Excellent article on why it’s better to wait on Email Analytics packages.  I’ve done some work in this area with www.thehousedesigners.com email list, so I’m become aware of some of the challanges regarding intergrating email metrics with other information you may be collecting, or want to collect.

I look at email as one of your marketing channels, along with your website, your offline publications, direct markeing, phone calls, etc.  There are many ways to sell, and often it’s difficult to track sales and conversions generated from one channel into another channel, or even, to know the true origin of your customers.

The WhatCounts article does a good job of outlining issues with email tracking and how a solution that measures your email metrics may/may not help.

"Unfortunately, the concept of driving your email segmentation strategy based on clickstream analytics data is not the silver bullet that vendors make it out to be.  Don’t get me wrong…analytics-email integration tests have yeilded some amazing conversion rates and WhatCounts has entered into partnerships with three leading Web analytics providers.  But, as a person who has worked in both the analytics industry and the email industry, I can tell you that there are some things that should be considered that will likely rain on your integration parade, regardless of the vendors or platforms you select"

I find this information useful and so I’m going to summerize it here, as this is a Web Metrics blog.

• "We would like to send a campaign to people who viewed a particular type of product but didn’t place it in their shopping cart"
• "We are going to remarket to people who abandoned their cart."
• "We are going to target people who viewed our Flash demonstration or downloaded our PDF."
• "We are going to try to re-engage people who got a null result from our search box"
• "We would like to weigh the effectiveness of each loyalty marketing method by calculating the true ROI of each email-send in terms of revenue-per-subscriber."

All to this is beyond what I can do with custom tagged urls and LiveStats.NET (what The House Designers use); but I can do many things that are similar, and do show me visitor intent and visitor interest - and I can go right down the the customer level.  We can know what each customer who clicked on an email link looked at, how long they stayed, and to a certain extent, what plans they looked at. From there, I can generate a list of "hot" customers (those who spent more than a minute looking at plans from the newsletter AND/OR those who had more than 5 pageviews - that’s my metric).  We then send those customers a special email offering them (just the "hot ones" a discount if they act now).

"Cookie-based identification is hit-and-miss at best, as some people reject first-party cookies, some people reject third-party cookies, 40% of people clean out all of their cookies at least once per month, some people disable JavaScript, some people share computers (or email addresses) and some people use multiple computers or browsers during the course of their buying cycle".

I found this to be true.  I think about 40% of people actually do clean out all t heir cookies once a month and that explains the large number of "Direct Sales" we were getting when we used KeywordMax.  We tried hard to find out more about the Direct Sales visits, but it was hard to get much because so many people are now wiping out cookies on a regular basis.

"We all know that accurate personalization leads to higher conversion rates, right?  How effective do you think your assumptions about my browsing habits will be if you only capture half of my sessions?  Just because I logged a session duration of six minutes on the weather section of your online newspaper one day doesn’t mean I want a bunch of info about the weather all the time.  Maybe I got sidetracked by a phone call and I left my computer unattended during those six minutes.  I may spend at least thirty minutes reading the sports section and the entertainment section during an anonymous session the next day, and your analytics-email integration will be missing the mark.  If effective personalization is exponentially more effective than broadcast messaging, is incorrect personalization based on faulty assumptions exponentially less effective than broadcast messaging? "

Very interesting!   I am dealing with this very issue in some of my work. 

Personalization is very tricky and somewhat dissipointing in click through rates from my own and other’s findings.  What you need to track is the conversion rate of those who click on the personalized content vs. those that don’t.  The problem is the numbers of click throughs for personalized content is usually too low to get a meaningful comparison.  If you have 4 impressions and two click throughs, that’s a 50% click through rate.  If you compare that against 4000 unpersonalized impressions and 200 click throughs, you’d have a 5% click through rate.   Can the 50% for personalized content CTR really be compared with the 5% for non-personalized content?  I don’t think it’s a good idea to compare them as the numbers for the personalized content are far too low.

The WhatCounts article ends with the advice to double check the claims venders are giving when deciding to buy their email metrics products.

"Make sure to do your due-diligence when vendors tell you that "they are integrated."  Ask them what exactly is integrated…don’t just take it at face value.  Find out if they have "press release integration (we plan to do it)" or "case study integration (feel free to talk to our customers who use it)."  If you talk to those "integrated" customers, find out if there truly was a delta in their conversion rates as a result of the work.  What was the opportunity cost of doing the project?  Did they put less emphasis on other segmentation strategies that could have been equally effective? "

I think it’s better to wait another year or so, before tying your email analytics with the rest of your analytics.  In my case, I’ve gone the low tech way and just used the web analytics and some custom tagging of urls in email newsletters to do the trick.  Sure, it’s not as sophisticated as many of the packages out there…but it does what it needs to do for my client and that’s what counts.