Start-up Metrics for Convicted Pirates
Inspiration from Dave McClure & David Jones
Stephen Merity
smerity@smerity.com
Who am I?
Graduated
BIT (Hons) First Class + University Medal
from University of Sydney
Previously I've worked at start-ups including
Google & Freelancer.com
& done consulting in machine learning + data mining.
All these start-ups live & breath metrics!
Why metrics?
A start-up without metrics
is like
a car without a windshield
is like
a car without a windshield
Most companies have no clue
what's happening on a daily basis...
Don't be one of them! ಠ_ಠ
what's happening on a daily basis...
Don't be one of them! ಠ_ಠ
Why metrics?
Commonwealth Bank can tell when ads are
on TV thanks to the visits to NetBank
on TV thanks to the visits to NetBank
Freelancer.com have a list of public holidays for each country on earth so they can explain traffic dips
What's most likely to kill a start-up is not a competitor, a crashed server, a hacker, ...
It's not knowing what customers are doing and
whether they're happy!
It's not knowing what customers are doing and
whether they're happy!
Build, Measure, Learn
Speed of iteration ➡ successful product fit
Start-up Metrics for Pirates
-
Acquisition
-
Activation
-
Retention
-
Referral
- Revenue
Pirates define your funnel
-
Acquisition
200 users sign up -
Activation
100 users enter profile details -
Retention
30 users return within a week -
Referral
5 users send invites ➡ 10 sign-ups -
Revenue
10 users sign up for $8 per month
Acquisition
How do you get new users?
How do you get the right new users?
Different channels have different conversion rates
Stop targeting TechCrunch!
The first step in determining the cost of acquisition
(special note: a user who refers another free = ½ CPA)
Stop targeting TechCrunch!
The first step in determining the cost of acquisition
(special note: a user who refers another free = ½ CPA)
Activation
...or "Oh, you have 4 million users, eh?"
A user isn't a user unless they've actually used it!
What do you need users to do before they're useful?
How can you encourage those activities?
How can you encourage those activities?
- Amazon.com ➡ Enter (valid) credit card details
- LinkedIn ➡ Fill initial profile details
- DropBox ➡ Added one file & syncing to one computer
Retention
What indicates they're coming back in 7/14/30 days?
(Facebook realised photo uploads + tagging are vital)
(Facebook realised photo uploads + tagging are vital)
Cohort analysis helps answer the question:
"How do all the new changes impact users?"
"How do all the new changes impact users?"
Referral
Social proof / word of mouth is amazingly strong:
if you have such a customer base use it!
Don't solely rely on it for user acquisition!
Fraught with danger & potential:
Twitter/Facebook can block you/change their API
You can be perceived as spammy & lose user trust
Revenue
What's the life time value of the average customer?
Single transaction? Recurring subscription?
If it's subscription, how long do users remain with you?
assert(CPA < LTV)
Work out your
cost per acqusition (CPA)
cost per acqusition (CPA)
and your
life time value (LTV)
This defines how much you can spend in acquisition & whether you'd benefit from "spamming money"
Your Key Metrics
Don't measure everything. It'll send you insane.
What are the key business factors
you need measured?
you need measured?
How can you understand them with metrics?
A/B Testing
is all the rage, but...
You likely won't (yet!) have enough traffic.
You likely won't (yet!) have enough traffic.
Hints & Tips
Track your history in Google Analytics
(Freelancer.com)
(Freelancer.com)
Gamification is over-hyped but stunningly useful
(·: CoderWall)
Really real time analytics for PR
(Are you on Sunrise? Watch the rise & fall)
Really real time analytics for PR
(Are you on Sunrise? Watch the rise & fall)
Hook metrics into your support system
(You live and die by how you respond to customers!)
Easy Mistakes in Metrics
If metrics aren't easy to check,
they won't be checked!
they won't be checked!
Metrics might not reveal the root cause of a problem
Metrics without statistics are indications:
a tick up or down does not mean that new feature is vital!
The Viral Loop
Viral growth factor:
(% of users who invite others) × (average # invited)
× (% of invited who accept invitation)
(% of users who invite others) × (average # invited)
× (% of invited who accept invitation)
If factor > 1 then you have organic exponential growth
Main point: don't rely on virality.
Use referrals to supplement your traditional methods.
So... Make metrics a habit!
Metrics are the key
to understanding your start-up
to understanding your start-up
and
they drive the Lean Start-up's
Build-Measure-Learn feedback loop
Build-Measure-Learn feedback loop
Every action should be metric measurable:
then you can properly celebrate success! =]
Stephen Merity
smerity@smerity.com
smerity@smerity.com