Multichannel Attribution

I love this graphic from Joel Book, ExactTarget's Education Chief because it illustrates how the topic of multi-channel attribution often seems to me like the search for the Holy Grail (the Last Cup of Christ).  The result can be elusive. Given the proliferation of channels, the complexity of the issue, cross-firm differences in resource allocation and other aspects of implementation, is there likely to be a broad-based answer to the questions of how to 1) attribute marketing results to marketing actions and 2) develop marketing programs that maximize spending? 

The four ways that I see academics are addressing and can further address these issues are as follows:

• By investigating an industry:  In the Journal of Direct, Data and Digital Marketing Practice, Schultz, Block and Ramn have studied consumer media usage in three product categories, computers, automobiles and fast-food restaurants. (Schultz, et. al. 2009). In these studies as in other marketing studies elsewhere, we often see that where the marketer is spending their dollars (mass advertising) and where the customer is spending their time (social media) do not match.
•  By examining a company:   Recently in  the Journal of Marketing (May 2012), Avery, Steenburgh, Deighton and Caravella investigated data from a high-end retailer to see if introducing a retail store in specific areas would affect catalogue and online sales.
• By analyzing the customer:  Venkatasen and Kumar (2005) and Venkatasen, Kumar and Ravishanker (2007)  have developed a number of models from the point of view of the customer to explain and predict the adoption of multiple channels and discuss financial implications. 
• By analyzing multiple firms (strategy):  Neil Morgan in a recent issue of JAMS (2012) outlined a conceptual framework to link all the different elements of marketing strategy and business performance, including implementation, which is often overlooked in research but critical in practice.

Although these efforts are great first steps toward answering marketing attribution questions, none of them really provide guidance as to how to attribute the results of marketing spending on brand advertising, social media, search etc. to performance.  The type of research conducted to date is a good start, but usually does not focus on the complexities facing the marketer today in a generalizable way and certainly does not look at cross-firm differences and performance.  So we as academics have an opportunity not only to apply rigor to these issues but to work on topics of real relevance to marketers today.

And the question of how to measure performance in this context is interesting also.  Should we look at ROI, or event CLV, as practitioners do, in hard numbers way?  Wouldn’t customer equity ala Rust and Lemon and Zeithaml (2004) be more appropriate or at least a metric that should be evaluated in conjunction with other measures (Stahl, Heitman, Lehmann and Neslin 2012).

Multichannel Marketing Class June 11: Final Case and Wrap Up

We had a great last class with several guest evalulators from industry and Dr. Geoff Gordon from NIU. Students presented the final case which was to put together a multichannel media strategy for an insurance company. The additional challenge was to work through the agent network. The students did a fabulous job. They recognized the importance of the database as a central respository of information across channels. Several groups suggested interesting segmentation strategies. I was impressed with their creative suggestions for using social media, although with the age of the insurance agents it might be difficult to get them on board. Also, Rob Jackson pointed out with all the regulation in the insurance industry and privacy regulation and concerns it might be difficult for agents to really participate in such an open media. We learned in four weeks about all the major media channels, direct mail, email, search, internet, etc., and about databases, analytics and measurement related to multichannel marketing. I think media channels will become more and more fragmented and marketers will face more challenges in the future. Thanks again to NIU's Marketing 660 CE1 MBA Class, Special Topic: Multichannel Marketing, for a great four weeks!

Multichannel Marketing Class June 10: Measuring across channels and PPC

We talked about a number of different ways to measure responses across channels. I likened the search for the ability to measure across channels and attribute response to specific media actions to the search for the Holy Grail. There are so many new media and it is difficult to know if the response came from the last sales contact, banner ad, or search stream. Approaches vary from surveying consumers to field experiments (treating one group differently from the control to see if there is a change in sales or response rates) to building sophisticated models to simulate and measure response. I shared some research from Don Schultz at Northwestern where he has used survey data from an online panel to then determine media influence. In the example Don gave us, Auto Makers were spending too much money on television advertising and not enough on the internet, in terms of perceived influence. Industry data is useful as a starting point but many marketers want to see what the effect is on response in their particular company situation. The search will continue for the best models.

We also looked at Pay Per Click in Google and saw how expensive some keywords can be and discussed bidding strategies. It can be a good strategy to go for the third or fourth spot to save money and still be 'above the fold.'

Multichannel Marketing Class June 8: Trends in Multichannel and SEO

Bob Arkema from Johnson & Quin led us in a lively discussion of how the direct and interactive industry is changing in a post financial-meltdown world. For the first time since data has been kept, direct mail spending decreased last year. Search and social media spending are predicted to grow but most other media are flat or declining. Companies are taking a good, hard look at what they mail and how much they mail. Mail is seen increasingly as a companion to other channels, such as online or email, which is consistent with the multichannel theme of our class. I have heard similar comments from other direct mail providers. Specialty houses that can do smaller print runs using variable digital printing for highly personalized communication so far seem to be holding their own, although there has been a lot of industry consolidation. It is my impression that we are living through an unprecedented change in the industry. Much of this change was driven by the collapse of our financial institutions, many of whom were heavy direct mailers.

The DMA (Direct Marketing Association) in New York itself has gone through several rounds of layoffs. I recently was nominated for a position on the board of CADM (Chicago Association of Direct Marketing) so I will get to help guide this organization through these troubled waters. In times of downturn, it always seems to me that basic marketing principles are key. Best Buy survives because it has a clear differentiation strategy and by the way makes great use of its customer data. Circuit City tried to play catch up but just could not distinguish itself in the marketplace.

Catching up from last week, we also talked about SEM (Search Engine Marketing) and SEO (Search Engine Optimization). Search is divided into paid versus organic search. Keywords are the terms users employ when searching for something. Organic search is free. Search engines use algorithms to determine page rank. Web spiders or 'crawlers' search the web and analyze pages to determine page rank. In natural search, these programs are looking to see that the blue title onthe top matches web site content, that the content of the web site matches the keywords used in search and to some extent the keywords in the HTML code, and that there are sites linked to the site and that the site links to others. These crawlers cannot read pictures so it is important to 'tag' pictures with descriptions that match the keywords the company wishes to identify with the site.

We also talked about the importance of good writing in business communication. I will spot each group three and only three poorly constructed sentences, typos and/or grammar mistakes in deference to our short-class format. After that, it is a point taken for each problem noted by me. We talked about the proper use of demonstrative pronouns. Except in rare cases, pronouns such as this and they in formal writing need to be substituted for a proper noun or have some kind of additional noun used with them. Instead of saying "This is important in multichannel marketing,' for example, substitute "This trend" or "This concern." Avoid addressing the audience as 'you' except in persuasive communication.

Multichannel Marketing Class May 26th: Multichannel Retailing

We covered multichannel retailing using a webinar case study from Exact Target on Johnston and Murphy. Johnston and Murphy is an upscale shoe retailer, primarily men's shoes, with three fullfillment channels, store, catalog and internet (I am assuming that they also have a dial-in call center for fulfillment but it was not mentioned). J&M contacts its customers via direct mail and email, in-store promotions and its web site. It did not seem that the company used brand advertising. I would think that the economics of its business is such that direct communication is more cost effective.

The company has 700,000 customers on its database and knows who they are and can contact them directly so direct and interactive marketing make sense. J&M started by segmenting its current business into eight key buying segments with different characteristics. By capturing 60% of customer email addresses at point of purchase it can then use purchase data to determine which segment a customer belongs in and tailor offers to them. All interaction is captured in the customer database; the company uses match back on address, phone number, email, etc. to put all possible information in one place. Three emails are sent to new customers within 40 days and current customers also get a three email cycle, offer, discount 1, discount 2. "Lost" customers are those who have not ordered in two years and they also receive email treatment.

The program has been successful for J&M. The email program is set up and then adjusted every six months or so so the communiation is automated. Data collected from transactions is modelled to determine the best combination of communications to elicit the maximum response. Students in class presentations talked about other retailers doing the same type of marketing effort by building a database, capturing transactions, segmenting and targeting. Student also reinforced the research mentioned in class that multichannel shoppers are typically more profitable shoppers.

This webinar is a good transition to the next two sections of our class. Email and Direct Mail are channels that are 'push' channels and are acquisition oriented. We use military language like 'target' and 'capture' to describe our efforts and the marketer is still somewhat in control. As we began to disucss last night, with the internet as a channel, control shifts to the consumer and we will hear more about that next week with our discussions on internet and social media. Next week we will also learn more about data analytics, which will aid understanding of the modelling efforts J&Q mentioned.

The real multichannel marketing challenge

I attended the Direct Marketing Association Conference and presented at the associated Direct Marketing Educator's Research Summit in early October. It seems that marketing practitioners are struggling with the entire multichannel concept. The fact is, analyzing customer responses across channels to figure out the best way to communicate with them (how many touches, over what time frame and over what channel) to drive response is not an easy task. In the first place, most firms still don't have their data from various marketing programs in one place. In the second place, the level of sophistication of data analysis needed to solve these problems is considerable. Although there are software firms out solving this problem, to a large extent the marketer must rely on response models and statistical techniques unique to the firm and requiring a comfort level with statistics that most marketers do not possess. I think this trend just reinforces the fact that marketing is becoming a more technical discipline. In terms of the educational field, I think the recent students from University programs, particularly those who have taken courses like the database course I teach, will be better prepared to meet the multichannel marketing challenge.