Key Takeaways from this presentation include:
- How data is used to run day to day operations
- How data is used to influence product decisions and marketing strategies
- Which skills are necessary to become self-serving in data tasks regardless of core responsibilities
11. Background
● Born and raised in Los Angeles
● Graduated Princeton University in 2013
○ Majored in Operations Research and Financial Engineering
○ Competed for the college swim team
● Worked on Wall Street as an Analyst at Citigroup
● Joined Zoomer Delivery
○ Was a City Launcher who launched the Chicago and Boston markets
○ Transitioned to being a Data Analyst
● Scale operations with Abra
○ Develop and scale company’s analytics capabilities
○ Optimize and automate operational processes
13. ● Because the world of data is vast, where and how do I start?
○ What tools are available to me and which ones are the most useful?
○ Who are data professionals, and what do they do?
○ How can I develop a data driven culture in my organization?
How do I get started?
14. ● Because the world of data is vast, where and how do I start?
○ What tools are available to me and which ones are the most useful?
○ Who are data professionals, and what do they do?
○ How can I develop a data driven culture in my organization?
How do I get started?
Start small and continuously build on your
experiences!
16. How can I develop data skills?
● There are many tools out there for many different purposes:
○ Basic data modeling and ad hoc analysis
■ Microsoft Excel and/or Google Sheets
17. How can I develop data skills?
● There are many tools out there for many different purposes:
○ Basic data modeling and ad hoc analysis
■ Microsoft Excel and/or Google Sheets
○ Large data aggregation and queries
■ SQL
18. How can I develop data skills?
● There are many tools out there for many different purposes:
○ Basic data modeling and ad hoc analysis
■ Microsoft Excel and/or Google Sheets
○ Large data aggregation and queries
■ SQL
○ Advanced data analytics
■ Python, R or Matlab
19. How can I develop data skills?
● There are many tools out there for many different purposes:
○ Basic data modeling and ad hoc analysis
■ Microsoft Excel and/or Google Sheets
○ Large data aggregation and queries
■ SQL
○ Advanced data analytics
■ Python, R or Matlab
○ Visualization
■ Periscope, Tableau or Shiny
21. Data professionals in an organization
● Data engineers and data architects
○ Develop, build, deploy and maintain data assets
22. Data professionals in an organization
● Data engineers and data architects
○ Develop, build, deploy and maintain data assets
● Data analysts
○ Understand the data assets of the business
○ Bridge between the data assets to various business divisions
○ Perform analysis that are used to drive various business decisions
23. Data professionals in an organization
● Data engineers and data architects
○ Develop, build, deploy and maintain data assets
● Data analysts
○ Understand the data assets of the business
○ Bridge between the data assets to various business divisions
○ Perform analysis that are used to drive various business decisions
● Data scientists
○ Are extremely technical
○ Mine for deep insight in large datasets
○ Constantly looks for obscure patterns that will add value to the business
27. Steps to telling a good story:
1. Define the question
2. Determine key factors affecting the system
Source: Turning Numbers into Knowledge by Jonathan G. Koomey
28. Example (cont.):
N. America Lat. America Europe Asia
App Downloads
Signup Rate
Activation Rate
Retention (LTV)
29. Steps to telling a good story:
1. Define the question
2. Determine key factors affecting the system
3. Evaluate factors by importance and uncertainty
Source: Turning Numbers into Knowledge by Jonathan G. Koomey
30. Example (cont.):
N. America Lat. America Europe Asia
App Downloads
Signup Rate
Activation Rate
Retention (LTV)
32. Steps to telling a good story:
1. Define the question
2. Determine key factors affecting the system
3. Evaluate factors by importance and uncertainty
4. Choosing the scenarios and adding details to the scenario set
Source: Turning Numbers into Knowledge by Jonathan G. Koomey
34. Example (cont.):
Abra Baseline
Expand into
new region
Scenario 2 Scenario 3
App
Downloads
A A + ∆
Signup Rate B% B%
Activation
Rate
C% C%
Retention
(LTV)
$D $D
35. Example (cont.):
Abra Baseline
Expand into
new region
Optimize user
experience
Scenario 3
App
Downloads
A A + ∆ A
Signup Rate B% B% B% + ∆%
Activation
Rate
C% C% C% + ∆%
Retention
(LTV)
$D $D $D
36. Example (cont.):
Abra Baseline
Expand into
new region
Optimize user
experience
Add to new
investments
App
Downloads
A A + ∆ A A + ∆
Signup Rate B% B% B% + ∆% B%
Activation
Rate
C% C% C% + ∆% C%
Retention
(LTV)
$D $D $D $D + $∆
37. Steps to telling a good story:
1. Define the question
2. Determine key factors affecting the system
3. Evaluate factors by importance and uncertainty
4. Choosing the scenarios and adding details to the scenario set
5. Determine the implications
Source: Turning Numbers into Knowledge by Jonathan G. Koomey
38. Example (cont.):
Abra Baseline
Expand into
new region
Optimize user
experience
Add to new
investments
App
Downloads
A A + ∆ A A + ∆
Signup Rate B% B% B% + ∆% B% + ∆%
Activation
Rate
C% C% C% + ∆% C% + ∆%
Retention
(LTV)
$D $D $D $D + $∆
39. Example (cont.):
Abra Baseline
Expand into
new region
Optimize user
experience
Add to new
investments
App
Downloads
A A + ∆ A A + ∆
Signup Rate B% B% B% + ∆% B% + ∆%
Activation
Rate
C% C% C% + ∆% C% + ∆%
Retention
(LTV)
$D $D $D $D + $∆
Revenue Uplift R $R + $∆1 $R + $∆2 $R + $∆3
40. Steps to telling a good story:
1. Define the question
2. Determine key factors affecting the system
3. Evaluate factors by importance and uncertainty
4. Choosing the scenarios and adding details to the scenario set
5. Determine the implications
6. Define metrics for success
Source: Turning Numbers into Knowledge by Jonathan G. Koomey
42. ● Data literacy is becoming a requirement in business
○ Everyone should be a data professional
● Keep up with various data tools out there to give yourself
competitive advantages
● Seek out knowledge!
44. www.productschool.com
Part-time Product Management, Coding, Data, Digital Marketing and
Blockchain courses in San Francisco, Silicon Valley, New York, Santa
Monica, Los Angeles, Austin, Boston, Boulder, Chicago, Denver,
Orange County, Seattle, Bellevue, Toronto, London and Online
Editor's Notes
Graduated from Princeton in 2013 with an Operations Research and Financial Engineering Degree
People would ask, what does this major entail, and to simply put it, its a major that combines statistics, computer programing and economics into one
Following Princeton, I decided to venture out into the world of finance.
I joined a banking group at Citigroup in NYC.
Although it was not data heavy at first, I had access to a vast amount of data, from Bloomberg terminals, client data, and data from past transactions in our space.
So as I started to create marketing and pitching materials for our clients, I saw how powerful numbers and analyses are in terms of winning client business
Instead of pitching a major client like Ford, issuance recommendations based on a gut feeling, creating succinct and accurate market and competitor analysis really establishes us as an industry authority which then gives us a competitive advantage when winning over clients
Following Citi, I worked in more traditional data roles with Zoomer Delivery and currently here with Abra.
These two companies were fairly early in their lives so I had the opportunities to work with teams across the board, from operations, marketing, engineering and last but not last, product.
Because the companies were so small, I had the rare opportunity to see how different teams leverage data:
marketing uses data to target and influence user behavior
operations and engineering uses data for optimizing existing infrastructure and keeping the business running,
product uses it to research customer and market opportunities to build great products.
But despite working so cross functionally, the same theme keeps coming up again and again: the one who could leverage data the best is usually the one who always had the most clout in the room, and by that, I mean he/she usually had the most compelling and persuasive argument in the room.
Data can be daunting.
Many companies are moving to be “Data driven” but what does that mean?
How can you be prepared for a changing business landscape where data is quickly becoming king?
For those who do not have a strong background in data, scripting or coding, finding and learning the various tools of the trade can be daunting.
My development in the various tools were not formal, as I discovered and learned to use much of these tools on the job or on my free time.
With that, I want to share some of this experiences with you to help you both focus some of your short term data development and hopefully give you a starting point that will lead you to explore other tools that exist in the data and analytics industry
To start, I recommend getting acquainted with Microsoft Excel or Google Sheets
These two tools are extremely useful to analyze small to medium sized data sets
Additionally, these tools are quite useful in helping you build data models before implementing them in more advanced tools
However there are limitations:
Google Sheets and Excel are not good for large data set analysis.
For example, Google Sheets has a 2 million cell limit
There are better tools for this
Skills that are good for large data analysis will revolve around SQL.
SQL is a tool that will allow you to join various data sources, perform aggregations (sum, average, min, max) and manipulate large data sets for exporting.
A few lines of SQL code can allow you to find out high level to fairly granular data quite quickly
For example, you have a table in the database that has all the sales that occur within your business. You can determine:
The region with the highest sales
Your top 50 most valuable customers
And in some instances, the times of the year when the business sees the highest sales volume
There are a few nuances and limitations to these languages that can be frustrating for those trying to learn SQL
There are a few SQL based languages out there - postgres, MySQL, Redshift - that may have syntaxical differences to keep in mind, but as a whole do the exact same thing
Advanced analytics, such as identifying treadlines and developing predictive analytics is extremely difficult to do in SQL
This is where scripting languages come in.
Of the tools listed here, Python and R is free and are both very powerful for advanced analytics. Matlab is a paid tool and used quite frequently in academia
These tools have many free importable libraries, which are mainly functions, that will enable you to both manipulate data in a more complex way and implement algorithms (e.g. Machine Learning) to find deeper insights from your data sets
Take for example, I have a large dataset of commodities prices and S&P 500 closing price. I can easily implement some code to help me determine how correlated commodity prices are in relationship to how volatile the broader market is.
However, no matter how good the analysis is, it is difficult to show your results with just numbers alone. Data visualization helps with this. You can write additional code in Python, R and Matlab, but there are more user friendly platforms that will allow you to do this.
All these tools listed here (although there are many out there) specialize in being a one stop shop for your data.
You can query, manipulate and visualize your data in one single product.
User interfaces here are very user friendly to help less technical people perform quite advanced analytics
As with all things, it is always important to get feedback for your analysis. Although you can learn and implement these tools, there are many pitfalls where your analysis could mean different things
Sanity check numbers and get implementation feedback from ore technical individuals within your organization
Although there were many different tools discussed in the previous section, chances are there are many individuals in your organization that you can leverage in helping you implement these tools in your own day to day. We will summarize them here
First of all, data engineers and data architects.
These are the professionals who develop, build and maintain the company’s data assets. In other words, they build the data infrastructure of the company.
They are extremely helpful for helping you understand how the data infrastructure works. For example:
They can help you understand the links between tables within a database or help you understand what table values or columns mean in relation to laymen business definitions
Data analysts (or sometimes business analysts) are those who translate data into something that people who are not as familiar with data can understand.
A good business analyst are able to work with data engineers effectively to understand the company data infrastructure
Subsequently, they are able to use that knowledge to determine information that will help make everyday business decisions
A business/data analyst is a good resource to look for high level reporting that can give you good insight into the the happenings of the business
Lastly, we have Data Scientists
These professionals implement scientific methods and algorithms for data analysis
Data scientists perform a lot of Machine Learning algorithms to help companies constantly monitor trends in their businesses and industries they opperate in
Examples of analysis that data scientists will perofrm:
Use user details and behavior data to assess the probability of a sale
Use user past behavior to suggest recommendations
In conclusion here, there are many professionals in your organization who specialize in various data verticals. Understanding what they do and knowing how to leverage them effectively can ultimately help you be more effective in your own role.
Now I have talked a bit about what tools we can use to analyze data and who we can talk with in our organizations about data, how can we personally be more data driven.
I believe the answer to that is to alter our own paradigms to use data to tell stories.
Whether we want to convince the head of sales to expand to a new region, or asking the operations director to switch to a new vendor that will reduce cost, data will enable you to be more persuasive by helping you develop compelling argument with hard evidence.
Being data driven means being structured in how you approach using data to tell your story.
It helps ensure that your analysis methods are robust and easy to relay to those who are not performing the analysis themselves
The framework that I will discuss is adapted from Jonathan Koomey’s book Source: Turning Numbers into Knowledge
Questions could include:
Where should the company expand into next?
Which business partner is the most strategic for our business?
And for or example, and probably is what most companies ask themselves these days: Which product feature(S) will impact the business the most?
In order to determine the scenarios or decisions that we want to analyze, we want to be able to find a set of factors that would be affected given the scenarios
This is particularly important, as it will let us compare our scenarios apples to apples
If this is not clear, dont worry, it will be as we go through the next few steps
For the example at Abra, as we are focused on our question on identifying high impact product features, the best way to assess them is how they affect our user funnel.
In our funnel, from a high level can include country or regional numbers for downloads, signup rates, first-transaction rates, and lastly retention in the form of customer Lifetime Value
Once you have identified key factors affecting the system that your decision affects, it is important to evaluate which factors are the most important ones
This is important for many reasons but the one that comes top of mind:
Sometimes you have to build models to realistically see how decisions will affect that factor
Given that, some factors are hard to account for or are highly uncertain given the amount of information you have
Given the example earlier with Abra, when we start our analysis, we do not have enough data to model how our product will perform in various regions
Given this, we decided that we will not include regional analysis but rather simplify the effects to how it changes the various factors at an overall company level
The variables in the baseline columns reflect basic levels that we think will get affected by the decisions that we are evaluation as part of this process
I have replaced the other columns with scenarios, which i will explain in the following slides
Given the original question that we defined, we must come up with a scenario set (or set of options) that we want to evaluate and make decisions on.
We use the factors that we determined in the previous step to help us out.
To apply this to Abra, we want to see what scenarios (or product features) we want to analyze based on these factors we want to affect.
As a product, we have many different ways we can improve and expand the product.
Our first scenario or product expansion could be adding new user support in a new region
Given the factors that we have defined, we determine through our models and assumptions that we could increase the overall number of downloads that we would get in a month by some delta amount noted in the table
Our second option could be optimizing the user signup experience
Our hope here is that this initiative could increase the signup completion rate and have trickle down effects that would also help us improve the activation rate of these new customers
Lastly, as a crypto investment app, another way would could provide value to our customer is by add additional assets customers can invest in
In our eyes, this addition will help increases the number of downloads by tapping into a user base who is interested in that particular added asset
Additionally, because the added group is a more targeted audience, this would also have an added positive effect on improving user retention
Once we have determined the scenario sets and how we believe each scenario would affect our key factors, we need to determine a metric that normalizes all these changes so we can compare the implications of each scenario.
The way you would do this would vary depending on the question you are trying to answer
For example, if you were a part of a taskforce responsible for lowering the carbon footprint of your company, the metric would most likely be the percent reduction in carbon emissions by the company year over year
Now, I want to bring this back to our Abra example. Given our factors, we could calculate the various factor changes would in theory have on the revenue of the company.
In observing our funnel, you can see how likely app downloads could convert into an activated user. Subsequently, you can then use the expected LTV to determine the revenue contribution that activated user will have for Abra
Given this, we can calculate from a high level how these changes could result in an uplift in the overall company revenue
By being able to come up with a standardized way of looking at projects all vying for company resources, we can then prioritize which projects to work on in the coming months
Last but not least, we want to track the results of our decisions.
This will allow us to evaluate and recalibrate our methodologies and assumptions so that when we perform our next iteration of the process, we can get better at figuring out what data models best help us make good product decisions at the end of the day.
As businesses evolve, it is important to be literate in data and data terms.
Because professionals are confronted with data everywhere, everyone should be a data professional.
Everyone should have the basic skill to be able to take some data and make sense of it
As company data needs get more complex, tools will evolve as a result. It is important to understand the tools available and which ones would be beneficial for your organization
Lastly, seek out knowledge! The world of data is quite large and daunting but there are so many resources out there to get you started!
I hope that if there is anything you can take away from my talk, I hope i have given you a seed to help you grow the data and analytics tools that you will nurture as you grow your career. Thank you!