Response Modeling

Marketers for years have tried variances of mapping response to timing and trigger messaging.  It absolutely makes sense in some cases. Yet if it’s so successful, why isn’t response modeling and behavioral targeting through email more pervasive in the space?  

I believe that if the industry operated on a pay-per-performance model we’d see this type of practice as a standard.  For many, it’s hard to stray from the “grow your list” and “send to it as many as you can“ mindset.   When you hear stories about JC Penney’s CEO opting into his own company’s list and getting over 300 emails the first month, you wonder how many other brands have this fragmented, high-cadence communication regimen.  I know by simple trends in our business that this is happening across the board. 

The volume of email sent in our space continues to double every two to three years, and yet we’ve seen a decline in response since 2009.  There will definitely be much more emphasis on real-time triggers and decisioning.   Here are a few that are top-of-mind:

First-party site behavior:   What happens on your site, stays on your site, or at least that should be your goal.  If you are to target site behavior, you should pick your highest value “conversion” or “shopping” paths and leverage retargeting to enable this behavior, balanced with page views, search source, even keyword trends.    I warn you, this is one of the hardest things to put into production with small staffs and a bunch of point solutions.  It requires specific guidelines and you must have very strict cadence and governance, not to mention that you need production people x3.

Social behavior:  It’s amazing how far along this tactic has come. In the past, not many had actionable plans that could prove any near-term impact.  Marketers and agencies are getting more creative though leveraging Facebook’s open graph to integrate apps to email for a better connected experience across brand sites, first-party sites and email. I still believe success depends on an episodic view of the fan, customer and engagement.  This becomes more of a challenge to model, but if you put in the effort, you can model your best and top segments to social behavior, even overlay if necessary and project on your customer base some degree of influencer, connector or advocate status.   Unfortunately, we have found that much like promotions, social network behavior is quite transient.  Projecting a fan or some loyalty factor from an activity over 90 days is like “grasping from straws” to answer a business question.  

Email response models: Can you predict future behavior from longitudinal modeling through clicks & opens?   I believe you can, but with a realistic view of what you’re actually forecasting. For commerce driven goals, response modeling will help answer near-term “in-market” business questions, and will need to be revisited several times a year.  For broader brand engagements where there is multichannel engagement, it becomes a reach factor, an index to govern segments cross channel.  Response modeling should be broken down and managed monthly to understand “fatigue” factors by segment, by offers that run on the network.  And it comes down to essentially a frequency/category based model that translates to monetization.  

The net effect should yield a positive business result and some improvement over yesterday.   Models are just that: a point-in-time view.  Don’t think you can do this once and run off it for years.  This is a dedicated discipline that I feel is going to change our rules.  Some platforms support rudimentary models, and they can be valuable, but each business is so different, this has to come from within.  

If you don’t have the resources to do this, I’d find them.  All trends in the industry are shaping towards a fragmentation of email.  We have to be smarter, more coordinated and better predictors of the consumer experience. Intelligence over Volume should be 2012’s banner.

 

Tags: email
Recommend (6)
4 comments about "Response Modeling".
  1. Tom O'leary from GroupMail , January 30, 2012 at 12:47 p.m.
    Great food for thought (and possible prescience) David. Having recently started managing our AdWords campaigns for GroupMail, this is in line with my recent thoughts about restructuring those high-impression, low conversion (or at least low value conversion) campaigns. The energy and resources spent on those high frequency, low return efforts are more productive when we sacrifice the comfort of reach and visibility for attention to those more meaningful and productive behavioral engagements. I understand firsthand the fear of abandoning the comfort of wide presence, high impressions and high clicks -- but if we tether ourselves to that model, we will undoubtedly be in synch with the slow decline of email ROI, which is set to drop below $40. Our [email] evolution requires adaptation to the social behaviors and interactions of modern online relationships and our ability to respond to them in real and meaningful ways. Have a great week! Tom -- With best regards from the GroupMail Team Tom O'Leary (marketing) ------------------------------------------------- GroupMail Newsletter Software http://www.group-mail.com Celebrating 15 Years!
  2. Nova F Misseyer from None , January 30, 2012 at 1:13 p.m.
    This absolutely works..
  3. Arthur Einstein from Loyalty Builders , February 3, 2012 at 1:56 p.m.
    All the right stuff. but really complicated. there has to be a simpler way.
  4. Mark Klein from Loyalty Builders LLC , February 3, 2012 at 2:52 p.m.
    Response modeling is difficult and not even the best way to do targeting. I outline an alternative in my recent blog post titled "Why response models are the wrong way to target" on the Loyalty Builders website http://loyaltybuilders.com/blog Here's why: First, they only use the treated customers to build the model, ignoring valuable information from the rest of the population. Second, and more important, they only represent their slice of the customer world at a moment in time. Customers are always in motion, and today’s model doesn’t match tomorrow’s customers.