Less than 1% of potential customers come to your website. Few sign-up, even less convert. It’s not a funnel, but a pasta strainer. In the past Guillaume “G” Cabane has shown how to automate advanced marketing personalization once prospects land on a website. Now G will reveal his master plan and first steps towards automating marketing based on intent … before the website. Contacting the right people with the top companies, at the right time, through the best channel with the perfect message. At scale.
2. Y I E L D B A S E D A U T O M AT I O N
M A R G I N B A S E D A U T O M AT I O N
Guillaume “G” Cabane
VP OF GROWTH @DRIFT
ADVISOR @SEGMENT
@GUILLAUMECABANE
#GHCONF18
MARGIN > REVENUE > LEADS
3. LONG TERMVISION
Create an AI
Based on a multiplicity of buying signals
Predicts the next best action
to maximize margin
4. Y I E L D B A S E D A U T O M AT I O N
AI
WEB TECHNOGRAHICS
TOPIC SURGING
MOBILE TECHNOGRAHICS
CUSTOM SITE SCRAPING
Chat
Email
Web content
Advertising
Account, User and
Campaign Record
Physical goods
Long term vision
VISITED WEBSITE
SCORE PROSPECTING
+
5. Y I E L D B A S E D A U T O M AT I O N
Progress towards vision
Value based
predictions
On-site
optimization
Prediction based
outreach
Unexpressed
intent discovery
2017
Value based
outreach
AI Based
sequencing
Margin
prediction
6. Y I E L D B A S E D A U T O M AT I O N
2017 RECAP
From spray & pray to intent driven automation
7. The days of Spray & Pray are over
Quality > Quantity
8. Marketers use spray & pray because it’s cheap and
they don’t know much about their contacts
Knowledge is what changed in 2017
9. Y I E L D B A S E D A U T O M AT I O N
SalesVelocityin$/month
Company size
5 50 200 1,000 10,000
Sales Velocity
Cost of sales
leadsProfitable
4 Metrics of Sales Velocity
# Number of leads
# x
$ Average contract value (ACV)
$ x
% Conversion rate
%
L Time to conversion
L
10. Y I E L D B A S E D A U T O M AT I O N
4 Metrics of Sales Velocity
# Number of leads
# x
$ Average contract value (ACV)
$ x
% Conversion rate
%
L Time to conversion
L
SalesVelocityin$/month
Company size
5 50 200 1,000 10,000
Sales Velocity
Cost of sales
leadsUnprofitable
11. Y I E L D B A S E D A U T O M AT I O N
SalesVelocityin$/month
Company size
5 50 200 1,000 10,000
Sales Velocity Concept and Expectation
Sales Velocity
Cost of sales
12. Y I E L D B A S E D A U T O M AT I O N
SalesVelocityin$/month
Company size
5 50 200 1,000 10,000
Sales Velocity Concept and Expectation
Sales Velocity
Cost of sales
leadsProfitable
13. Too often we think…
“Growth = Automation
Automation = Lower costs”
15. For “Quality > Quantity” to be true…
We need to know more about our leads
And we need to predict the sales velocity
16.
17. Y I E L D B A S E D A U T O M AT I O N
GET IP
RETURNS DOMAIN
2017: post-visit Optimizations
18. Y I E L D B A S E D A U T O M AT I O N
0%
19%
38%
56%
75%
Low Medium Good Very Good Low Medium Good Very Good
Won Opportunities
MadKudu Customer Fit
Percentage of Lead Types
79% of won opportunities
16% of leads make up
19. Y I E L D B A S E D A U T O M AT I O N
SalesVelocityin$/month
Company size
5 50 200 1,000 10,000
Sales Velocity Concept and Expectation
Sales Velocity
Cost of sales
20. Y I E L D B A S E D A U T O M AT I O N
Company size
5 50 200 1,000 10,000
Enterprise
ABM
Self-Service
Marketing
Automation
21. It is now possible to know the quality
of anonymous traffic per source…
…and take action!
22. Y I E L D B A S E D A U T O M AT I O N
Send emailFirmographic ScoreGet IP Address Returns Domain
COMPANY OBJECT
Hundred of high fit
companies per day
> 80% open rate
> 20% response rate
IF SCORE > 0.9
Prospection
23. Y I E L D B A S E D A U T O M AT I O N
Display LivechatFirmographic ScoreGet IP Address Returns Domain
COMPANY OBJECT IF SCORE > 0.9
We’ve helped companies like
Instacart, Crate and Barrel, and
IBM to integrate tools like
Mixpanel, GA, and GTM. Want
to schedule a demo of Segment?
24. Y I E L D B A S E D A U T O M AT I O N
Change
HomePage
Firmographic ScoreGet IP Address Returns Domain
COMPANY OBJECT IF SCORE > 0.9
25. Y I E L D B A S E D A U T O M AT I O N
Account, User and
Campaign Record
VISITED WEBSITE
SCORE PROSPECTING
COLLECTION ACTION
ACTION
ENRICHMENT & FIT SCORING
Chat
Email
Web content
Advertising
Physical goods
26. Seeing the success of those strategies, we went on a
path to discover other sources of intent.
Telling us who is expressing buying behavior
before they hit our site.
We found a TON of data
27. Y I E L D B A S E D A U T O M AT I O N
Web technographics: SDK install history
Alert on competitor install
- 1 month after, chance of POC failure ->
“how’s the POC going with x?”
- 10 month after, warn about contract
renewal
- Multiple competitor install? “Can we be
part of your RFP”?
28. Y I E L D B A S E D A U T O M AT I O N
Get all upvotes on a post.
Find contacts.
29. Y I E L D B A S E D A U T O M AT I O N
Capture IP traffic of visits on
G2Crowd pages. Find company.
30. Y I E L D B A S E D A U T O M AT I O N
Discover companies surging on specific topics
-> What are they reading about?
-> re-process all existing customers
31. Y I E L D B A S E D A U T O M AT I O N
Mobile App data: extract release schedule
time
App version
3.7
3.8
3.9
Predict next major release.
PM will have pressure to succeed
32. Y I E L D B A S E D A U T O M AT I O N
Installed
Web Tech
Reading About
“Livechat”
Hiring Sales
Director
TIME
Working backwards from the visit
Visited
Website
Upvoted
Post
Visited
Category Page
Outbound
Email
=> We detect intent before it’s expressed to us
33. Problem: the same user/company can pop up
on multiple signals in a short period of time
34. It’s a simple strategy: start with free channels, then increase pressure.
aka. send emails before you deliver a gift
Hitting the user like that works … but isn’t optimal.
It’s a mediocre experience for them and a loss of margin for us.
35. Y I E L D B A S E D A U T O M AT I O N
Visited
Website
Installed
Web Tech
Reading About
“Livechat”
Hiring Sales
Director
TIME
Demo
Request!
Capturing & measuring increasing intent…
Before it is expressed to us
SCORE: 0.12
Intent
SCORE: 0.33
Intent
SCORE: 0.61
Intent
SCORE: 0.88
Intent
36. Y I E L D B A S E D A U T O M AT I O N
WEB TECHNOGRAHICS
TOPIC SURGING
MOBILE TECHNOGRAHICS
CUSTOM SITE SCRAPING
Chat
Email
Web content
Advertising
Account, User and
Campaign Record
Physical goods
Output Channels
VISITED WEBSITE
SCORE PROSPECTING
COLLECTION ACTION
Intent capture
Enrich & Fit scoring
37. Cut the nice designs G, what does the martech stack look like?
38.
39. Y I E L D B A S E D A U T O M AT I O N
In 2017, We Observed Behavior.
Now, We Understand & Predict It.
40. Y I E L D B A S E D A U T O M AT I O N
time
1
0.2
User fit
User
Behavior
Conversion event
Intent to
Convert (p)
Visit my
website
Visit 3rd
party websites
Assisted discovery (nurture campaigns)
Register for
trial
Self discovery of
product
Discovery of
Market
Fit & Behavior modeling
h/t to @dariusmc
behavior enables us to guess intent.
Best case scenario
Belief is truth, and fit is our best
guess at modeling it
41. Y I E L D B A S E D A U T O M AT I O N
time
1
0.2
User belief
User
motivation
Intent to
Convert (p) Marketing campaign goal
Actual campaign effect
Marketing Campaigns = Usually Late to the Game
42. Y I E L D B A S E D A U T O M AT I O N
time
1
0.2
User belief
User
motivation
Intent to
Convert (p)
injection insufficient / late
“Come Back and Get 20% OFF”
=> Cost incurred but no revenue
injection too high
“Get a Free Tesla”
=> Cost > ARRPU = negative Margin
optimal time
of campaign (t)
43. Y I E L D B A S E D A U T O M AT I O N
fit prediction (f)
behavior
analysis (b)
optimal time to
inject motivation (t)
Fit & Behavior modeling
44. Y I E L D B A S E D A U T O M AT I O N
Required Momentum injection
to reach conversion tipping point (i)
45. Y I E L D B A S E D A U T O M AT I O N
time
1
0.2
User belief
User
motivation
Intent to
Convert (p)
Optimal Time of
Campaign (t)
Fit Prediction (f)
Behavior
Analysis (b)
Motivation
Injection (i)
Prospect reaches conversion tipping point
46. Y I E L D B A S E D A U T O M AT I O N
time
1
0.2
User belief
User
motivation
Intent to
Convert (p)
Early detection & appropriate injection
Maximizes margin
optimal time
of campaign (t)
47. REMEMBER: the game isn’t
hitting as hard as possible or as early as possible.
48. Y I E L D B A S E D A U T O M AT I O N
“I need to solve pain x
- Hey, have you seen feature y and z?”
Angle of
confusion
f+b
conversion
i
behavior
fit
Vector to conversion & Angle of confusion
time
Likelihood
to buy
49. Y I E L D B A S E D A U T O M AT I O N PROBLEM EXAMPLES VISION
2018 is the Year
to Automate MARGIN
M A R G I N B A S E D A U T O M AT I O N
50. The desired outcome isn’t only conversion or leads.
It’s max margin. Highest revenue at lowest cost.
Making money without money,
isn’t that what Growth Hacking really is?
51. Y I E L D B A S E D A U T O M AT I O N
Injection of momentum to tip over
conversion (cost to convert)
Capturing buying signals enables
prediction of fitf
Behavior analysis gives us optimum
timing (t)
Predicted revenue of user
b
i
1- (p+b) = i
=> momentum to inject, and at
which angle: Offer + message
(f+b)*r => $
=> value of prospect
r
if $ > i
=> Go ahead if margin is positive
Max efficiency is i @ t
=> when i is smallest
52. Y I E L D B A S E D A U T O M AT I O N
Visited
Website
Installed
Web Tech
Reading About
“Livechat”
Hiring Sales
Director
TIME
Demo
Request!
Predicting intent Value
VALUE: $130
Prediction
VALUE: $200
Prediction
VALUE: $350
Prediction
VALUE: $600
Prediction
VALUE: $1200
Prediction
53. Y I E L D B A S E D A U T O M AT I O N VISION
Visited
Website
Installed
Web Tech
Reading About
“Livechat”
Hiring Sales
Director
TIME
Demo
Request!
The Right Action at the Right Time
VALUE: $130
Prediction
VALUE: $200
Prediction
VALUE: $350
Prediction
VALUE: $600
Prediction
VALUE: $1200
Prediction
COST: $0.12
ADVERTISING:
Install {Tech} &
200+ Tools
COST: $0.01
EMAIL:
Discover the power of
great communication
COST: $20
Send branded gift
COST: $4
LIVECHAT:
Anything I can
help with?
Predicted next best
Action
54. Y I E L D B A S E D A U T O M AT I O N
Let’s start at the
beginning.
? We know there are
visitors poking around on
Drift’s G2 Crowd Page.
That’s intent.
55. Y I E L D B A S E D A U T O M AT I O N
An IP Address can
help us build a profile.
? 98.217.201.178
56. Y I E L D B A S E D A U T O M AT I O N
growthhackersofficial | likes: 23544
company/growthhackers-com
@GrowthHackers - The Official
Twitter Account of GrowthHackers.
Premiere destination to collaborate
and get inspired to help grow your
business. Producer of #GHConf18.
followers: 187267 | following:
56947
growthhackers.com
Helping teams unlock their company’s full
growth potential through a combination of
software, services and community. -
GrowthHackers
Marketing & Advertising, SAAS, B2B
alexaUsRank: 27794
alexaGlobalRank: 58492
employees: 19
Company Info Social Info Tech Stack
57. Y I E L D B A S E D A U T O M AT I O N
Clearbit tells us a lot
about a lead.
Madkudu tells us the
likelihood that lead
will convert.
?
58. Y I E L D B A S E D A U T O M AT I O N
Firmographic ScoreGet IP Address Returns Domain GrowthHackers
Company Object
59. Y I E L D B A S E D A U T O M AT I O N
Customer Fit Positive Signals
# of SDKs found on website
value: 15
web traffic volume
value: medium
twitter followers
value: 187276
Negative Signals
employee count
value: 19VERY GOOD
60. Y I E L D B A S E D A U T O M AT I O N
Firmographic ScoreGet IP Address Returns Domain GrowthHackers
Company Object
VERY GOOD
Finds persons
within company
61. Y I E L D B A S E D A U T O M AT I O N
Clearbit also helps us
build a profile around
a person
Sean Ellis
Founder, CEO
62. Y I E L D B A S E D A U T O M AT I O N
sean@growthhackers.com
Newport Beach, CA, US
Author of Hacking Growth. Coined the
term "growth hacking" after using it in
early days to ignite breakout growth for
Dropbox, Eventbrite, LogMeIn and Lookout.
@SeanEllis
followers: 122,789 - following: 40,680
in/seanellis
Sean Ellis
Founder, CEO
63. Y I E L D B A S E D A U T O M AT I O N
• Momentum to close: 70%
• Plan: Enterprise
• Revenue: $12,000 / year
• Close date: 45 days, 3hours
• Best next action: invite for lunch
• Cost to convert: $ 200
• SalesVelocity: $700
Predictions
64. Y I E L D B A S E D A U T O M AT I O N
Sean Ellis
Founder, CEO
Overview:
This person works like a general: Bold,
pragmatic, skeptical, extremely goal-driven,
and most comfortable with control.
Qualities:
Competitive,Assertive, Strategic, Self-reliant
Tips:
Meet them head-on and speak logically.
Move quickly, but be precise.
Challenge ideas and assert yourself.
Respect their time and speak logically.
Dc
75% confidence
d: 60
i: 0
s: 0
c: 40
65. Y I E L D B A S E D A U T O M AT I O N
Sean Ellis
Founder, CEO
When emailing:
This person often invokes a pragmatic,
objective perspective and can be extremely
skeptical of anyone that is too enthusiastic
or long-winded.
• Write 3 sentences or less
• Use data to prove a point
• State your purpose for the email in the
first sentence
Dc
75% confidence
d: 60
i: 0
s: 0
c: 40
66. Y I E L D B A S E D A U T O M AT I O N
Sean Ellis
Founder, CEO
When selling:
This person will most likely try to win an
argument, but you can persuade them more
easily by playing to their more logical
qualities.
Focus the conversation on how you can
help them use time and resources efficiently.
• Ask a tough question or issue a
challenge
• Expect heavy skepticism",
• Avoid too much hype",
• Don't be thrown off by a very blunt
comment
Dc
75% confidence
d: 60
i: 0
s: 0
c: 40
67. Y I E L D B A S E D A U T O M AT I O N
Firmographic ScoreGet IP Address Returns Domain GrowthHackers
Company Object
VERY GOOD
Finds persons
within company
Personalize your
Message
Send Sean a message tailored
specifically to him
68. Y I E L D B A S E D A U T O M AT I O N
Sean Ellis
Founder, CEO
69. Y I E L D B A S E D A U T O M AT I O N
2018 Roadmap:
- Margin Prediction
- Rule based outreach to build a training set
- Measure impact of (i) against dataset
Value based
predictions
On-site
optimization
Prediction based
outreach
Unexpressed
intent discovery
2017
Value based
outreach
AI Based
sequencing
Margin
prediction
2018 2019
70. Y I E L D B A S E D A U T O M AT I O N
Today we’re announcing revenue reporting
71. 5 year plan:
Build a system that not only computes the next best action,
but also it’s content.
Based on all aggregated signals, that content will be perfectly crafted
and vastly superior to what a human could write.
72. We reject automated emails,
because today their value is lower that of emails written by a human.
But when we reach a point AI fueled emails are more valuable than
handcrafted emails, how will that perception change?