Digital advertising industry is based on data. But without context, data is meaningless. Data and context should be used to improve the business. Otherwise we are useless.
Very often there is a friction when we have to explain to the decision makers what the data and context are telling us. It's not because there is no will to improve the business, but because we don't speak the same language.
We have to re-learn the ancient skill of storytelling.
3. Who am I? A GRANDFATHER 4.0
Miroslav Varga mech. eng. MEUS
miroslav@escapestudio.hr
4. ‘... Showing a story with data is more
valuable than the data alone ...’
Stolen from: https://www.ppchero.com/the-1-job-skill-for-digital-marketers-in-2020/
6. Human history in short
〜100.000 years ago we started developing our language to transfer knowledge
usually gathered around fire (and we still do so)
7. Human history in short
〜30.000 years ago we started to transfer knowledge
using (cave) paintings
8. Human history in short
〜4.000 years ago we started to transfer knowledge using text and writings
<script>
(function() {
// Change 'script' to whatever element you want to create, e.g.
'img' or 'input'.
var el = document.createElement('script');
// Add any standard attributes you need, e.g. 'src', 'width',
'type', 'name'.
// The syntax is el.setAttribute(attribute_name, attribute_value).
el.setAttribute('src', '/myScript.js');
// Add any non-standard attributes with the same method.
el.setAttribute('data-Simo-Ahava-script', 'My Script');
// Finally, inject the element to the end of <body>
document.body.appendChild(el);
})();
</script>
9. Human history in short
〜30 years ago, Powerpoint as a knowledge
transfer platform was born.
Today, there are still 3 Analytics experts alive
that witnessed that moment in History
10. Human history recap in short
〜100.000 years ago we started developing our language to transfer knowledge
〜30.000 years ago we started to transfer knowledge using cave paintings
〜4.000 years ago we started to transfer knowledge using text and writings
〜30 years ago, Powerpoint was born.
Which one do you think our brain is mostly adapted to?
This I learned from David JP Phillips
11. 1. Stories are part of our history
Now let’s check the structure of a good story.
What have we learned so far?
12. What can we learn from the best storytellers?
Let’s analyze a blockbuster structure:
Once upon a time _____
Every day ____________
One day _____________
Because of that _______
Because of that _______
Until finally ___________
This I copied from Michael Dolan
14. Let’s fill the blanks using NEMO
1. Once upon a time there was a widowed fish, named Marlin, who was extremely
protective of his only son, Nemo.
15. Let’s fill the blanks using NEMO
1. Once upon a time there was a widowed fish, named Marlin, who was extremely
protective of his only son, Nemo.
2. Every day Marlin warned Nemo of the ocean’s dangers and implored him not to
swim far away.
16. Let’s fill the blanks using NEMO
1. Once upon a time there was a widowed fish, named Marlin, who was extremely
protective of his only son, Nemo.
2. Every day Marlin warned Nemo of the ocean’s dangers and implored him not to
swim far away.
3. One day in an act of defiance, Nemo ignores his father’s warnings and swims into the
open water.
17. Let’s fill the blanks using NEMO
1. Once upon a time there was a widowed fish, named Marlin, who was extremely
protective of his only son, Nemo.
2. Every day Marlin warned Nemo of the ocean’s dangers and implored him not to
swim far away.
3. One day in an act of defiance, Nemo ignores his father’s warnings and swims into the
open water.
4. Because of that he is captured by a diver and ends up in the fish tank of a dentist in
Sydney.
18. Let’s fill the blanks using NEMO
1. Once upon a time there was a widowed fish, named Marlin, who was extremely
protective of his only son, Nemo.
2. Every day Marlin warned Nemo of the ocean’s dangers and implored him not to
swim far away.
3. One day in an act of defiance, Nemo ignores his father’s warnings and swims into the
open water.
4. Because of that he is captured by a diver and ends up in the fish tank of a dentist in
Sydney.
5. Because of that Marlin sets off on a journey to recover Nemo, enlisting the help of
other sea creatures along the way.
19. Let’s fill the blanks using NEMO
1. Once upon a time there was a widowed fish, named Marlin, who was extremely
protective of his only son, Nemo.
2. Every day Marlin warned Nemo of the ocean’s dangers and implored him not to
swim far away.
3. One day in an act of defiance, Nemo ignores his father’s warnings and swims into the
open water.
4. Because of that he is captured by a diver and ends up in the fish tank of a dentist in
Sydney.
5. Because of that Marlin sets off on a journey to recover Nemo, enlisting the help of
other sea creatures along the way.
6. Until finally Marlin and Nemo find each other, reunite and learn that love depends
on trust.
20. 1. Stories are part of our history
2. Stories have a structure.
Now let’s check is there any added value (money) in
creating stories?
What have we learned so far?
21. How stories add value? Can we measure it?
● Significant Objects by Rob Walker and Joshua Glenn,
● A literary and anthropological experiment,
● Demonstrated that narrative on any given object’s
subjective value can be measured objectively.
22. ● The project auctioned
off 200 thrift-store
objects via eBay.
● For item descriptions
they hired, short story
purpose-writers (over
200)
Significant Objects
24. 1. Stories are part of our history
2. Stories have a structure.
3. Stories can add measurable value.
Let’s make money from data!
What have we learned so far?
25. Inspired by: Dave Birss
How to make money from this data?
‘... every Friday millennials have a twentyseven
percent higher chance of wearing odd socks ...’
26. The context:
‘ ... the reason behind it is that Thursday is the
new Friday, it’s the new going out and drinking
night, and what happens is they’re hungover on a
Friday morning, and the alarm goes off and they
just pick up whatever is on the floor, and put on
their feet and they go ...’
27. Stories need data & context to make money
1. Produce a sensor system that rings an alarm on the doorsteps if the socks are
odd. The Friday morning odd socks eliminator or Friday MOSE*.
*MOSE in Italian = Moses but also MOdulo Sperimentale Elettromeccanico, Experimental
Electromechanical Module). It’s a project intended to protect the city of Venice, Italy, and the Venetian
Lagoon from flooding.
28. 1. Produce a sensor system that rings an alarm if the socks are odd at the
doorstep. The Friday morning odd socks eliminator or Friday MOSE.
2. Develop a special transfer plan: ‘Thursday evening taxi ride incentive
special’ - TETRIS - all rides before midnight will be 27% off for people who ride
with us also on next Friday morning. (We can even develop a plan that includes
the employers in some way - Hangover solution for employees - HSFE*).
*HSFE has a lot of meanings. It stands for Health, Safety, Fire and Environment protection but also for
High speed front end. The definition I like best stands for Hispanic Sports foundation for Education.
Stories need data & context to make money
29. 1. Produce a sensor system that rings an alarm if the socks are odd at the
doorstep. The Friday morning odd socks eliminator or Friday MOSE.
2. Develop a special transfer plan: ‘Thursday evening taxi ride incentive
special’ - TETRIS - all rides before midnight will be 27% off for people who ride
with us also on next Friday morning. (We can even develop a plan that includes
the employers in some way - Hangover solution for employees - HSFE).
3. Offer a new extra strong coffee called ‘Friday morning sucks - FMS* only
serving on Fridays with 27% higher caffeine, or quantity, or both.
● FMS stand for Free movement systems or Functional Movement Screen depending on Google’s results
Stories need data & context to make money
30. 1. Produce a sensor system that rings an alarm if the socks are odd at the
doorstep. The Friday morning odd socks eliminator or Friday MOSE.
2. Develop a special transfer plan: ‘Thursday evening taxi ride incentive
special’ - TETRIS - all rides before midnight will be 27% off for people who ride
with us also on next Friday morning. (We can even develop a plan that includes
the employers in some way - Hangover solution for employees - HSFE).
3. Offer a new extra strong coffee called ‘Friday morning sucks - FMS only
serving on Fridays with 27% higher caffeine, or quantity, or both.
4. Produce a strong anti-hangover drink called Thursday’s memory eraser TME*
(TME, or thiometaescaline, is a series of lesser-known psychedelic drugs)
Stories need data & context to make money
31. 1. Stories are part of our history.
2. Stories have a structure.
3. Stories can add measurable value.
4. Stories need data and context to make money
Now let’s add humans to the equation
What have we learned so far?
32. Humans can think associatively ...
Which phrase best describes two American gentlemen, one
Australian lady, one podcast and Analytics?
33. Humans can think associatively ...
Which phrase best describe two American
gentlemen, one Australian lady, one
podcast and Analytics?
34. Humans can think associatively ...
Which phrase best describe two American
gentlemen, one Australian lady, one
podcast and Analytics?
35. Humans can think associatively ...
Who is the person hiding behind this description?
- Brilliant mind,
- Whiskey lover,
- New Yorker from Austin (Tx)
- Has hard-to-understand presentations at Superweek
36. Humans can think associatively ...
Who is the person hiding behind this description?
Brilliant mind, Whiskey, New Yorker from Austin (TX) and
hard-to-understand presentations at Superweek
37. Humans can think associatively ...
Who is the person hiding behind this description?
Brilliant mind, Whiskey, New Yorker from Austin (TX) and
hard-to-understand presentations at Superweek
38. Humans can think associatively ...
What have William Shakespeare and a cuttlefish in common?
This was at a lecture from Dr Ranko Rajović
39. Humans can think associatively ...
What William Shakespeare and a cuttlefish have in common?
Why geese helped the European’s preserve all their knowledge?
This was at a lecture from Dr Ranko Rajović
40. 1. Stories are part of our history
2. Stories have a structure.
3. Stories can add measurable value.
4. Stories need data and context to make money
5. Stories should use people's ability of associative thinking
Now let us humans compete with the machine
What have we learned so far?
41. Let’s compete with the machine
Can you remember these 14 words in 90 seconds?
The Sun
Van Gogh
Painter
Three hundred
One
The Red Vineyard
Brother Theo
Handful
Food
Medical treatment
Getting Married
Right Ear
Poor
One Billion
This example was inspired by a lecture from Dr Ranko Rajović
42. The human body sends 11,2 million bits per second to
the brain for processing, yet the conscious mind seems
to be able to process only 50 bits per second.
What Each Human Senses Processes?
eyes - 10,000,000 bits per second
skin - 1,000,000 bits per second
ears - 100,000 bits per second
smell - 100,000 bits per second
taste - 1,000 bits per second
43. Maybe there is a solution ...
Now, I’ll show you 4 slides.
You will remember the 14 original (and +90 more) words
in the next ≈ 75s!
The beauty is - without almost any effort!
44. The power of a story:
This is the famous painting The Sun by Vincent Van Gogh
45. He was a painter who painted
three hundred paintings during
his lifetime. Unfortunately, he
sold only one while alive. It was
the painting The Red Vineyard.
The buyer was his younger
brother Theo who helped
Vincent financially and
emotionally as much as he could.
A handful of paintings Vincent
was forced to exchange for food
or medical treatment.
46. When his brother Theo announced he’s getting married, Vincent cut off his
right ear in an act of despair.
47. The paradox is that although Vincent died poor and misunderstood, today
only one of his paintings (The Starry Night) is about 1 billion U$ worth.
49. Now again: Test the importance of stories and
images
The Sun
50. Now again: Test the importance of stories and
images
The Sun
Van Gogh
51. Now again: Test the importance of stories and
images
The Sun
Van Gogh
Painter
52. Now again: Test the importance of stories and
images
The Sun
Van Gogh
Painter
Three hundred
53. Now again: Test the importance of stories and
images
The Sun
Van Gogh
Painter
Three hundred
One
54. Now again: Test the importance of stories and
images
The Sun
Van Gogh
Painter
Three hundred
One
The Red Vineyard
55. Now again: Test the importance of stories and
imagesThe Sun
Van Gogh
Painter
Three hundred
One
Red Vineyard
Brother Theo
56. Now again: Test the importance of stories and
imagesThe Sun
Van Gogh
Painter
Three hundred
One
Red Vineyard
Brother Theo
Handful
57. Now again: Test the importance of stories and
imagesThe Sun
Van Gogh
Painter
Three hundred
One
Red Vineyard
Brother Theo
Handful
Food
58. Now again: Test the importance of stories and
imagesThe Sun
Van Gogh
Painter
Three hundred
One
Red Vineyard
Brother Theo
Handful
Food
Medical treatment
59. Now again: Test the importance of stories and
imagesThe Sun
Van Gogh
Painter
Three hundred
One
Red Vineyard
Brother Theo
Handful
Food
Medical treatment
Getting Married
60. Now again: Test the importance of stories and
imagesThe Sun
Van Gogh
Painter
Three hundred
One
Red Vineyard
Brother Theo
Handful
Food
Medical treatment
Getting Married
Right Ear
61. Now again: Test the importance of stories and
imagesThe Sun
Van Gogh
Painter
Three hundred
One
Red Vineyard
Brother Theo
Handful
Food
Medical treatment
Getting Married
Right Ear
Poor
62. Now again: Test the importance of stories and
imagesThe Sun
Van Gogh
Painter
Three hundred
One
Red Vineyard
Brother Theo
Handful
Food
Medical treatment
Getting Married
Right Ear
Poor
One Billion
63. The original 14 keywords are easy to remember
when part of a story
The Sun
Van Gogh
Painter
Three hundred
One
The Red Vineyard
Brother Theo
Handful
Food
Medical treatment
Getting Married
Right Ear
Poor
One Billion
65. We are visual, associative and storytelling creatures:
The Sun
Van Gogh
Painter
Three hundred
One
Red Vineyard
Brother Theo
Handful
Food
Medical treatment
Getting Married
Right Ear
Poor
One Billion
66. That is our greatness.
It can be hidden.
But that doesn’t mean it doesn’t exist.
67. They thought, Van Gogh was crazy. But he knew
there can’t be any stars visible behind the moon!
69. 1. Stories are part of our history
2. Stories have a structure.
3. Stories can add measurable value.
4. Stories need data and context to make money
5. Stories should use people's ability of associative thinking
6. Stories can help us to catch up with the machine
Be careful with stories based on spreadsheets. About 24% of
all spreadsheets with mathematical calculations are wrong!
And above that, often people mix correlation with causation.
What have we learned so far?
70. Quote by Chad Sunderson before involving Analytics
‘... it's so important to develop a foundation of statistics that
allows you to understand what the data is actually telling
you.
And being able to read it, interpret it and translate it back to
people who matter...’
71. Reichenbach’s (1956) Common cause principle:
If two random variables x and y are statistically dependent then either:
(a) x causes y,
(b) y causes x, or
(c) there exists a third variable z that causes both x and y.
Further, x and y become independent given z
(Pearl, J.): A variable x has a causal influence on y if changing x leads to changes in y. This
position is very useful in practice.
Not everybody agrees with this (e.g., Cartwright).
72. Reichenbach in real life.
Relation between smoking and lung cancer?
To check the common cause principle we could:
(a) Force the general population to start smoking (x) and measure the
impact on the number of lung-cancers (y).
(b) Force all people with lung-cancer (y) to start smoking and measure the
impact on smoking of the general population (x).
(c) Assume that smoking (x) and lung-cancer (y) are caused by a third
variable - e.g. smog or stressful live (z).
Stress (z) causes people to start smoking (x) and stress increases the
number of all types of cancers and therefore also lung-cancers (y).
(x) and (y) are independent given (z)
73. Croatia would be
here if Tesla had
accepted the
Nobel Prize
Often people
assume causality
between correlated
datasets.
Eg. Impact of Chocolate
consumption on number of
Nobel Prizes per capita
per Country
Published by Messerli
I recommend to visit Tyler’s Vigen web.
74. Sometimes Mathematics
is counter-intuitive
A real-world example:
Looking at people who are single and
who are in a relationship as separate
groups, being attractive (x) and being
intelligent (y) are two independent traits:
Published by Dablander
75. Z is called The collider
Let’s make the reasonable assumption that both,
being attractive and being intelligent, are positively
related with being in a relationship.
What does this imply?
76. Z is called The collider
Let’s make the reasonable assumption that both,
being attractive and being intelligent, are positively
related with being in a relationship.
What does this imply?
First, it implies that, on average, single people are
less attractive and less intelligent (red data points).
77. Z is called The collider
Let’s make the reasonable assumption that both,
being attractive and being intelligent, are positively
related with being in a relationship.
What does this imply?
First, it implies that, on average, single people are
less attractive (x) and less intelligent (red data
points).
Second, and perhaps counter-intuitively, it implies
that in the population of single people (red), being
attractive and being intelligent are negatively
correlated.
(If the handsome person you met at the bar were
also intelligent, then she would most likely be in a
relationship!)
78. Keep in mind segmentation - the Simpson Paradox
More men and women recover when taking the drug (93% and 73%) compared to when not taking the drug
(87% and 69%).
When taken together, fewer patients who took the drug recovered (78%) compared to patients who did not
take the drug (83%). Should a doctor prescribe the drug or not?
Suppose you observe N=700 patients who either choose to take a drug or not
Adapted from Pearl, Glymour & Jewell
79. Does taking the drug have a positive effect on recovery?
If we would force everyone to take drugs, the benefit is:
p(R=1 ∣ do(D=1))=∑ p(R=1 ∣ D=1, S=s) p (S=s) =
= p(R=1 ∣ D=1, S=0)p(S=0)+p(R=1 ∣ D=1, S=1)p(S=1)=
= 81/87×(87+270)/700+192/263×(263+80)/700=
≈0.832
Answering this question requires knowledge about the interventional distributions (ask Matt):
p(R∣do(D=0)) and p(R∣do(D=1))
s
If we restrict everyone to take drugs, the benefit is:
p(R=1 ∣ do(D=0))=∑ p(R=1 ∣ D=0, S=s) p (S=s) =
= p(R=1 ∣ D=0, S=0)p(S=0)+p(R=1 ∣ D=0,
S=1)p(S=1)=
= 234/270×(87+270)/700+55/80×(263+80)/700=
80. Take a look at this Analytics data
This is a simple one-week data segment:
New Visitors generate 0,6 of all Sessions; Returning Visitors generate 0,4 of all sessions. For every 17
sessions, 10 are generated from New Visitors and 7 are generated from Returning Visitors.
New Visitors Revenue participation is 0,25 and Returning Visitors Revenue participation is 0,75.
This is the only part created by me
One Returning Visitor generate ≈ 4.24x more revenue than One New Visitor.
81. Drawing this data is easy
0,4
0,6
0,25
0,75
SESSIONS REVENUE
0 1
1 1
x = 0,29
Intersection is at:
(0.75-0.4)x +0.4=
=0.35x + 0.4 =
(0,25-0.6)x +0.6=
=-0,35x + 0,6 =>
0,7x = 0,2 =>
x = 2/7 = 0,286
NEW RETURNING
82. Doing some ‘what if’ mental exercise
0,4
0,6
0,25
0,75
SESSIONS REVENUE
0 1
1 1
0,28 0,50
≈ 0,15 more NV could
move the intersection to 0.5
NEW RETURNING
83. Let’s create our story based on Analytics data:
This is a simple one-week data segment.
One possible story:
If we increase the number of sessions generated by New Visitors for 10000, that could increase
the number of sessions generated by Returning visitors for 7000 and maybe increase the overall
revenue for ≈ 50% (21821/2 + 64468/2 ≈ 43000)
84. 1. Stories are part of our history
2. Stories have a structure.
3. Stories can add measurable value.
4. Stories need data and context to make money
5. Stories should use people's ability of associative thinking
6. Stories can help us to catch up with the machine
7. Stories are helpful, even in statistics
What have we learned so far?
85. Wrapping everything up:
Try at your next meeting with your decision maker
explain data in a story like this:
86. Once upon a time Google Analytics was installed.
87. Once upon a time Google Analytics was installed.
Every day data has been checked.
88. Once upon a time Google Analytics was installed.
Every day data has been checked.
One day comparing New vs Returning visitor sessions an interesting fact
stand out. For every 17 sessions on our site, 10 are from New and 7 from
Returning visitors.
89. Once upon a time Google Analytics was installed.
Every day data has been checked.
One day comparing New vs Returning visitor sessions an interesting fact
stand out. For every 17 sessions on our site, 10 are from New and 7 from
Returning visitors.
Because of that let’s try to increase the sessions of New visitors by 10.000.
90. Once upon a time Google Analytics was installed.
Every day data has been checked.
One day comparing New vs Returning visitor sessions an interesting fact
stand out. For every 17 sessions on our site, 10 are from New and 7 from
Returning visitors.
Because of that let’s try to increase the sessions of New visitors by 10.000.
Because of that we could expect also to increase the number of sessions
from Returning visitors by 7000. Returning visitors spend 4.24 times more
than New Visitors.
91. Once upon a time Google Analytics was installed.
Every day data has been checked.
One day comparing New vs Returning visitor sessions an interesting fact
stand out. For every 17 sessions on our site, 10 are from New and 7 from
Returning visitors.
Because of that let’s try to increase the sessions of New visitors by 10.000.
Because of that we could expect also to increase the number of sessions
from Returning visitors by 7000. Returning visitors spend 4.24 times more
than New Visitors.
Until finally we achieve a 50% increase in Revenue.