M across 12 customers
- One client identified where cart removal increased by 300%. Needed to know why. The analysis identified two specific products and the tag management solution. Tag Manager was kicking it out of the cart.
- Anotehr client used it to identify an important change in a campaign. Saw a decrease in revenue for the brand. One campaign was killed because they used the wrong metrics for measurement. They reactivated and closed the gap
- Audience clustering
- Uncover relevant, valuable, sizeable audiences of customers for targeting
- Move beyond simple segmentation towards actionable audiences
- Forget educated guesses… uncovers meaningful patterns
- Need Analytics Premium. It’s part of Workbench
You are now halfway to Predictive Analytics
Predictive Analytics
It falls into two areas. First is making predictions around your business. (forecast revenue.) Second is making predictions about your customers (churn, buying patterns, etc.)
Challenges
- measuring true impact of marketing is back-breaking
- What is likely to happen next?
Objectives
- Connect the entire cstomer
- Predict future customer actions
Outcomes
- Assist with better budget planning
- Targeted customer experience
Gear
- Algorithmic Attribution
- Ensures impartiality through a data-driven approach. Not just you deciding what whether first touch is the biggest indicator.(or some other model)
- Accurately quantifies the impact of marketing in driving success
- Combines interactions across paid media and brand-owned experiences
- Propensity Scoring
- Predict how likely customers are to complete a future action
- Market to the most relevant audiences
- Predict ho much lift to expect before you target an audience
- need Analytics Premium Complete or Premium Predictive Intelligence
- Decision trees (classify)
- Rapidly analyst the probability of all possible outcomes
- Easy to interpret and explain
- Deploy the models for real-time decisions and personalization
- need Analytics Premium Complete or Premium Predictive Analytics
Elevating the Role of Analytics
Don’t forget one key aspect here. You have to organize for success. You can take a variety of approaches like centralized, COE, consulting, functioal, and decentralized. Turns out consulting is the least effective. Centralized or COE is the most able to allow you to mature.
Key Consideration: nurture a mutualistic relationship between analysts and data scientists. (get data scientists if you don’t have them.
Next steps
- Get strategic. Evangelize the insights and become the secret weapon to the C-Suite
- Key strategies
- be a trust builder
- go the extra mile to foster learning
- Be the guide to which the enterprise turns
- Speak your audience’s language
- Tell the enterprise what’s possible
