Last week's series on Identifying and Solving Client Pains was concluded with Eric Peterson. I'd like to continue with that same topic this week with Akin Arikan.

Akin Arikan is Sr. Segment Manager (Internet Marketing) at Unica Corporation.

Here are Akin's thoughts...

It always drives me crazy when traditional marketers question the accuracy of web metrics while pouring a gazillion marketing dollars into TV, print, and outdoor advertisements where they use panels or focus groups to calculate ROI. Ahmm, ... doesn't that seem a bit schizophrenic? But one area where traditional marketers do catch their Internet colleagues red handed is the measurement of the effects between the online and offline. Or more precisely, the typical ignorance thereof.

You are not alone if it seems ironic to you that we use complex algorithms to optimize how many cents exactly we will spend on online advertisements or search keywords, yet we use the online results as our sole input. Meanwhile, studies conducted by Comscore and earlier by Dieringer Research Group show that (depending on the industry) 60% or more of the purchases resulting from Internet activity may actually happen offline rather than online. Trying to optimize your online investments without taking the offline results into account is a bit like trying to sail across the ocean orienting yourself by the position of the moon instead of using your compass or GPS.

That's nice Akin. So how do we go about it then, wise guy? I separate techniques into two groups, namely 1.) those that aim to measure effects on a 1:1 basis, i.e. for each individual customer, and 2.) those that don't go down to the 1:1 level but measure aggregate effects. I recommend you start with 2.) first before you go down to level 1.) even though you you will find the biggest rewards there.

For ROI assessment at the aggregate level there are multiple techniques that you will have heard about at conferences such as SES. For example you can inject unique 800 telephone numbers into your web site when visitors click through from particular advertisements. Anyone who calls can be attributed back to that advertisement. But what if you are buying many paid keywords, would you need an 800 number for each? No, there are companies that offer clever tricks for making it unnecessary to have a different number for each keyword. But how about store sales? You can offer online coupons that people will apply offline. You can offer "buy online, pick up in store". You can offer online product configurations that people can print and bring to the store if they represent a sufficiently high value to them. If nothing else works, you could experiment by shutting down your online advertising in a particular region to measure how offline activity will go down with it.

The proposed techniques help with estimating the effects of online to offline, but are not precise enough for 1:1 marketing. For example, if you wanted to e-mail a discount promotion to someone who browsed but did not complete a hotel reservation online, then wouldn't it be vital to first make sure that this person has not completed a reservation by phone for full price yet? Here you can use techniques that have been applied by your colleagues in direct marketing for many years. Namely, as long as you can identify the customer on each channel you can reconcile their interactions on the back-end. For example, take a hotel web site. You cannot know who an anonymous visitor to your hotel web site is. But if I go to Extended Stay Dot Com for the umpteenth time to make a new reservation, more likely than not, they know exactly who has been browsing because I have completed reservations on their site recently and my cookie allows them to identify me. Likewise, when I call and make that reservation via the phone I identify myself with my name, even my credit card. So they could put 1 and 1 together to know that I just browsed the web site but should not be sent a discount offer because I have meanwhile completed a reservation by phone. This kind of cross-channel response attribution has been common practices in direct marketing for many years. If you want to spare yourself having to reinvent the technology from scratch take a look at software suites such as Unica's Affinium suite for Internet Marketing (http://www.unica.com/) that already comprises many multi-channel capabilities in this regard.