Menu

Fire Your Marketing Department (If They Don’t Use Cohort Analysis)

“A few years ago, my HR business partner said that she doesn’t fire people, they fire themselves. I couldn’t agree more!” — Marc Grabowski, COO of Nanigans.

Underperformance is a typical reason for employee dismissal, but with performance marketing it’s not as simple as looking at results and making a judgment call. Performance is based on the metrics a marketer has at hand. If they are looking at the wrong metrics, their success will be falsely represented.

Smart marketers are continuing to shift from CPA based bidding to optimizing on revenue against ad spend (True ROI). You can read more about this in our blog post on Price Elasticity: Why your CPA is Broke as a Joke. But the challenge for marketers who are focusing on lifetime return is that they are looking at the wrong time periods for revenue generation versus the ad spend responsible for that revenue.

For example, if an ecommerce advertiser spends $10,000 on advertising today and their website sales for the same day show $12,000 (assuming that all sales are margin positive), did today’s expenditure prove profitable? Should a brand be looking at conversions that took place that day as reason to adjust future investment? Should they target the same user profiles? Most marketers would say “yes”, this was a good day of spend versus revenue and will use it as a guide to increased spending tomorrow.

In a functional organization, this person gets fired and won’t have the opportunity to waste any more of their organization’s money. Harsh, but you get the point.

In this scenario, revenue captured on that day is reflective of ad spend and optimization practices from previous days. The revenue column aggregates customers who were captured on this specific day but it unfairly dips into the ad spend from customers who may have been acquired on previous days. Targeting principles that may have been implemented (and even removed) in the past will be represented in this revenue sample. This marketer is looking at the wrong information and it will cost their company money.

The worst part of this scenario is that the company would believe that the marketer did everything right and would attribute the loss in revenue to an anomaly. Even more concerning is that the marketer won’t really know which segments of customers (or profiles) really drive their business.

Talented marketers use the right tools to make the right decisions. First, a marketer should not inform future tactics based on “Time of Conversion” but rather “Time of Exposure or Click” (depending on the attribution model to which the marketer subscribes). Second, a cohort analysis report shows (in a customized timeframe) metrics such as marketing spend, the number of customers’ acquired/reached and the amount they spent on a daily, weekly or monthly basis. It is flexible enough to show the targeting and parameters that were responsible for driving these customers. Cohort analysis allows a marketer to replicate successful segmentation and remove unnecessary targeting that did not generate the desired result.

Another benefit of cohort analysis coupled with a predictive model is reporting on customer maturity. It reveals the how much a given cohort of customers is going to spend over time and answers the following:

  • Do these customers spend on the day they are reached by the advertising?
  • Do these customers come back over time?
  • When a customer continues to shop over time, will they buy for one week or continue to buy for months?
  • What elements create a good return customer versus an opportunistic one-time buyer?
  • When is the segment 50% mature and when will it stop producing returns?

All of these variables help marketers budget, target, measure and re-deploy ad spend. So as you assess the results of your campaigns, consider how cohort analysis can maximize the impact of every marketing action in order to find and acquire more customers.

A Step-by-Step Explanation of Cohorts

1) Time of Conversion

The chart below reports the number of conversions that were recorded on 11/30/2012 by this advertiser.

On 11/30/2012, This advertiser spent $21,982 on marketing.
This advertiser sold to 684 customers on 11/30/2012 and $152,556 was spent.
These customers and this revenue were not all attributable to the $21,982 spent on this day.

Cohort01

2) Time of Click

The chart below reports the number of conversions that were recorded based on dollars spent on 11/30/2012 by this advertiser.

On 11/30/2012, This advertiser spent $21,982 on marketing.
This advertiser acquired 607 customers attributed to spend on 11/30/2012 and $143,415 was spent.
These customers and this revenue were all attributable to the $21,982 spent on this day.

cohort22

3) Revenue by Month

Over the First 4 months (all time in this case), the $21,982 of spend from 11/30/2012 generated $143,415 and below is how this broken out by month. The purchasing users clicked on the ad on 11/30/2012 but then converted at some point over the next four months.

cohort3333

4) Revenue by Day

Over the First 14 days , the $21,982 of spend from 11/30/2012 generated $72,882 and below is how this broken out by day. The purchasing users clicked on the ad on 11/30/2012 but then converted at some point over the next 14 days.

cohort44

5) Daily Yield

This chart shows the daily yield generated from a week of cohorts allowing you to compare strong cohorts to weaker ones. Yield Revenue is calculated by dividing the revenue generated through a day by the spend used to purchase the media.

Example: On Day 1 of the Nov 30 Cohort, this advertiser’s revenue exceeded spend by 115.83%.

cohort5

Leave a reply

Your email address will not be published. Required fields are marked *

* Copy This Password *

* Type Or Paste Password Here *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>