Diagnosing a Common E-mail Malady: ‘Revenue Failure’

Your e-mail program is like a person: sometimes it’s robust and healthy, and at other times, it’s sick. Last year, my wife woke up with the right side of her face paralyzed. The ER quickly ruled out a stroke and then ruled out a brain tumor through a sequence of tests. Thus determining the cause was neither a stroke or brain tumor, but Bell’s Palsy.

Just like my wife’s situation, the best tests are those that allow you to rule out the most serious problems. So, to diagnose what’s ailing your e-mail program, begin by using a methodical approach to find out what’s not causing the problem.

Identify the initial symptoms (spam complaints, falling conversion, falling order value) and develop an initial set of hypotheses. Then apply a series of tests to rule out those that aren’t contributing causes.

A tool that’s useful for diagnosing revenue failure is a so-called “revenue tree,” the e-mail equivalent of the CAT scan (also available for Web site evaluations).

How the Revenue Tree Works

E-mail professionals have a wide range of data to review in a standard set of metrics (subscription/unsubscribe rates, open-click-conversion rates, average order value, etc). However, just looking at rows and columns of numbers in a spreadsheet doesn’t help you connect the dots in a way that the true picture of your situation becomes clear.

The revenue tree for e-mail lays out the metrics in a way that preserves the equation that relates the metrics.

The relationships are basic:

  1. Revenue = circulation (messages delivered) x RPE
  2. Circulation = number of subscribers x frequency of message sends
  3. RPE = average order value x Web site conversion rate x click rate
  4. Click-through rate = open rate x click-to-open rate

click to enlarge

This graphic lays out a sample e-mail analysis for a retailer tracking revenue from e-mail over time. The revenue tree places total revenue in the first bracket, then breaks it out into circulation (messages delivered) and revenue per e-mail (RPE), with bar graphs showing the data for each of four quarters, 2008 in blue and 2009 in red to easily compare year-over-year performance.

The revenue tree then breaks the circulation bracket into number of subscribers and message frequency, and RPE into average order value (AOV), Web site conversion rate, and e-mail click-through rate, again with bar graphs showing values over the time period.

Finally, it breaks the click-rate bracket into the open rate and the click-to-open rate.

Applying the Revenue Tree

In looking at this revenue tree, we can see that revenue started to slide in Q3 2009. The decline in Q3 was due to low conversion (sample hypotheses: landing page, Web site, or inconsistent offers). But in Q4, the decline was not due to conversion, but click-through, most of which was a click-to-open problem (sample hypotheses: irrelevant content, poor segmentation, non-competitive offers).

As a result of this analysis, we isolated factors for further review and correction. We haven’t yet diagnosed what caused the revenue to begin falling, but we know where to go to find the answers, and we can develop hypotheses that are relevant to the facts at hand. That way we can reduce the number of tests we run to find the answer.

It helps to filter these graphics by key variables to isolate issues. For example, if you start doing a lot of prospecting, you’d expect revenue to go up, but overall productivity (revenue per e-mail) to go down. So you’ll want to look at the revenue tree for customers and prospects separately. Similarly, you can look at various campaign types or other customer segments.

The Takeaway: Begin Analysis Before Symptoms Appear

You don’t have to wait until your e-mail program starts to falter to do a revenue tree analysis. It can be the start of a proactive wellness program, where you can build up strengths and identify weaknesses before they begin to affect revenue.

You’re the doctor. Study the data and prescribe the right remedies so your e-mail program can thrive.

For further reading on developing hypotheses and using more general “logic trees” to break apart problems into clearly connected but separate parts, see “The McKinsey Mind.”

Related reading