At my company, Aspen Marketing Services, we do a lot of work around direct marketing (direct mail, events, email, etc.). I’ve had a great opportunity to learn more about marketing in the offline world and have also been able to contribute to projects that have and online component.
Since it’s the end of the year, I thought I’d do a wrap up on some of my experiences and share my thoughts on how the gap between online and offline marketing can be closed (or at least tightened) with analytics.
Anytime you interact with someone offline and then try to drive them online, you need to know where (what campaign, event, referring source, etc.) they came from. A lot of marketers watch for a spike in site traffic and say, “Ahh…it’s because of the campaign.” Well maybe, but how do they go on to tell how conversion rates for the campaign were affected, what behavior is exhibited by people who come in through the campaign vs. another source, etc.? Not very well, I think.
Tie offline interactions to online actions.
In direct mail, we often use PURLs (personalized URLs), campaign codes, etc. to link a visit to a piece of offline marketing. Sometimes a vanity URL can be used, though there’s no guarantee the visitor will bother to type in anything after your domain name (www.example.com/campaign). Instead, I prefer to use completely new vanity URLs (www.mycampaign.com) or subdomains (campaign.example.com) since entering the code is unavoidable. I also like to use unique toll-free numbers for the calls related to the campaign, but that’s the topic of another post.
By knowing exactly where the traffic is coming from, you can set up segments to account for site and campaign level measurement as well as behavioral interaction analayis.
Capture conversion separately.
You can always roll-up conversion numbers but having them broken out separately for each campaign provides you with a number of benefits. Here’s a brief list of some important ones:
Cost per Conversion for the campaign
Campaign costs / # conversions
This tells you how much your campaign cost in terms of each conversion you got from it. So if you know that Campaign A cost you $25,000 and you got 100 conversions, you’d have a cost of $250 per conversion. This may also be expressed as CPS (cost per sale) or CPL (cost per lead). Knowing this information lets you compare campaigns on a cost/benefit basis and also have historical, easy-to-compare values when you’re thinking about your next campaign.
Value of Campaign Customers
Let’s say you launch a campaign that offers one month of free service before the customer has to pay anything. You may notice a strong adoption rate during the campaign, but what about when people churn (cancel their service with you)? You need to be able to tie the customer acquisition source back to the campaign to better understand future behavior. Leveraging the acquisition source data provides the opportunity to measure the value of the campaign customers separately from customers having come in from other sources. It also provides an additional piece of information for any predictive modeling that might be done to do things like determine channel preference or likelihood to accept an offer.
Determine Behavioral Patterns
Behavioral observation of the segment of customers responding to a campaign may provide future insights into how best to interact with those customers. If you don’t have these customers broken out seperately (I would argue that you should be looking at both response and conversion here), it will be very difficult to tell what makes these people tick.
Provide a cohesive marketing experience.
I hate it when I click a link or go to a site and enter a code and all I get is a product page or (worse) the homepage. The link should take them to an extension of the offline experience. To take that a step further, if you’re doing any sort of multivariate or segment-based testing in your offline creatives, you should have online matching creatives. We stress this with clients and have proven that a cohesive experience lowers bounce rates, increases conversion rates, and results in more interactions per visit — each an opportunity to learn more about the visitor.
Value more than the just the responses and conversions.
Often we focus on the bottom line. Did they buy or not? Are we increasing new subs or not? However, proper analysis provides much more value than that. For example, based on web analytics data, can conclusions be drawn on what visitors’ product preferences are (rhetorical question… yes.)? But what would a research study into product preference cost? Looking at it from that perspective just made implementing analytics a whole lot cheaper since you’re getting a whole lot more value. Not to mention that the data doesn’t lie. Not saying that the people do, but we’re dealing with observed behavior versus someone saying what they might do in a situation.
Collecting and analyzing this data offers up more value to the organization than simply revenue recognition. The knowledge you gain is valuable for future marketing efforts and, when looked at as a quantifiable component of your Cost per Conversion, further justifies an analytically driven campaign.
Test, Test, Test….
Whether a simple A/B test or a more advanced multivariate test with design of experimentstechniques, testing your offline targeting, segmentation, and creative design along with your online conversion process is essential. Don’t be afraid to mix things up and experiment. Not all tests will succeed but if done thoughtfully and correctly, you’ll learn what works and what doesn’t. Small lifts in incremental gain are important here and your tests should build on each other.
I hope those bits of advice are useful and would love to have others share their advice as well. They’ll all likely be expanded on into larger posts of their own, just not on a late Friday afternoon. :)
*The views expressed in this post are my own and do not necessiarly refelect those of my employer, spouse or anyone else.
**Since the views are my own, so is the content of this post and copyright is asserted. Linking permission is expressly allowed.
Comments on this entry are closed.