Intro

This is an example of using googleAnalyticsR with RMarkdown to generate reports.

Authentication

Generating a token

To generate the token first time:

  1. Create your Google Cloud project as detailed on the website (see video guide on website)
  2. Set your environment arguments for your client_id and client_secret e.g. GA_CLIENT_ID="XXXXX.apps.googleusercontent.com" GA_CLIENT_SECRET="XXXXXX"
  3. Run ga_auth() in an R session.
  4. Move the generated auth file (usually .httr-oauth in your working directory) to a secure location - can also rename it to something more specific e.g. ~/auth/mywebsite.httr-oauth
  5. Specify the auth file location in GA_AUTH_FILE in the .Renviron file

By now you should have an .Renviron file with these entries:

GA_AUTH_FILE="/Users/mark/dev/auth/googleAnalyticsR.httr-oauth"
GA_CLIENT_ID="XXXXXX.apps.googleusercontent.com"
GA_CLIENT_SECRET="XXXXXX"

This file will be used by the RMarkdown (and your other googleAnalyticsR calls) when authenticating and calling the API.

Using the token

You can then use the following in your RMarkdown to authenticate each time you knit the document:

library(googleAnalyticsR)
accounts <- ga_account_list()
accounts[1,c("accountName","websiteUrl","webPropertyName")]
##   accountName               websiteUrl     webPropertyName
## 1     cloudyr http://cloudyr.github.io the cloudyr project

Warnings

If you see something like this:

## Warning: Token token$app$key != getOption('googleAuthR.client_id') 
## #>Token: XXXXX.apps.googleusercontent.com 
## #>Option: YYYYY.apps.googleusercontent.com

It means the auth token key was generated with different settings than what you have live.

Getting data

Fetching data can use R’s date functions to always be up to date:

# create date range from start of 2017 to yesterday
my_date_range <- c(as.Date("2017-01-01"), Sys.Date() - 1)

# replace with your GA view Id:
ga_id <- 81416156

# get your data
gadata <- google_analytics_4(ga_id, date_range = my_date_range, 
                             metrics = "sessions", dimensions = c("date", "medium"))
## 2017-11-03 13:27:45> Downloaded [949] rows from a total of [949].

Caching data

If you only want to fetch data when its not present (useful for big data calls), you can save the data to an .rds file so the RMarkdown will only fetch from the API when needed (e.g. when the date changes). If you want it to fetch a new batch, delete the .rds file.

# only get data if the data is not already here for this day
my_data_file <- paste0(Sys.Date(),"my_data.rds")

if(file.exists(my_data_file)){
  message("Using existing data file ", my_data_file)
  gadata <- readRDS(my_data_file)
} else {
  message("Calling API")
  gadata <- google_analytics_4(ga_id, date_range = my_date_range, 
                               metrics = "sessions", dimensions = c("date", "medium"))
  saveRDS(gadata, file = my_data_file)
}
## Using existing data file 2017-11-03my_data.rds

You may want to delete old cache files.

Multiple GA accounts

If you need to fetch from several different Google analytics accounts, the easiest way to handle authentication is to have the websites all available under one email.

If that is not possible, then you can run through the authentication steps twice, and name your auth tokens different names.

You will need to unset the GA_AUTH_FILE in your .Renviron if you have one by either commenting it out or removing the line.

You can then authenticate seperatly using googleAuthR::gar_auth()

library(googleAnalyticsR)
googleAuthR::gar_auth("~/auth/client_one.httr-oauth")

client_one <- google_analytics_4(ga_id_one, date_range = my_date_range, 
                                 metrics = "sessions", dimensions = c("date", "medium"))

googleAuthR::gar_auth("~/auth/client_two.httr-oauth")

client_two <- google_analytics_4(ga_id_two, date_range = my_date_range, 
                                 metrics = "sessions", dimensions = c("date", "medium"))