Mar 2015

All marketing operation rests on the principles of STP (Segmentation, Targeting and Positioning). While the constant struggle for  the marketers is to identify the right segments to get the right message across to the target audience. Google has made online segmentation easier by adding in a Cohorts analysis to its free Google Analytics offering. This kind of analysis is extremely useful in determining the success of new ads or changes in messaging for web visitors. Retailers, Hotels, Banks, Telcos and Ecommerce companies stand to gain from this new metric.

While this is still in beta stage, some users can see it as a link under the Audience tab.

What is a Cohorts Analysis?

In statistical analysis, a cohort is a group of people who share some time-bound event, characteristic, or experience. From a Google Analytics/Web Analytics perspective these can be visitors to a site on a given date, first time visitors to the site.

The cohort analysis report can be configured around cohort type, cohort size, metric, and date range.

  • Cohort type. Currently Google only offers acquisition date under Cohort type.
  • Cohort size: The cohort size can be set to day, week, or month. Currently there is no option to manually define a Cohort size.
  • Metric: The metric indicates what currently needs to be measured. It can range from page views per users to goal completions and user transactions.
  • Date range. The relative date range displayed in sets of 7 days upto 30 days. If you select the month option, then you can analyse data upto 3 months. There is no custom range currently.

Running a Cohort Analysis on Google Analytics

Having covered the basics of setting up a Cohort Analysis, now let’s look at running a cohorts analysis and review the results on Google.

For our analysis, I am considering the data from one of our clients who does ecommerce transactions through their website.

The company runs several new promotions every week and they want to check if they should be changing prices/offers every week. For our analysis lets consider a hypothesis that the running new campaigns every week is not a good marketing strategy. We can use Cohorts analysis to check the validity of the hypothesis.

For this blog I have run a Cohort Analysis, analysing the cohort size by week, measuring the goal completions for the last 6 weeks. In this report Week 0 indicates the first interaction with the website as we only have a Cohort type of acquisition date right now.

This is the standard output from Google Analytics, the left column has the cohort size which was split up in two weeks in this case.

We are reviewing the data for a period of 6 weeks which is indicated in the top row.

You will notice that the bulk of the transactions have happened in the first week during the first interaction with the website. Week 1 to Week 6 shows the data for the next six weeks from the marketing campaign. You will notice that users have been interacting with the ad and transacting with the website even 6 weeks after the release of the advertisement. Wow! That’s good and shows the effectiveness of the message/creative.

Most of the transactions happen at the first contact, with about 1/5th of the transactions occuring during the first week and slows down as expected. That looks fine; maybe they can run an offer for 2 weeks.

However when we look at this analysis in conjunction with the pageviews/user, we realise that most of the transactions happen in the first interaction.

Now an user needs to visit atleast 4-5 pages to complete a transaction. Thus, we realise that most of the transactions do happen in the first week and for the company to maintain the sales momentum, it will need to run ads/marketing programs every week.

So, our current hypothesis stands nullified, the company does need to run marketing programs every week, to keep transactions going and meeting revenue numbers.

Hope this adds in an additional dimension to look at your marketing programs and how it impacts your bottomline.

Got questions, use the comments section or tweet us!