Bringing Programmatic
In-House
Jonathan Lau
Programmatic Lead
| 2
Summary
Who am I?
Why In-house?
Team Structure
In-house vs Managed
Programmatic Creative Innovations
Bid Models
What’s Next?
Who am I?
| 4
7-years in performance marketing
4-years leading teams
Father of two
Husband of one
Enjoys lifting heavy things
Quick Intro of Myself
| 5
Why
In-house?
| 6
We do not want to be chauffeured. We want to learn to
drive ourselves.
● Belief that Programmatic is an enormous growth
opportunity
● Lower marketing cost over time
● We want to diversify from black box, platform
partners
● In-house and managed partners are not mutually
exclusive
Investing in Our Future
| 7
Think and answer each of the following:
● What are your goals and how much time will it take
to achieve them (It’s going to take time)
● Monetary Investment (Initial sunk costs)
● Proper resource management
● Performance expectations (No immediate results)
● DON’T DO TOO MUCH!!!
Setting clear expectations
Team
Structure
| 9
Who We Started With
Dan Morris
(Mad Scientist)
● Cooks up new
ideas, creative
concepts
● Programmatic
strategist
Alexa Wieczorek
(Ops Ninja)
● Campaign
management
● Programmatic
strategist
Jon Lau
(The Dude)
● Team Lead
● Resource
scavenger
● Budget/Goal
Management
| 10
Our Current Team (I wish we had this when we started)
Dan Morris
(Mad Scientist)
● Cooks up new ideas,
creative concepts
● Programmatic strategist
Alexa Wieczorek
(Ops Ninja)
● Campaign management
● Programmatic strategist
Jon Lau
(The Dude)
● Team Lead
● Resource scavenger
● Overhead
Nick Olmanson
(Data Scientist)
● Focus on training bid
algo
● Build data pipeline for
our bid engine
Web Developer/Coder
● Build dynamic ad concepts
● Debug and deploy
iterations
Lauren Skarup
(Baby Ops Ninja)
● Second Ops person as we
scale our channel
| 11
In-house vs
Managed
| 12
Amazing Partner
Room for customization
Constant communication
and feedback loop
Provided crucial insights
when we first started
Scale as you grow
Working with a White-Label Bidder
| 13
Managed DSP CPM $4.76
In House DSP CPM $15.58
● Significantly lower CPM, but questionable inventory
● In-house efforts drove higher user quality and coming in with a lower CPA
Over 3x Lower CPM
In-house vs Managed DSP CPMs
Programmatic
Creative
Innovations
| 15
Dynamic Ads
Scratch
Card
320x50
Expandable
Drive higher
intent
Drive higher
quality
Create a
different user
ad experience
Multi-Sport
Live Lines
| 16
Funnel Optimizations
Current
Refreshed
Default
Prelim data
shows
+61% CTI
Conversion
*
Refreshed
Slots
| 17
Bid Models
| 18
What We Learned
Do not overcomplicate your approach when building
your bid model.
● Start with something simple and layer features
over time
● Training the bid model takes time and money to
achieve significant learning
● Capitalize on your learnings and iterate
accordingly (Even if it means reverting backwards)
● Does not replace a human touch
What’s Next
| 20
Our Wishlist for the Future
● Automation
● User LAL Modeling
● Dynamic Ad creation engine
● Retargeting
Questions?

Jon Lau

  • 1.
  • 2.
    | 2 Summary Who amI? Why In-house? Team Structure In-house vs Managed Programmatic Creative Innovations Bid Models What’s Next?
  • 3.
  • 4.
    | 4 7-years inperformance marketing 4-years leading teams Father of two Husband of one Enjoys lifting heavy things Quick Intro of Myself
  • 5.
  • 6.
    | 6 We donot want to be chauffeured. We want to learn to drive ourselves. ● Belief that Programmatic is an enormous growth opportunity ● Lower marketing cost over time ● We want to diversify from black box, platform partners ● In-house and managed partners are not mutually exclusive Investing in Our Future
  • 7.
    | 7 Think andanswer each of the following: ● What are your goals and how much time will it take to achieve them (It’s going to take time) ● Monetary Investment (Initial sunk costs) ● Proper resource management ● Performance expectations (No immediate results) ● DON’T DO TOO MUCH!!! Setting clear expectations
  • 8.
  • 9.
    | 9 Who WeStarted With Dan Morris (Mad Scientist) ● Cooks up new ideas, creative concepts ● Programmatic strategist Alexa Wieczorek (Ops Ninja) ● Campaign management ● Programmatic strategist Jon Lau (The Dude) ● Team Lead ● Resource scavenger ● Budget/Goal Management
  • 10.
    | 10 Our CurrentTeam (I wish we had this when we started) Dan Morris (Mad Scientist) ● Cooks up new ideas, creative concepts ● Programmatic strategist Alexa Wieczorek (Ops Ninja) ● Campaign management ● Programmatic strategist Jon Lau (The Dude) ● Team Lead ● Resource scavenger ● Overhead Nick Olmanson (Data Scientist) ● Focus on training bid algo ● Build data pipeline for our bid engine Web Developer/Coder ● Build dynamic ad concepts ● Debug and deploy iterations Lauren Skarup (Baby Ops Ninja) ● Second Ops person as we scale our channel
  • 11.
  • 12.
    | 12 Amazing Partner Roomfor customization Constant communication and feedback loop Provided crucial insights when we first started Scale as you grow Working with a White-Label Bidder
  • 13.
    | 13 Managed DSPCPM $4.76 In House DSP CPM $15.58 ● Significantly lower CPM, but questionable inventory ● In-house efforts drove higher user quality and coming in with a lower CPA Over 3x Lower CPM In-house vs Managed DSP CPMs
  • 14.
  • 15.
    | 15 Dynamic Ads Scratch Card 320x50 Expandable Drivehigher intent Drive higher quality Create a different user ad experience Multi-Sport Live Lines
  • 16.
    | 16 Funnel Optimizations Current Refreshed Default Prelimdata shows +61% CTI Conversion * Refreshed Slots
  • 17.
  • 18.
    | 18 What WeLearned Do not overcomplicate your approach when building your bid model. ● Start with something simple and layer features over time ● Training the bid model takes time and money to achieve significant learning ● Capitalize on your learnings and iterate accordingly (Even if it means reverting backwards) ● Does not replace a human touch
  • 19.
  • 20.
    | 20 Our Wishlistfor the Future ● Automation ● User LAL Modeling ● Dynamic Ad creation engine ● Retargeting
  • 21.