"r" for ROI

12
“r” for ROI (for Google adWords campaigns) Andrea Dodet - MSc International Business Copenhagen Business School

description

Showing clearly the effect of a Google AdWords campaign is one of the most important task involved in a digital consultancy. Here's a quick method to impress your clients with outstandingly clear content. Just a little "r" scripting required (definitely not rocket science). Enjoy

Transcript of "r" for ROI

Page 1: "r" for ROI

“r” for ROI (for Google adWords campaigns)

Andrea Dodet - MSc International Business Copenhagen Business School

Page 2: "r" for ROI

HOW CAN WE ASSESS THE

IMPACTWITHOUT KNOWING THEIREFFECT

?

OF GOOGLE ADWORDS CAMPAIGNS

Page 3: "r" for ROI
Page 4: "r" for ROI

The Language

R is a free software programming language and software environment for statistical computing and graphics.

Page 5: "r" for ROI

`

The Package

The model predicts how the response metric would have evolved after the campaign as if the

campaign had never been performed.Simple, uh?

install.packages("devtools") library(devtools) devtools::install_github("google/CausalImpact") library(CausalImpact)

Page 6: "r" for ROI

Easy, quick, intuitive

# Load data data <- read.csv (“data.csv”) !# Specify pre-post intervention periods pre.period <- c(1,70) post.period <- c(71,100) !# Causal impact analysis library(CausalImpact)

impact <- (data, pre.period, post.period) !# Show results summary(impact) plot(impact)

!

Page 7: "r" for ROI

The Output

Page 8: "r" for ROI

Observed and predicted response to the original scale.

Black line is the data affected by the campaign. Blue area is the situation in which nothing happened.

In this case the line is well above the blue area =

The campaign has produced a positive effect on the variable

Page 9: "r" for ROI

The difference between the two:

Causal effect produced by the campaign each point in time.

Page 10: "r" for ROI

Individual causal effects added up in time (cumulative)

Page 11: "r" for ROI

…RECAP

• R installed (R Studio for comfortably interface) • CausalImpact package !A dataset formed by: !• Entries with clicks (variables) in non-affected markets or

clicks on other sites etc. • Entries after the launch of a given adWords campaign.

What you need