Session 5 additional analytics operations
1 like380 views
This document summarizes a session on additional analytics and metrics for international new business development. It discusses choosing the right metrics at different stages of business, including acquisition, activation, retention, and revenue. It provides industry-specific examples and lessons on developing dashboards to report metrics. Finally, it covers when analytics can and cannot be used, how to compete on analytics as a company-wide effort, potential errors, and using a balanced scorecard approach.
![Feel the pain! Suggestions for ranking results
from convervent problem investigations: Rule
of 31
1. Did the interviewee successfully rank the problems you presented?
[10 / 5 / 0]
2. Is the interviewee actively trying to solve the problems, or has he
done in the past? [10 / 5 / 0]
3. Was the interviewee engaged and foucssed throughout the
interview? [8 / 4 / 0]
4. Did the interviewee agree on a follow-up meeting (solution meeting? [
8 / 4 / 0]
5. Did the interviewee offer to refer others to you for your interview? [4 /
2 / 1]
6. Did the interviewee offer to pay you immediately for your solution? [3
/ 1 / 0]
8-7-2014 5
Croll&Yoskovitz, p. 53](https://image.slidesharecdn.com/qgwzw4s2sxwymehpnvop-signature-8126d699322c1c0821543643346e600bd6c75281309d374bdfbb3dafb9042c83-poli-140708033834-phpapp02/85/Session-5-additional-analytics-operations-5-320.jpg)
![8-7-2014 11
Background
What are you trying to achieve?
Next action
What’s the next action
Falsifiable Hypotheses
Declare your expected outcome.
Use this format:
[Specific Repeatable Action] will [Expected
Measurable Outcome]
Details
How will you set up this experiment?
Validated learning
Summarize your learning from the experiment
Results
Enter your qualitative/quantitative data
EXPERIMENT REPORTTitle: [TITLE] Author: [NAME] Created: [DATE
[VALIDATED or INVALIDATED
Lean Stack by Spark59.com](https://image.slidesharecdn.com/qgwzw4s2sxwymehpnvop-signature-8126d699322c1c0821543643346e600bd6c75281309d374bdfbb3dafb9042c83-poli-140708033834-phpapp02/85/Session-5-additional-analytics-operations-11-320.jpg)
Session 5 additional analytics operations
- 1. 1 International New Business Development Session 5 Additional analytics Dr. Rainer Harms 8-7-2014
- 2. Previously on INDB You investigated customer needs (segment & problem) You developed a solution (solution and unique value proposition) You are going to investigate whether the solution could meet the needs and to discover the „power of analytics“ in other situations. 8-7-2014 2
- 3. What metrics to choose? 8-7-2014 3 The One Metric That Matters Croll&Yoskovitz, p. 53
- 5. Feel the pain! Suggestions for ranking results from convervent problem investigations: Rule of 31 1. Did the interviewee successfully rank the problems you presented? [10 / 5 / 0] 2. Is the interviewee actively trying to solve the problems, or has he done in the past? [10 / 5 / 0] 3. Was the interviewee engaged and foucssed throughout the interview? [8 / 4 / 0] 4. Did the interviewee agree on a follow-up meeting (solution meeting? [ 8 / 4 / 0] 5. Did the interviewee offer to refer others to you for your interview? [4 / 2 / 1] 6. Did the interviewee offer to pay you immediately for your solution? [3 / 1 / 0] 8-7-2014 5 Croll&Yoskovitz, p. 53
- 6. Sticky as sticky can get: Metrics at the stickyness stage Stickyness: from Acquistion to activation / retention Key question 1: „What is an active user?“ Key question 2: „How do I measure „activity“ Key question 3: How do I enhance activation? 8-7-2014 6 Croll & Yoskovitz
- 7. Industry-specific metrics. Consider this case „Consider an online luxury goods store. Subscribers to the site get exclusive deals at reduced prices for items that are curated by the site’s operators. Visitors to the site can browse what’s available, but must sign up to place an order or put something in a shopping cart; by signing up, they agree to receive a daily email update. Visitors can also tweet or like something they see on the site.” What key metrics does the company care about? 8-7-2014 7 Croll&Yoskovitz, p. 75-76
- 8. Lessons learned from industry-specific metrics Differ by stages Differ by business (model) In the end, have a simple logic: Profit = (p*x) – (c+x) - F 8-7-2014 8
- 9. Reporting metrics: use a dashboard 8-7-2014 9 1852 people 35% 31% 23% 2% Acquisition Activation 7 day retention % Referred Revenue
- 11. 8-7-2014 11 Background What are you trying to achieve? Next action What’s the next action Falsifiable Hypotheses Declare your expected outcome. Use this format: [Specific Repeatable Action] will [Expected Measurable Outcome] Details How will you set up this experiment? Validated learning Summarize your learning from the experiment Results Enter your qualitative/quantitative data EXPERIMENT REPORTTitle: [TITLE] Author: [NAME] Created: [DATE [VALIDATED or INVALIDATED Lean Stack by Spark59.com
- 12. From customer discovery to going concern: Competing on analytics „At a time when firms in many industries offer similar products and use comparable technologies, business processes are among the last remaining points of differentiation. And analytics competitors wring every last drop of value from theses processes“ (Davenport 2006, p.1) 8-7-2014 12
- 13. Common applications and example companies 8-7-2014 13 Davenport (2006) p.105
- 14. Key questions of Business Analytics 8-7-2014 14 Information What happened? (Reporting) What is happening now? (Alerts) What will Happen? (Extrapolation) How and why did it happen? (Modeling, experimental design) What’s the next best action? (Recommendation) What’s the best/worst that can happen? (Prediction, optimization, simulation) Insight Past Present Future
- 15. 8-7-2014 15 Demand forecasting, capacity planning, alternate supply Demand forecasting, capacity planning, alternate supply Bottlenecks, equipment failures, yield variations Product quality, order performance, asset utilization Dynamic routing, order combination, preventive maintenance Process control, quality control, bottleneck analysis
- 16. When you COULD NOT do analytics 8-7-2014 16 There’s no time No precedent History is misleading When our assumptions no longer apply Variables can’t be measured Decisionmaker has considerable experience BUT … mind the sitation Human beings can make accurate and quick decisions on personality and intentions
- 17. When you COULD do analytics: Nature of business processes 8-7-2014 17 Data-rich Information-intensive Asset Intensive Labour Intensive Cross-functional or cross-business in scope Low average success rate Dependant on speed and timing consistency and control distributed decision making
- 18. If you want to compete on analytics, do 8-7-2014 18 Recognize limiations Time Entrepreneneurship? Company-wide effort Company wide: scale econ. No data islands CEO level sponsorship Move to analytics as cultural change Employees with high level of proficiency in: Analytical skills (IT, quantitative methods) Business skills (professional fields) Communication (relationship) skills “quantitative literacy” http://treasure.diylol.com/uploads/post/image/444616/resized_jesus-says- meme-generator-in-god-we-trust-all-others-bring-data-e073f5.jpg
- 19. Be aware of: I 8-7-2014 19 STRATEGIC errors: Not the right questions Use analytics to justify what you want to do, instead of letting the facts guide you „Statistics are often used as a drunken man uses a lamppost – for support rather than illumination” (Lang in Davenport, Harris & Morrison, 2010, p.12) Fail to understand all alternatives Fail to interpret the data correctly LOGIC errors:
- 20. Be aware of: II 8-7-2014 20 LOGIC errors Incorrect assumptions (mortgage lending models relied on the assumption that house prices would continue to rise) Careless mistakes (errors in the datasets) Insufficient decision-making criteria Too late to be of use Postponing decisions
- 23. A Business Analytics mind map 8-7-2014 23
Editor's Notes
- #4: C&Y point out that the selecion of a key metric is a function of both the key goal of a venture at a certain time (which they call „life cycle“, but we dont) and „Business model“. They have this nice flipbook analogy of a business model. For any busienss model element, there are a discrete choice of options. „element-option“ combinations can be comined at will to create a number of iterations. C&Y have: product type, delivery model, revenue model, selling tacti, acquisitoin channel (and theis subtypes) on p. 68 That means taht finding the PARTICULAR metric seems to be somewhat of a art rather than just logic.
- #5: Upper row: Popular terms form startup management authors. We have worked with Maurya‘s lean canvas. Very similar to th is is Eric Ries’s “lean start up”. Other authors are Dave McClure and Sean Ellis (who coined the term “Growth hacker”). You see that the beginnig and end points are very similar. They start with undersatnding thecustomer problem in terms of quality (what is it) and intensity (how severe is it). Then, elements of the business model are carefully scrituinzed by hypothesis-driven approaches. If things worked well, you can scale the business to a larger market. For each of these steps, Croll and Yoskovitz suggest a set of metrics and measurement approaches. We are going to (re-)discuss “Emphathy” assessment in the beginning of a startup process and “Stickiness” where it is about getting the first, leading customers in.
- #6: I suggse that the „customer problem“ type metrics are pretty similar for all business models. After that it gets more specific. C&Y: suggest to take notes about these elements of an interview. „31“ is font scientifc, but „from experience“ Check the book pages for examples on what 10 points / 5 points / 0 points actually mean.
- #7: Quidid –a tool for simple survey. Frist approach: those that are invited need to create an account. Well, most dont. Second appraoch: You can answer any question without an account, and reqister later if you want to use the tool for your porposes. KQ 1: If Business Model is a „sales“ model, where people come in one time and buy stuff & then leave, the concept does not make much sense (or does it?). Many other BM are however geared towards having customers in for a little longer. Then you can indeed get information on „activation“ e.g. in terms of time spent on a website, time until return, etc.
- #8: DONT FORGET THE OFFLINE COMPONENTS!!! Like shipping time etc. Conversion rate: The number of visitors who buy something Purchases per year: The number of purchases made by each customer per year. Average shopping cart size: The amount of money spent on a purchase. Abandonment: The percentage of people who begin to make a purchase, and then don’t. Cost of customer acquisition: The money spent to get someone to buy something. Revenue per customer: The lifetime value of each customer. Top keywords driving traffic to the site: Those terms that people are looking for, and associate with you—a clue to adjacent products or markets. Top search terms. Both those that lead to revenue, and those that don’t have any results. Effectiveness of recommendation engines: How likely a visitor is to add a recommended product to the shopping cart. Virality: Word of mouth, and sharing per visitor. Mailing list effectiveness: Click-through rates and ability to make buyers return and buy.
- #11: From „Lean Startup“ online course itself
- #16: Examples: Supply Chain
- #19: Company wide: from targeted approaches to company-wide analytics. Example: UPS who have traditionally modeled the logistics, now have move to apply analytics to predicting the behaviour of their own customers. No data islands: In many organiziatins, data are collected in an non-systematic way: single persons and single work groups collect their own spreadheets. They contain errors, are badly documented etc. Scale economies: when you go compnay wide, you can afford methods experts! CEO sponsorship: Often a cultural change issue Limiation: allow persons to override decisions in unusal circumstances (Hotel booking & hurrican Kathrian) or when the algorithm sucks (pictures of naked women increase web traffic)