A guide to helping start-ups with building and operating data systems for growth. Highlights what makes a good metric, how to define the right metrics for your business, and then how to build data infrastructure so that you can collect the relevant data.
Presented at Grow Camp 2018 in MaRS Discovery District in Toronto, Canada.
2. Who here has heard about
experimentation in Grow
Camp so far?
3. Build -> Measure -> Learn
What you have learned so far
1. BRAINSTORM 2. PRIORITIZE 3. TEST
4. IMPLEMENT
5. ANALYZE
6. SYSTEMIZE
4. Build -> Measure -> Learn
Why you still can’t act on it
1. BRAINSTORM 2. PRIORITIZE 3. TEST
4. IMPLEMENT
5. ANALYZE
6. SYSTEMIZE
No clear KPIs
Poor tracking
Bad instrumentation
Wrong conclusions
6. Data can tell you any story
with some creativity (and
manipulation).
7. Agenda
What I’ll be covering today
Mindset
What makes a good metric
Metrics related to your business
Toolset
Building data infrastructure
Recap / Lunch
Skillset
In-depth examples
Refining your metrics
8. Who am I?
A tidbit of my work
Monetization
Referral programs
Customer development
Product analytics
Data infrastructure
Metrics & reporting
Web analytics
Channels and campaigns
Conversion rate optimization
9. Agenda
What I’ll be covering today
Mindset
What makes a good metric
Metrics related to your business
Toolset Building data infrastructure
Recap / Lunch
Skillset
Analyze your business
Find the metrics that matter
10. To lead into the workshop, can
you take 2 minutes now to jot
down a few metrics you are
tracking currently?
11. Qualitative vs. Quantitative
Types of metrics
Qualitative
• Observable
• Behaviours
• Quality
• Surveys, customer
interviews, etc.
Quantitative
• Measurable
• Numbers and stats
• Quantity
• Tracking usage, clicks,
etc.
12. Example: Survey vs. button clicks
Types of metrics
What people thought about the website. How many clicks did a button get?
13. Vanity vs. Actionable
Types of metrics
Vanity
Numbers or stats that look
good on paper, but don’t
really mean anything
important.
Actionable
Stats that tie to specific and
repeatable tasks you can
improve and to the goals of
your business.
14. Example: Total users vs. users registered per month
Types of metrics
0
100
200
300
400
500
600
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Total Users
0
20
40
60
80
100
120
Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Net New Users Per Month
15. Lagging vs. Leading
Types of metrics
Lagging
Historic metric that shows
how you’re doing
Leading
Number today that shows a
metric tomorrow.
16. Example: Sales results vs. new leads
Types of metrics
Sales results for the past quarter.
New sales leads can help predict
revenue for the quarter.
17. Correlation vs. Causation
Types of metrics
Correlation
Statistical measure (a
number) that describes the
size and direction of a
relationship between two or
more variables.
Causal
Indicates that one event is
the result of the occurrence
of the other event.
19. Output vs. Outcome
Types of metrics
Output
Tells the story of what you
produced or your
organization's activities.
Outcome
The level of performance or
achievement that occurred
because of the activity or
services your organization
provided.
20. Example: 10 blog posts vs. +30% time on site
Types of metrics
Write 10 blog posts Increase time on site by +30%
21. What makes a good metric?
§ Understandable: What does it mean?
§ Comparative: Against time, users, etc.
§ Ratio/Rate: X variable / Y variable
§ Behaviour Changing: Does this invoke change?
Understanding metrics
22. Example: Free Trial Conversion
§ Understandable: How many users converted from free
trial to paid per week
§ Comparative: 10% free trial to paid conversion vs. 10%
free trial to paid conversion per week
§ Ratio/Rate: Total users who completed free trial vs.
converted customers / free trials started
§ Behaviour Changing: It is low, so we need to test to
improve the trial experience
Understanding metrics
23. Agenda
What I’ll be covering today
Mindset
What makes a good metric
Metrics related to your business
Toolset Building data infrastructure
Recap / Lunch
Skillset
In-depth examples
Refining your metrics
24. Metrics depend on your business type
But they generally are similar: CPA, CAC, LTV, ARPU, Churn
Source: Lean Analytics
25. SaaS Product Example
Acquisition
Channel Traffic Conversion
CAC, CPA by persona
CAC, CPA by channel
Multi-touch
attribution
Landing Page
Optimization
Real time
personalization
Checkout flow
conversion
Drop-off paths
Pricing elasticity test
26. CPA vs. CAC
Acquisition
Cost Per Acquisition
Measures the cost to acquire
something that is not a
customer — for example, a
registration, activated user,
trial, or a lead
Customer Acquisition
Cost
Measures the cost to acquire
a customer.
27. Factors Included in CAC
Acquisition
§ Tools, overhead, salaries are all marketing expenses
§ Measure over time, by channel, persona, etc.
§ Non fully loaded CAC = understand channel performance
§ Fully loaded CAC = understand holistic marketing performance
!"! =
$%&'()*+, + .%/(. (01(+.(.
# 34 56.)37(&. %586*&(9
28. Analyze Channel Performance Alone
Display Campaign Example
Metric Value
Date Feb 1-7
Campaign Name Campaign-1
Spend $10,000
Impressions 4,000,000
Clicks 1,500
Subscriptions 32
Billings $16,000
ROAS ($16,000/$10,000)*100 = 60%
CAC ($10,000/32) = $312.50
30. Autodesk spends $ on display, affiliate,
and SEM with all owners reporting
positive and growing return on ad
spend (ROAS). But new seats are
falling. How can this be the case?
31. Using data to check view-through value
§ Run an experiment with a charity ad and same ad spend
§ Incremental impact that view-through has is +17%
Example display experiment
Test Cell Spend Orders via
Display
Orders that also
touched SEO/SEM
Campaign-1 $1,000 100 70
Red Cross
Campaign
$1,000 50 50 +17%
+20 orders
35. Test to Improve User Experiences
Activation
Analyze correlation data to find causations that
improve first user experiences.
Source: Amplitude
36. Examples of Activation Strategies
Activation
DropBox: Your first file is backed up from your computer into the
cloud
Facebook: You connect with 7 friends within your first 10 days
Stack Overflow: Your question is answered
Instacart: When your groceries are delivered for the first time
Instagram: Someone likes one of your photos
38. If someone isn’t using your product, they will churn
Usage-based retention
Source: Amplitude
39. Cohort analysis key to retention
Transaction based retention
Source: Christoph Janz
Aggregated retention rates for high level reporting. By cohort for improving it.
0
200
400
600
800
1,000
1,200
Jan-13 Feb-13 Mar-13 Apr-13 May-13 Jun-13 Jul-13 Aug-13 Sep-13 Oct-13
Jan-13 Feb-13 Mar-13 Apr-13 May-13 Jun-13 Jul-13 Aug-13 Sep-13 Oct-13
40. Customer retention chart
Transaction based retention
0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
70.00%
80.00%
90.00%
100.00%
J
a
n
-
1
3
F
e
b
-
1
3
M
a
r
-
1
3
A
p
r
-
1
3
M
a
y
-
1
3
J
u
n
-
1
3
J
u
l
-
1
3
A
u
g
-
1
3
S
e
p
-
1
3
O
c
t
-
1
3
Customer Retention Chart
41. Use this data to build LTV models
Transaction based retention
There are many ways to calculate LTV. Fully loaded or non fully loaded.
Source: Christoph Janz
Where:
• ARPU = average revenue per user
• Avg. Cust. Lifetime, n = the inverse of the churn, n=1/annual churn
• WACC = weighted average cost of capital
• Costs = annual costs to support the user in a given period, cost of goods sold
• CAC = customer acquisition costs
!"# = %
&'(
)
*+,-& − /0121&
(1 + 6*//)&
− /*/
47. Baselines
§ Pixlr was monetizing
via ads so spent bulk
of time on retention
§ Understood we were
just in line with
retention for photo &
video apps
Product Retention Rates
Source: Flurry Benchmark
51. Getlatka.com - tons of more benchmarks
Baselines
Source: GetLatka.com
Churn
2%/month
Free to paid
2% of free users
Page load time
<5 seconds
LTV: CAC
3:1
52. Agenda
What I’ll be covering today
Mindset
What makes a good metric
Metrics related to your business
Toolset Building data infrastructure
Recap / Lunch
Skillset
Analyze your business
Find the metrics that matter
55. Founding Stage (0 to 20 employees)
Data Infrastructure
Track only the metrics that move your business
forward. After your first marketing hire, use
UTM tracking.
Web tracking
Product tracking
Financial Reporting
56. Mid Stage (20 to 100 employees)
Data Infrastructure
Start forming the first data pipeline. Hire first
analytics lead.
ETL
Collect Load & Operate Transform & Analyze
Enrich
57. Extract, Transform, and Load
Data Infrastructure
Do not build in-house ETL. As employees leave, becomes really
difficult and costly to maintain.
58. Late Stage (100+ employees)
Data Infrastructure
Migrate existing event/web tracking tools to
Snowplow due to poor scaling costs. Extract data
from marketing tools to data warehouse. Document.
Collect Load & Operate Transform & Analyze
Enrich
ETL
60. Summary
Recap
1. Metrics should be understandable,
comparative, a ratio, and behaviour changing.
2. Find metrics that matter to your business,
measure them, slice and dice them, and
compare them to a baseline.
3. Start using data early and add more capability
as you scale.
61. Agenda
What I’ll be covering today
Mindset
What makes a good metric
Metrics related to your business
Toolset Building data infrastructure
Recap / Lunch
Skillset
In-depth examples
Refining your metrics
63. Agenda
What I’ll be covering today
Mindset
What makes a good metric
Metrics related to your business
Toolset Building data infrastructure
Recap / Lunch
Skillset
In-depth examples (20-30 min)
Refining your metrics (remainder)
65. Discuss the performance of this
company by the metrics
provided.
https://autodesk.box.com/v/grow-camp
66. Litmus test of any company’s financial
success is the ability to acquire many
high lifetime value (LTV) customers.
Being LTV-centric is at the heart of
being customer centric.
Daniel McCarthy, Asst. Professor, Emory University
Source: Daniel McCarthy
70. 72% of customers churn by month six
100% - 28% = 72%
45%
38%
34%
31%
30%
28%
27% 26% 25% 25% 24% 23%
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
1 2 3 4 5 6 7 8 9 10 11 12
Percent
of
Cohort
Still
Active
Number of Months Since Acquisition
Subscriber Retention by Number of Months Since Acquisition
Source: Daniel McCarthy
71. Market saturation shown by increasing CAC
9.5% growth in CAC from Feb to Mar 2017
$105
$92
$82
$73
$68
$64
$56
$52
$48
$56
$58
$61
$61
$63
$65
$64
$66
$67
$63
$62
$61
$72
$78
$84
$85
$90
$94
$105
$115
$124
$114
$113
$111
$134
$147
$161
$-
$20
$40
$60
$80
$100
$120
$140
$160
$180
A
p
r
2
0
1
4
M
a
y
2
0
1
4
J
u
n
2
0
1
4
J
u
l
2
0
1
4
A
u
g
2
0
1
4
S
e
p
2
0
1
4
O
c
t
2
0
1
4
N
o
v
2
0
1
4
D
e
c
2
0
1
4
J
a
n
2
0
1
5
F
e
b
2
0
1
5
M
a
r
2
0
1
5
A
p
r
2
0
1
5
M
a
y
2
0
1
5
J
u
n
2
0
1
5
J
u
l
2
0
1
5
A
u
g
2
0
1
5
S
e
p
2
0
1
5
O
c
t
2
0
1
5
N
o
v
2
0
1
5
D
e
c
2
0
1
5
J
a
n
2
0
1
6
F
e
b
2
0
1
6
M
a
r
2
0
1
6
A
p
r
2
0
1
6
M
a
y
2
0
1
6
J
u
n
2
0
1
6
J
u
l
2
0
1
6
A
u
g
2
0
1
6
S
e
p
2
0
1
6
O
c
t
2
0
1
6
N
o
v
2
0
1
6
D
e
c
2
0
1
6
J
a
n
2
0
1
7
F
e
b
2
0
1
7
M
a
r
2
0
1
7
Estimated Marketing Spend Per Customer Acquisition
Source: Daniel McCarthy
72. As customers become more loyal, they spend less
30% less monthly revenue per active customer over seven month span
$147
$124
$120
$115 $113 $109
$104
$-
$20.00
$40.00
$60.00
$80.00
$100.00
$120.00
$140.00
$160.00
$180.00
$200.00
Sep 2016 Oct 2016 Nov 2016 Dec 2016 Jan 2017 Feb 2017 Mar 2017
Expected Monthly Revenue Per Active Customer
September 2016 Acquisition Cohort
Source: Daniel McCarthy
73. LTV has decreased by 20% over two year span
There is only so much the most loyal customers can spend
$997
$977
$956
$936
$925
$914
$905
$895
$885
$873
$855
$837
$911
$889
$868
$846
$833
$821
$812
$830
$848
$864
$873
$881
$808
$789
$770
$750
$740
$729
$721
$724
$728
$729
$-
$200.00
$400.00
$600.00
$800.00
$1,000.00
$1,200.00
J
a
n
2
0
1
4
F
e
b
2
0
1
4
M
a
r
2
0
1
4
A
p
r
2
0
1
4
M
a
y
2
0
1
4
J
u
n
2
0
1
4
J
u
l
2
0
1
4
A
u
g
2
0
1
4
S
e
p
2
0
1
4
O
c
t
2
0
1
4
N
o
v
2
0
1
4
D
e
c
2
0
1
4
J
a
n
2
0
1
5
F
e
b
2
0
1
5
M
a
r
2
0
1
5
A
p
r
2
0
1
5
M
a
y
2
0
1
5
J
u
n
2
0
1
5
J
u
l
2
0
1
5
A
u
g
2
0
1
5
S
e
p
2
0
1
5
O
c
t
2
0
1
5
N
o
v
2
0
1
5
D
e
c
2
0
1
5
J
a
n
2
0
1
6
F
e
b
2
0
1
6
M
a
r
2
0
1
6
A
p
r
2
0
1
6
M
a
y
2
0
1
6
J
u
n
2
0
1
6
J
u
l
2
0
1
6
A
u
g
2
0
1
6
S
e
p
2
0
1
6
O
c
t
2
0
1
6
Acquisition cohort
Cumulative Revenue After 6 Months Per Active Customer by
Acquisition Cohort
Source: Daniel McCarthy
74. CAC: LTV Analysis
§ No COGS included in six month LTV
§ If variable contribution margin = 26% and CAC = $161, then $619 in net revenue
§ Net revenue = gross revenue minus returns and promotional discounts)
§ But wait… $619 in net revenue comes in five months
§ Correct, but remember only 30% of customers were around after 5 months
§ So on the remaining 70%, negative LTV:CAC
§ At $161 per customer, only 30% of customers actually are profitable
§ Break even moves further away as revenue per customer declines and CAC rises
$729: $114 = 6.39x ROI (Oct 2016)
75. Agenda
What I’ll be covering today
Mindset
What makes a good metric
Metrics related to your business
Toolset Building data infrastructure
Recap / Lunch
Skillset
In-depth examples
Refining your metrics
76. Pull back out that list of
metrics you made earlier. What
stage are you in, what metric is
it, and would you change it? If
yes, how?