Do you know what’s going on inside your online marketing campaigns? Are you spending precious budget and resources wisely?

In this article, I examine in detail the current web marketing efforts of Polar Inc. – a fictitious company — and recommend methods to optimize the effectiveness of its online campaigns. Let’s begin by examining Polar Inc.’s online banner advertising.

PPC/Banner Advertising

Between April 2005 and June 2006, Polar Inc. spent a total of $215,718 on Internet-based advertising. These efforts resulted in a total of 10,599 conversions (an average conversion rate was 3%). However, if we drill down into the data, we can see areas where our advertising campaigns can be optimized.

A total of $80,800 was spent on PPC advertising campaign which yielded a total of 6,551 conversions. By comparison, a total of $134,918 was spent on banner advertising which yielded only 4,048 conversions. Thus, banner advertising cost 67% more but yielded 38% fewer conversions.

Metrics
Several metrics are used to calculate the performance of the PPC and banner campaigns. The primary metrics and their formulas are:

• Cost Per Click (CPC) This is an assumption set at 0.37 cents. It is based on industry research from Efficient Frontier. See http://searchenginewatch.com/3629032
• Total Clicks: This is a calculation expressed as PPC Ad Spend / PPC CPC. For example, for April 2005, the formula would have been $650/0.37 = 1,757
• Conversion Rate This is an assumption set at 3%. It is based on industry information from ClickZ at http://www.clickz.com/3628276.
• Conversions This is a calculation expressed as Clicks x Conversion Rate. For example, for April 2005, the formula would have been 1757 * 0.03=52.7
• Cost Per Conversion This is a calculation expressed as Ad Spend / Conversions. For example, for April 2005, the formula would have been $650/53=$12
• Banner CPM This is an assumption set at $20.
• Click-Through- Rate (CTR)
• PPC Impressions This information is gathered from our PPC provider, Google AdWords.
• Total Ad Spend Per Subscriber This is a calculation expressed as Total Ad Spend / Subscribers. For example, for June 2005, the formula would have been $11,364 / 10,000 = $1.14

Compiling this information requires interdepartmental communication and collaboration both within Polar Inc. and with external data sources. Those sources include Polar Inc. IT department for set up the ad landing page signup form, which is considered the point of conversion for both PPC and Banner campaigns. Google AdWords provides data related to CPC and Total Clicks, CTR and Impressions.

Recommendation
Reduce banner ad spending and reallocate a portion of the budget to PPC campaigns.

E-mail Marketing – Recency

During 2005 and 2006, Polar Inc. sent a total of 37,589 reminder e-mails to subscribers. These are sent out monthly in an effort to get users re-engaged with using the website. User’s accounts are monitored for usage patterns and when a set “recency trigger” is reached — an assumption set at 60 days — an e-mail is sent encouraging the user to start using the website again.

We know through industry research that the more recent an event has occurred, the most likely that it will occur again. The measure of time passed since the last event is called Recency. For Polar Inc., this is assumed to be 60 days. For example, if a user has not logged into the website for at least 60 days, he/she is sent an e-mail.

However, according to the chart above, we can see that over time the number of e-mails being opened steadily falling. For example, the number of e-mails opened in November 2005 was 25% higher than the previous month — but that’s significantly less than the 78% increase in e-mails opened in September 2005 compared to the previous month of August 2005.

Several metrics are used to analyze the reminder e-mail metrics. They include:

• Reminder E-mail Recency Trigger (days)
  This is an assumption set at 60 days.
• Reminder E-mail Open Rate
  This is an assumption based on average open rates of e-mails sent by a large Toronto-based financial company. It is set at 30%.
• Reminder E-mails Opened
  This is a calculation expressed as Total E-mails Sent x Reminder E-mail Open Rate. For example, for March 2006, the formula would have been 3,987 * 0.30 = 1,196
• Reminder E-mail CTR
  This is an assumption based on average open rates of e-mails sent by DundeeWealth/Dynamic Funds. It is set at 20%.
• Reminder E-mails Cost Per Visit
  This is a calculation based on an assumption. The assumption is that it costs a flat-rate of $100 to sent out an e-mail campaign. The calculation is expressed as Cost / Reminder E-mails Clicked. For example, for August 2005, the formula would have been 100 / 60 = $1.66.

Compiling this information requires interdepartmental communication and collaboration both within Polar Inc. and with external data sources. Those sources include Campaign Monitor (www.campaignmonitor.com) for measuring total e-mails delivered, open rates, and CTR. Google Analytics is used for measuring hits to ad landing pages that are linked from within the e-mail campaign. Coordination with IT departments is required to set and execute the Recency Trigger action of 60 days.

Recommendation
Set and monitor the recency trigger (currently at 60 days).

E-mail Marketing – Latency

Polar Inc. does not currently have a latency set, and thus it’s difficult to create a LifeCycle for the website. Latency is the average time passed between events. Latency is a useful measure because industry research shows that a user is more likely to engage in bevaviour that is recent.

By setting a latency trigger based on a LifeCycle, we will know when the most opportunistic time is to promote our insurance products. This is based on the premise that a user is more likely to convert when they have recently engaged in some action.

Metrics
Several new metrics can measure recency and latency. They include:

• Frequency of visit
• Days since first Visit
• Average Recency
• Average Latency

Compiling this information will require mining Google Analytics so that we can set the average recency and latency cycles for our website. After this is established, we can set triggers that will sent out the e-mail campaigns.

Recommendation
Set a latency trigger for the item last uploaded and e-mail visitors who trip this in order to promote insurance.

Holiday Campaigns (December)

According to the analysis, Polar Inc. experienced a substantial decline in website use at the end of 2005 leading up to the holidays. This decline in usage corresponded with a decline in ad spending during the same time period.

As we can see, total ad spending was decreased by -70% between November and December 2005. As a result, there were -72% fewer clicks on our PPC campaigns and -68% fewer clicks on our banner ad campaigns. Similarly, PPC and banner ad conversions were down -72% and -68%, respectively.

This sudden reduction in ad spending has impacted website usage. During the same time period, items added to the system dropped by -60% and items per subscriber correspondingly dropped by -62%.

Metrics
Several metrics are used to calculate the performance of the PPC and banner campaigns. The primary metrics and their formulas are:

• Page Views Per Subscriber
  This is a calculation expressed as Pages Views / Subscribers. For example, for May 2005, the formula would have been 100,000 / 5000 = 20
• Items Per Subscriber
  This is a calculation expressed as Items in System / Subscribers. For example, for June 2005, the formula would have been 1,760,892 / 10,000 = 176
• Total Ad Spend Per Subscriber
  This is a calculation expressed as Total Ad Spend / Subscribers. For example, for June 2005, the formula would have been $11,364 / 10,000 = $1.14

Compiling this information will require mining Google Analytics so that we can set the average page views per subscriber (Google will provide this automatically, or we can calculate manually using the formula above).

Recommendation
Boost ad spending, not reduce it, leading up to the holiday season. It’s clear that a -70% drop in ad spending is correlated to the -60% drop in items added to the system for the same time period.

Conclusion

Based on the analysis we’ve done above, the following recommendations make the most sense for this client:

• Reduce banner ad use
• Increase PPC ad budget to capitalize on effectiveness
• Set recency trigger for e-mail to inactive users
• Set latency trigger for last item added/uploaded and e-mail users to promote insurance products (more likely to convert)
• Boost ad spending over the December holiday season