1. Produced by
Customer Success Summit 2014
Customer Success Summit 2014
In-product growth hacking to increase revenue
Harpreet Singh
Senior Director, Product Management,
CloudBees
@singh_harpreet hsingh@cloudbees.com
2. Produced by
Customer Success Summit 2014
• #1 Continuous Delivery PaaS
• Global startup:
– Aus, NZ, France, Switzerland,
Netherlands, Belgium, Ireland, US
(MA, VA, TX, IL, WA, CA)
• The most fun place I have ever worked
J
2
3. Produced by
Customer Success Summit 2014
What led me to Customer Success
• Setup GlassFish monetization – Sun’s
application server
• Customer Success
– “People like being helped and not sold to”
• @CloudBees have setup, managed the Jenkins
biz establishing it from 0 -> multi-million
3
4. Produced by
Customer Success Summit 2014
– improve and fine-tune what I’m offering?
– connect more intelligently with my users?
– help prospects succeed on the platform?
– convert free users to paying customers?
How can I…
4
5. Produced by
Customer Success Summit 2014
When my customers are…
under
immense
-me
to
market
pressures
smart
developers,
focused
on
delivering
cool
technologies
6. Produced by
Customer Success Summit 2014
Fairly
sta*c
customer
lists
Lifecycle Marketing
in the Packaged Software World
Lead
Nurturing
Considera*on
Sale
Post
Sale
They
no-ced
us!
This
is
why
we
are
special
Seems
like
they
might
buy
Sold!
Upsell!
Most
systems
in
house:
Email
marke-ng,
customer
lists
Marke*ng
7. Produced by
Customer Success Summit 2014
Lifecycle Marketing
in the Cloud Era (Freemium Model)
Lead
Nurturing
Considera*on
Free
Tier
Base
Tier
Add
ons
Enterprise
Tier
Post
Sale
Most
systems
external:
Hosted,
best-‐of-‐breed
services
Users
can
move
freely
up
and
down
the
-ers
Evangelist
“Growth
Hacker”
Customer
Success
Manager
7
Short
cycle,
as
users
can
try
the
free
-er
and
make
the
determina-on
on
product
feasibility
themselves
Marke*ng
8. Produced by
Customer Success Summit 2014
Lessons Learned: Fail fast
• Lesson #1
– Monitor everything – data, data, data…
• Lesson #2 –
– Health-based metrics. Flag and address dis-
engagements real time
• Lesson #3 –
– Activity-based Metrics: Hypothesis testing & fail fast
8
9. Produced by
Customer Success Summit 2014
Subscribed to a service
• Jenkins
• RUN@cloud (Deployment
PaaS)
• Forge (Git, SVN)
• DB service
• Partner service
– MongoHQ
– NewRelic
– PaperTrail
– …
Performed an action
• Web UI login
• Jenkins
– Created a Job, Installed a
plugin, Ran a job…
• Deployment PaaS
– Deployed an app, restarted
app, app stopped, app
hibernated…
• Others
– Canceled account, added a
paid DB, added a user
Monitor everything
Play: Sensors everywhere in the product
10. Produced by
Customer Success Summit 2014
• Time and activity
based
• Grade users based on
risk profile
• Separate onboarding
from established
users
• Offer hands-on help
for established users
Health-based metrics
Play: Flag and address dis-engagements real time
10
Green
Sign-‐Up
Yellow
Red
Lost
Use?
Use?
Use?
1
wk
2
wks
3
wks
4
wks
Time
Onboarding
Yellow
Red
Lost
Use?
Use?
Established
Green
Use?
Y
Y
Y
N
N
N
N
N
N
Y
Y
Y
Convert
11. Produced by
Customer Success Summit 2014 11
• Track number and types of activity
over time
– Scale and adjust “smart” communication
over time
• Users can become inactive and then
reactivate
• Technical, not sales communication
Green
Paid
Yellow
Alarm
Bells
Lost
Use?
Use?
Use?
1
wk
2
wks
3
wks
4
wks
Time
Paying
Y
Y
Y
N
N
N
Health-based metrics
Play: Flag and address dis-engagements real time
12. Produced by
Customer Success Summit 2014
• Popularity of Play!
– More articles here
– Reach out to the
community
– Better monitoring
– …
12
Play2
Java
EE6
WP
Hibernate+Tomcat
Spring+Hibernate
CF
Grails
PetClinic
CF
Node.js
Glassfish
AngularJS
Java
GAE
Ac*vi*
Explorer
Sencha+Clojure
Facebook
Backbone+JAX-‐RS
Scala+LiU
Java
Sauce
Clojure/Compojure
Java
EE6
Sauce
Activity-based metrics
Play: Hypothesis testing and fail fast
14. Produced by
Customer Success Summit 2014
Established
respond
2:1
over
onboarding
Insight into behavior with the pre-revenue funnel
30%
on-‐boarders
@
Risk
10%
are
sales
opportuni-es
30%
of
paid
customers
were
helped
by
developer
success
engineers
just
before
they
converted
So, as software developers, we really should be doing for our customers what we prefer ourselves.
In the cloud, as-a-service world, with a freemium (sign up for free and try it) model or a time-trial model, it’s different. First of all the evaluation cycle is short. People try it out and decide without ever talking to you or a salesperson whether they want to invest more in it. Once they get serious, they might move to different tiered pricing that is tied to the level of usage, users, or something. In this world, Marketing’s more traditional activities are to drive the leads from anonymous visitors into known users. The other parts are now engaged in by people with different titles – evangelists, growth hacker (a trendy silicon valley term!), and customer success managers.
Let’s make it more concrete by using CloudBees as an example. What we’re interested in is how people are using our service. Part of that is: what service are they using – like “what software has someone bought” in the old packaged software world. But more specifically, we want to know how engaged they are. Are they doing builds? Are they deploying new apps on a regular basis. Are they adding new users?
But in our as-a-service world, we can so a lot better job. (Forgive my making a garbage collection joke here on the slide by using name of object spaces and collectors!) In our world, we have people who sign up but don’t’ do anything after looking around. We classify them as “onboarding”, and we engage with them differently than we do with those who dig in and actually start to use the service.
But probably more importantly, once someone has stepped up to paying for the service, we have an opportunity to keep them happy using it in ways that are obviously just not possible with packaged software. Of course, part of the trick here is to know your audience, which being mostly developers, just want to be left alone.
And this gives us some data that helps us determine where to put our investments. Now this changes because we roll out new ClickStarts every week, and some are accompanied by more visibility than others, but this is actual data we also share on our developer blog. What you can see here is that Play is very popular on CloudBees, as are the traditional Java stacks mixed with JavaScript UI frameworks like Backbone. Node.js is also popular, primarily because people want to find out what it’s about as far as we can tell. At any rate, we’re getting better at sorting out causality on these technologies as we go along. Today the data is there, but it’s hard to interpret.