Today we discuss the 4 tiers that drive a data driven company via the "Data Pyramid". Included in this pyramid are key performance metrics, data warehousing , analytics and business intelligence and data science. Tune in to see how each of these tiers can play a role in helping you measure the success of your business.

1. F R A M E D DATA
The Road to Becoming a Data Driven Company

2. How Do I Create a Data Driven
Company?

3. A lot of companies these days like to say
that they are a data driven or metrics
driven company.

4. But what does that even mean?
Are they really data driven?

5. “Data driven” can cover a huge range of
goals and abilities.

6. So then, what makes a company
data driven?

7. First step:
Define your company’s Key
Performance Metrics (KPM)

8. These are the hard, quantitative numbers that are
attached to your business goals

9. For example, an e-commerce site might measure
revenue per day
While an online social gaming company might
prefer to measure daily active users or monthly
active users

10. Figure out what makes your company tick
and what drives its success.
After defining your quantitative goals, you
can align your entire company’s culture and
team around it.

11. A lot of companies will hire data scientists,
data analysts, and product managers
without necessarily figuring out the metrics
that define your business’ success.

12. Second step:
Data Warehousing (storing your data)

13. Data Warehousing is taking all of your
production data (from web pages, queries, user
behavior, demographics, etc.) and storing them
into one centralized source of truth.

14. Data Warehousing is taking all of your
production data (from web pages, queries, user
behavior, demographics, etc.) and storing them
into one centralized source of truth.
The schema (organization) of your data
warehouse will be centered around your KPMs.
i.e. If you’re tracking daily active users, it might
make sense to have a user login table.

15. (Yes it sounds a bit like database 101, but
you’d be surprised to see how many
companies skip that step)

16. Third step:
Now we can get into Analytics and Business Intelligence

17. Now that you have a data warehouse, you or
your analysts can use the information to
determine answers to key questions.
How many users logged on in
the last 7 days?
How many people logged on
from Idaho on Tuesday night?
How many women above 40
bought a virtual good from our
online game?

18. The next goal is to extract
numbers and generate
hypotheses, reports, and
dashboards to figure out
where we want to go next.

19. We did it!!
Many companies can stop here and
pretty much be a reasonably data
driven company.

20. However, in the last couple of years we’ve
rediscovered machine learning and have
made it part of our everyday arsenal.

21. Fourth step:
Data Science

22. Roughly, data science means taking features and
variables from the hypotheses we defined earlier
in analytics and BI, and using these to craft
machine learning models to either explain our
past or predict the future.

23. Machine learning models inform
product decisions, strategic
decisions, and form the basis for a
curiosity based company that is
now on the offensive.
How can we use our data to monetize the company
further?
How can we use our data as a key asset?
How can we use our data to figure out what our users
are doing?

24. (BTW, Framed Data provides painless data science services that any company can use to
increase their user retention. Give our free trial a shot and click the logo below!)

25. USER-FACING EXAMPLE #1: amazon.com

26. Can you spot the machine learning here?

27. Can you spot the machine learning here?

28. At the bottom is a section labeled
“Customers Who Bought This Item Also
Bought”

29. At the bottom is a section labeled
“Customers Who Bought This Item Also Bought”
Given what I’ve bought in the past, they use machine
learning to try to predict what I might be interested in
buying in the future.

30. At the bottom is a section labeled
“Customers Who Bought This Item Also Bought”
The main KPM they are trying to optimize for is revenue.

31. USER-FACING EXAMPLE #2: LinkedIn

32. Can you spot the machine learning here?

33. Actually there are 2, but we’ll cover this one first.

34. This is a 300x300 box that tries to predict
people I might know.

35. LinkedIn is optimizing for added
connections and friends, getting their users
further linked in. 

36. The articles in the news feed are also
based on my interests and previous
clicks.

37. The goal here is to optimize for
engagement.

38. But how do I determine the
best KPMs for my business?

39. Picture the business funnel
for an e-commerce website.

40. At the top you have all the
people visiting your site.
SITE VISIT

41. Out of those users, some may
look at an item.
SITE VISIT
ITEM VIEW

42. Then out of those, a few might
add the item to a shopping cart.
SITE VISIT
ITEM VIEW
SHOPPING CART

43. Finally, you have people checking
out, which translates to revenue.
SITE VISIT
ITEM VIEW
SHOPPING CART
REVENUE

44. In each stage of the funnel, you have people
leaving. This is the attrition rate of the funnel step.
SITE VISIT
ITEM VIEW
SHOPPING CART
REVENUE

45. Each stage is a potential Key Performance Metric.
SITE VISIT
ITEM VIEW
SHOPPING CART
REVENUE

46. Given your business, whether it’s online or brick
and mortar, there are various steps that a user
may take ultimately generate revenue.

47. It’s up to you to figure out what drives
your business.

48. H I @ F R A M E D . I OFramed Data