Path Analysis in Google Analytics 360

Product owners, product managers, designers, and developers invest countless hours improving their products.

Their goal? Create a flow that helps users perform a task with the minimal friction possible (e.g. checkout, quotes, lead generation, etc.).

In the real world, however, the paths users take may be far from ideal, and those deviations to the flow can be costly.

Read on to learn how to visualize the user path for your product/website/app.

A brief history on paths

During thousands of years, sail boats were used to transport goods and people. By the late 1800's there were well-established routes around the world oceans and navigation was somewhat sophisticated.

In 1852 Matthew Maury collected and analyzed hundreds of ships logbooks, looking for patterns of the most favorable winds and ocean currents.

The result of his work, "Wind and Current Charts", revolutionized how sailing ships traveled around the world, saving millions of dollars due to increased speeds by taking advantage of favorable winds and currents.

Step 1: Path Analysis Requirements

Identify the process to study

Your site/application is likely to serve multiple customer needs, if you look at the entire data, it will be confusing and overwhelming.

Instead, focus on a specific feature, process or section of the site that is focused on a particular function (e.g. checkout process, price simulator, quote generators).

Data Requirements

To recreate the paths users take in your website, you need raw event level data. In this example we will use both Google Analytics 360 BigQuery data and Snowplow Analytics data.

Step 2: Getting the Data for Path Analysis

Google Analytics 360 BigQuery

Open Google BigQuery and navigate to your Google Analytics 360 dataset, click on the ga_sessions_ table and then click on "Query Table" (top right).

Select one or two days that are representative of your website/app usage.

Edit the query in the WHERE statement to reflect the section to study and export the resulting data to a CSV.

  #legacySQL
  SELECT
  STRFTIME_UTC_USEC(SEC_TO_TIMESTAMP(visitStartTime+ hits.time/1000),"%Y-%m-%d %H:%M:%S") as hit.timestamp,
  visitId,
  fullVisitorId,
  hits.page.pagePath
  FROM [dataset_id.ga_sessions_YYYYMMDD]
  WHERE hits.page.pagePath LIKE '/<< SECTION YOU LIKE TO STUDY >>%'
  AND hits.type="PAGE"
  

Snowplow Analytics

Open your favourite PostgreSQL client and connect to your Snowplow database. Navigate to the atomic schema, execute the following query and export the data to a CSV.

    SELECT
    collector_tstamp,
    domain_sessionid,
    domain_userid,
    page_urlpath
    FROM atomic.events
    WHERE page_urlpath LIKE '/<< SECTION YOU LIKE TO STUDY >>%'
    AND event="page_view"
    

Step 3: Meet Disco, the process mining tool

Head over to Fluxicon website and download a trial version of Disco.

Open Disco and select Load your own data. (click image to zoom)

Map the Google Analytics 360 or Snowplow Analytics fields to Disco. You can also import additional fields for further segmentation.

Click start import. *Mandatory fields

In a few seconds you should see the most commons paths with the absolute frequency (number of events between paths) and case frequency (number sessions between paths) with mean durations between steps, in this case pages.

DISCO includes other functionalities like filters, statistics, case explorations and much more. I recommend downloading a trial version and exploring!.

Conclusion

Process mining has an array of applications, particularly in highly process-driven businesses, e.g.: call-centers, fast-food restaurants, factories, healthcare, manufacturing, car dealerships, etc.

Looking for a one time process audit? Contact us!.