Applying The Analytics Framework:: How To Get To Actionable Analytics Initiatives
Applying The Analytics Framework:: How To Get To Actionable Analytics Initiatives
1. Business objective
• Please briefly describe your top business priority. (At this point, please don’t think
about data/analytics and focus on these initiatives from a business perspective.)
2. Key actionable business initiative
• List some specific business initiatives that you are considering.
• Among your ideas, what is your most impactful business initiative?
• Is it sufficiently specific to be actionable? To make sure it is, how would you execute
the initiative?
3. Metrics of success
• How will evaluate your degree of success of your initiative? What are the (few) key
metrics that you and others will look at?
• Please prioritize these metrics (less than 3).
• What is your hypothesis of how and by how much your business initiative is going to
move your metrics.
4. Role of analytics
Describe the way(s) analytics might add value:
• Will it enable the business initiative?
• Will it help you ideate to refine the business initiative?
• Will it help you evaluate the success of the business initiative?
Please note that analytics might add value in more than one way.
5. Thinking through the analytics
Data
• Are you designing your own data or relying on existing data? Describe.
• What is your outcome/target and what are you explanatory variables/features?
• Describe the variation in your data (outcome/target and variables/features) that you
have to answer your question of interest.
Type of analytics
For each way you identified in which analytics ads value: Is it exploratory analytics,
predictive analytics, causal analytics or some combination? Explain.
Exploratory analytics
If your analytics is exploratory, do you have a good understanding how exactly the data was
collected and what it measures? If not, please connect with someone who does (and stress-
test your understanding of the data).
1
Think through the interpretation of what you find:
• Do you want to interpret the relationship between features and the outcome in a
causal manner? If so, you have just crossed over the causal analytics. See below for
follow-up questions.
• Are the features that strike you as interesting actionable? Remember that you might
be able to act on a feature even if you can’t manipulate it.
Predictive analytics
If your analytics is predictive, do you care more about the best prediction or about
answering "why" questions (or both)? Based on this answer, do you want to focus on
analyst or data-driven models (or do you need both)?
Causal analytics
If your analytics is causal, how are you planning to establish a causal relationship between
the factor or interest and the outcome? Specifically, are you planning to use an RCT, a
quasi-experimental approach (matching, diff-in-diff, phased roll-outs), or opportunistic data
with controls for confounds?
• If you are planning to use an RCT or quasi-experimental approach, describe your
design.
• If you are using opportunistic data, use the Causality Checklist to identify possible
confounds. Describe why they might be a confound. How do you plan to deal with
these possible confounds?
Impediments
• What impediments do you face in executing your analytics? How will you overcome
these impediments?
6. Executing the Analytics
• Who in your organization is going to be responsible for executing the analytics?
• Who will collect the data? Who will run the models? Who will implement and evaluate
the experiments?
• How will you involve them in defining metrics and thinking through the analytics?
7. Implementation
• Once you have the results, what decisions will this influence? – What would do
differently because of the analytics developed?
• Have you thought through about existing or new workflows and how would you embed
analytics to ensure adoption?
8. Scale
• What organizational challenges may limit the success of scaling up your analytics-driven
business initiative? (Data, People, Systems, Culture, etc.)
• What are you going to do address these organizational challenges?
• Do you have a plan in place to keep improving your business initiative with analytics or
was your analytics a one-shot deal?