groupon_controlled experimentation_panel_the hive

17
Rajesh Parekh Controlled Experimentation to Guide Product Innovation

Upload: the-hive

Post on 12-Jul-2015

231 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Groupon_Controlled Experimentation_Panel_The Hive

!Rajesh Parekh!

Controlled Experimentation to Guide Product Innovation!

Page 2: Groupon_Controlled Experimentation_Panel_The Hive

Controlled Experimentation (A/B Testing)!•  Method to study effects of a treatment#

•  Concept:!- Randomly split users into two groups#

� A : Control#

� B: Treatment#- A and B are identical to each other except

for the treatment being evaluated#- Collect performance metrics from the

experiment#- Run statistical tests to determine if

differences between A and B are purely by chance#

2

Randomly)Divide)

A)(Control)) B)(Treatment))

Measure)&)Evaluate)

Controlled)Experimenta=on)Panel)

Page 3: Groupon_Controlled Experimentation_Panel_The Hive

Why Run Controlled Experiments?!•  Commonly used approach in clinical trials!- What is the effect of a particular drug / treatment?#

•  Systematically validate hypotheses with data!!•  Concurrently run the treatment and control!- The difference (if any) is#� Because of the treatment OR#� Due to random chance#

•  Determine if a treatment is causal in nature!- E.g., Making the search box bigger causes increase in queries / user#

3 Controlled)Experimenta=on)Panel)

Page 4: Groupon_Controlled Experimentation_Panel_The Hive

Controlled Experimentation: Use Cases! #

4 Controlled)Experimenta=on)Panel)

A"B"Stract"Widget"Company"

_________________)_________________)

_________________)_________________) BUY)NOW)

A"B"Stract"Widget"Company"

_________________)_________________)

_________________)_________________) BUY"NOW"

Website)Variants)

Page 5: Groupon_Controlled Experimentation_Panel_The Hive

Controlled Experimentation: Use Cases! #

5 Controlled)Experimenta=on)Panel)

Free)Trial) Play)Now)

Mobile)Call)to)Ac=on)

Page 6: Groupon_Controlled Experimentation_Panel_The Hive

Controlled Experimentation: Use Cases! #

6 Controlled)Experimenta=on)Panel)

Top)deal)highlighted)

Email)Template)Design)

Page 7: Groupon_Controlled Experimentation_Panel_The Hive

Controlled Experimentation: Use Cases! #

7 Controlled)Experimenta=on)Panel)

Backend)changes)(e.g.,)Personaliza=on)Algorithm))

Page 8: Groupon_Controlled Experimentation_Panel_The Hive

8

Controlled Experimentation: Use Cases!

# •  Follow-up message for users who previously clicked on an ad#

•  Incentive campaign to re-engage lapsed users#

•  Think of this as placing filters / guards on a randomly chosen user population#

Controlled)Experimenta=on)Panel)

Custom)Defined)User)Segments)

Page 9: Groupon_Controlled Experimentation_Panel_The Hive

9

Key Components of an Experimentation Platform!Hashing function!!

!

!

!

!

Logging!!

!

!

!

Metrics – suite of KPI!!

!

!

!

!

Dashboard!

F())))))))))))))

Group)0)

Group)1)

Group)NU1)

Time"Spent"

Revenue"

Click>Through"Rate"

Session"Length"

Abandonment"

Purchase"Rate"

•  Metric)improvements)and)Sta=s=cal)Significance)in)a)central)place)

•  Detailed)logging)of)all)user)interac=ons)

Controlled)Experimenta=on)Panel)

Page 10: Groupon_Controlled Experimentation_Panel_The Hive

Ensure Identical Control and Treatment!

•  Custom Segments#

10

CONTROL" TREAMENT"

Gender""

Male)

Female)

CONROL" TREATMENT"

Region"Size"

Small)

Medium)

Large)

CONROL" TREATMENT"

Prior"Exposure"

No)

Yes)

δ%"

Controlled)Experimenta=on)Panel)

•  Frequency Distribution#

•  Large Difference in Prior Exposure Rate violates assumptions#

Page 11: Groupon_Controlled Experimentation_Panel_The Hive

A/A Tests!•  Run an experiment with two identical variants#

•  Helps to determine if:#- Users are being split uniformly at random#- Correct data is being logged#- Variance between identical populations of users is acceptable#

•  Challenge:!- Few purchases of high value deals render statistically significant

difference between treatment and control#

11

SPAIN"TRIP"$1,999"

Controlled)Experimenta=on)Panel)

Page 12: Groupon_Controlled Experimentation_Panel_The Hive

Monitor Each Variant!•  Place yourself in each variant

to validate the experience#!

!

12

Carefully)inspect)each)variant)

Controlled)Experimenta=on)Panel)

•  Wrong sort order!!

!

Page 13: Groupon_Controlled Experimentation_Panel_The Hive

Objective Function!#

#

Conversion" Revenue"

P(conversion)))

•  Favors)lower)price)deals)

E(rev))=)P(conversion))*)price))

•  More)expensive)deals)can)dominate)

Need)to)balance)mul=ple,)oZen)conflic=ng)objec=ves))

13 Controlled)Experimenta=on)Panel)

Page 14: Groupon_Controlled Experimentation_Panel_The Hive

Measure Overall Impact!•  Test focuses on#- A particular area of

the website#- A sub-population of

users#

•  Measure!- Improvement on the

sub-segment AND#- Entire population!#

14

Measure)overall)impact)to)guard)against)cannibaliza=on)

Controlled)Experimenta=on)Panel)

Page 15: Groupon_Controlled Experimentation_Panel_The Hive

Panel Discussion: Questions!

15 Controlled)Experimenta=on)Panel)

Page 16: Groupon_Controlled Experimentation_Panel_The Hive

Thanks to many talented individuals at Groupon I am privileged to work with!#•  Data Science#•  Engineering#•  Marketing / Market Research#

Acknowledgements#

16 Controlled)Experimenta=on)Panel)

Page 17: Groupon_Controlled Experimentation_Panel_The Hive

Rajesh [email protected]!!

17 Controlled)Experimenta=on)Panel)