Is Your Customer Base Changing?

A couple of weeks ago I began to write an entry about a phenomenon we often speak to our clients about: how it is important to focus on cultivating and retaining best customers, but at the same time, it is just as important to watch for changes in the customer base that either indicate a serious threat or a new opportunity. As with about half my blog entries, it sat unposted; in this case, because I struggled with a good example that would not be too obviously one of our clients.

Fortunately, the news media came to the rescue, with an article in the Wall Street Journal about Burger King's current challenges. The article describes Burger King's successful strategy of catering to "super fans", which are characterized as 18- to 34-year-olds that comprise about half the visits to their restaurants and that visit fast-food restaurants an average of 10 times per month. For those not familiar with Burger King's history, their market position had eroded over many years in the 1990s and 2000s. In response, the company devised a new strategy in the mid-2000s that led to five straight years of increases in same-store sales.

All seemed well, and it seemed that the company had implemented one of marketing's mantras – focusing on one's best customers – successfully. There were some naysayers, such as some franchisees that felt the strategy was eroding other customer segments, such as families with young children. And there were critics of some of Burger King's marketing campaigns which stretched the norms of the fast-food category. Yet, overall the strategy paid off, until same-store sales began to fall in 2009.

Burger King, in its financial reports has pointed to the fact that the wider economy has an adverse effect generally and has hit the 18- to 34-year-old segment particularly hard. On the other hand, some analysts have pointed out that this segment was already reducing its fast-food trips before the recession, as consumers sought healthier alternatives. McDonald's, which worked through market challenges a few years ago by introducing healthier menu items and emphasizing its family-friendly environment and value menu items, saw same-store sales increase by 2.5% in the quarter ended September 2009 versus Burger King's decline of 4.6% during the same period.

Burger King is a special example of a larger dilemma many companies face. We need to appeal to segments that either spend disproportionately within a category or that prefer our brand. If executed well, this strategy brings short-term success, but carries with it long-term risks, because consumption patterns and brand preferences shift. The recession accelerated such shifts, but the danger is always there.

There are ways to mitigate the effects of such changes. The first and most important is to monitor the customer base with an unbiased eye. I say "unbiased," because as marketers, we tend to fall in love with valuable customer segments and tend to filter reality through those emotions. In a sense, marketers fail to recognize that "he's just not that into you" when enthusiasm among their best customers cools. Of course, this does not mean we want to abandon customer segments, but we do need to recognize changes before they become a wider problem. And, metaphorically speaking, we did to think about dating some other segments that represent an opportunity to build a lasting relationship.

Despite the common use of the words "customer portfolio," marketing practice often does not implement the thought contained in the phrase. Good financial portfolios are designed to spread financial risk to achieve long-term stability. Similarly, a customer portfolio should be managed so that there is always some balance in customer composition. Again, that is not to say that we should not exploit shorter opportunities to grow some segments, but not if it results in unbalancing the customer mix too far.

It will be interesting to see how Burger King and others that have seen their core customer segments eroded by the recession and other factors respond to these pressures. And the current winners – the Wal-marts and McDonalds of the world – will need to avoid some of the same potential landmines over time.

 

Posted at 08:43AM Feb 04, 2010 by David King in Data  |  Comments[0]

The Next Big Thing in Interactivity?

The evolution of interactive marketing may be close to taking its next big leap forward.

Let's face it, despite the fact that we use "interactive" to describe digital marketing, there's relatively little true interactivity. Most interactions still involve people sitting at a keyboard (or maybe a touchscreen) and typing or clicking away.  Speed of interaction has certainly increased  – live chat, for instance, allows us to communicate in nearly real time – but basically, we are still pecking away at a keyboard. Low-cost video conferencing, such as what we can do with Skype and similar services, lets us add a face to a phone call.

The hardware and software that we use to communicate has been the chief obstacle to achieving something that approaches a really personal interaction. But that may be about to change.

Augmented reality (AR) technology is reaching a point where it may be affordable enough for a wide range of applications, including marketing. The simplest way to understand the technology is to see a demonstration...I suggest you check out the video about Microsoft's Project Natal Xbox360. The aim of this system is to continue the evolution of gaming interactivity that Nintendo's Wii began by doing away with controllers entirely.

The current Xbox360 is already a beast of a computer,  and it's ability to render high-definition graphics far exceeds the capacity of most of our personal computers. The same goes for other consoles, such as Sony's PS3, so it will not be surprising that gaming consoles may represent some of the first practical mass-market augmented reality technologies.

We also see the beginnings of practical implementations on the Internet, although local hardware (web cams and graphics processors) and network bandwidth are still limiting factors. For example, there are now toolkits available for interactive developers, such as the FLARtoolkit developed by the developer Saqoosha and the tools from commercial developer Total Immersion. Other technology providers include Boffswana, Flat Out Media, and metaio.

Some of the early web applications have been the 3D views of cars, in which a user can rotate a vehicle and interact with some of the car's features. Cool, but not game-changing. But there are many potential applications, some of which have been tried on a limited basis, that may represent a competitive advantage to companies that adopt such technology.

• Online shopping with realistic virtual "dressing rooms"; 
• Interactive user training...imagine an application in which a diabetic patient gets help using a new blood glucose meter interactively;
• Marketing "videos" in which the customer can interact with the product in complex ways, maybe even with friends or family members;
• A whole new set of social networking applications that don't restrict us to avatars, as in Second Life.

It is always hard to gauge the effect of an emerging technology, particularly how long it will take to become disseminated widely enough, but augmented reality seems to be on the cusp of a mainstream breakthrough.  Check out some of the links above and decide if there is a role for the technology for your company in the next few years.

Posted at 11:38AM Feb 03, 2010 by David King in Technology  |  Comments[0]

Directing Segmentation

With the arrival of the new year (and new marketing budgets), more than a few companies may be contemplating developing and implementing a new customer segmentation. Having seen my share of such projects over the years, I'd like to offer a specific suggestion for how marketers can plan for a more useful segmentation.

The "Tell Me Everything" Approach

Segmentation is one of the fundamental tools of marketing...we simply could not function effectively without being able to classify markets, prospects, and customers into useful categories. Too often, though, the drive to build customer insights is expressed as a vague concept, which might be summarized in the following statement: "our new segmentation should tell me everything about customers, so that the best marketing programs can be designed and rolled out."

There are several potential pitfalls here. For example, there is no objective definition of  what is meant by "best" marketing programs: imagine developing a creative brief that had only this level of detail. Yet it is surprising how many segmentation initiatives are launched around such vague notions.

Similarly, we face problems, both theoretical and practical, in developing a segmentation that is supposed to tell us "everything." For example, one issue we face is that as we try to incorporate more dimensions into the segmentation, the more complex it becomes. Often, the complexity becomes significant enough that the decision is made to "collapse" the many resulting micro-segments into a more manageable and smaller set of segments. For this reason, many companies have arrived at enterprise segmentation schemes with six to eight segments.

I characterize this approach to segmentation as "unguided," in the sense that there is a desire to identify "natural" segments, from which we are supposed to infer insights and strategies. In other words, the analysts are supposed to find the true meaning locked within the data and reveal it to the organization. Yet, the best insights come when we are looking for particular things or when we have a specific outcome.

Setting A Direction

So now to my bit of advice to anyone embarking on segmentation: focus on the desired strategic outcome and you will achieve better customer insights.

Here is a quick example. Let us say that an important goal for the company is to reduce customer attrition from 20% annually to 18% annually; if successful, the company would see a significant increase in its profits. With this goal in mind, it is easier to formulate a segmentation that supports the company's strategy. At a high level, we could ask that the segmentation be able to answer a few fundamental questions:

• Which customers are most likely to leave in the future? Too many segmentation schemes are static and fail to yield any ability to forecast the future. In this example, what if attrition rose to 22%?
• Why are a fifth of all customers leaving each year, and are there any discernible segments that cluster around specific reasons (e.g. some customers believe the products are too expensive, while others believe that the company's technology is outdated, and still others have been angered by poor customer service)?
• What will induce customers to remain with the brand? Here, we may also different segments, with some customers potentially representing little opportunity for retention, while others need to have the right response from the company in order to induce them to stay.
  

We could certainly look at other dimensions, but by focusing on the strategic outcome of improving attrition rates, we can hone in on the most relevant questions and formulate a response.

I used the example of a strategic segmentation built around attrition, but we could just as easily use something even more general, such as customer value. We would begin by placing customers into different value segments, then use other data to expand the segmentation. In particular, we would want to understand how to increase value across various segments. This means we need to understand what will enable us to persuade customers to spend more, and then devise marketing programs that use these insights. We also may have segments, in which maintaining value is the most critical aspect; for example, highly valuable customers may not have a lot of potential for more growth, but we want to make sure to retain the value they represent.

We are also great advocates of what I would call "tactical segmentation". For example, let's say we want to build a basic response model, something that would probably employ logistic regression. In nearly every case, that model can be improved by using latent class regression, which will produce a segmentation local to the dependent variable we are trying to predict. Plus, we now can exploit the differences between segments to develop more effective campaigns. Here is an even more clear example of "directing" segmentation around a more specific problem.

Posted at 01:00PM Jan 13, 2010 by David King in Data  |  Comments[0]

A New Decade, A New Census

My summer job in 1980 was as a census enumerator for the U.S. Census Bureau, a position that involved visiting homes of people that had failed to return their census form or had for whom the information was incomplete. It was a fun job (other than the occasional run-in with aggressive family pets) and it paid well: I was compensated for each completed census form, as well as for completing certain other tasks.

Another aspect of the job was verifying the existence of homes. The area I grew up in was undergoing a major housing boom and large tracts of land had been subdivided and already had been assigned a postal address. In many cases, there were no actual houses yet, so I would dutifully note this on the forms I submitted.

For marketers, census data is a part of the data we manage about markets, prospects, and customers. If we want to understand the demographics of a market, our analysis will rely heavily on information derived from the Census. Census data or attributes derived from this data (such as "lifestyle clusters") is appended to many customer databases to provide additional variables for reporting and analysis. And a good number of vendors make their living by reselling, repackaging, and augmenting the base data.

This year's census form contains only 10 questions (a sample can viewed here), and focuses really on household composition, age, sex, and ethnicity. Many of the detailed attributes we associate with census data is collected via the American Community Survey, which is an ongoing survey conducted each year against a sample of the population. Back when I was a census taker, the ten-year census had a short form and a long form; the short form was still lengthy compared with today's survey and the long form was analogous to the American Community Survey.

The forms are now in the mail to American households, and the Bureau expects about 70 million forms to be returned, leaving many others that will need to be gathered by census takers. Once the data is collected and analyzed, all sorts of things will flow from it. First there will be the predictable debate about over- or under-counting: the census tends to over-count the affluent, especially those that own multiple homes, and it under-counts minorities and non-citizens. States may gain or lose congressional seats in the U.S. House of Representatives and census data will guide congressional redistricting, which may represent one of the most creative uses of demographic data.

Marketers will be using it to understand more about markets and customers and eventually to target consumers for marketing, a use of the census that I doubt the writers of the U.S. constitution foresaw when they wrote: "The actual Enumeration shall be made within three Years after the first Meeting of the Congress of the United States, and within every subsequent Term of ten Years, in such Manner as they shall by Law direct." But then that is true of so many things that have taken place since the first census in 1790, which covered thirteen states and resulted in an estimated official population of 3,929,214.

Posted at 12:49PM Jan 04, 2010 by David King in General  |  Comments[0]

Managing Advertising Privacy

With U.S. regulators and legislators continuing to look at targeted online advertising,  the networks whose businesses rely on serving such ads find themselves in a potentially difficult position. 

The networks need to be concerned that legislation or regulation will be put forth to restrict the collection or use of data to target advertising at online users. An important part of the appeal of online advertising is that it has been presented as more targeted than traditional media, such as a TV, print, and radio, and this has been a factor in the shift to online. If the use of browsing behavior and other attributes is curtailed, then online advertising may no longer be as attractive, with the result that advertising rates and overall growth may decline.

So far, the networks have responded cautiously.  For example, both Google and Yahoo! have released tools to allow users to manage their advertising preferences. Google's Privacy Dashboard permits users to see what Google is tracking about them (note that you need a Google account in order to view this information). Yahoo! has released its Ad Interest Manager, which similarly provides information about the user, but which also permits users to modify what types of ads are served. In order for the Ad Interest Manager to retain your settings, you must have a Yahoo! account and accept a cookie from Yahoo! Both organizations, along with most of the other ad networks, also belong to the Network Advertising Initiative which has also released an add-on to Firefox, which essentially helps preserve the cookies from NAI member sites, so that consumers' preferences can be maintained.

Another potential danger for the ad networks is consumer awareness and action. If significant numbers of users were to restrict targeted advertising, then the networks' reach would be reduced, which in turn could slow the growth in advertising revenue. Fortunately, if history is a good guide, this scenario is unlikely; most consumers do not use such browser features as deleting cookies or "private browsing."

How all of this plays will play out over the next few years is unclear. Again, if history is a guide, then we are likely to wind up with notification requirements, rather than outright bans on behaviorally-targeted ads. In this scenario, sites using targeting methods would need to clearly inform users about what data is being collected and how it is being used. Since much advertising is delivered across networks of affiliated sites, all sites within a network would need to comply with a common set of policies.

There is evidence to show that younger online users care less about privacy than older ones. And even among the wider population, only a small minority actively use existing privacy tools. In the end, these reforms are likely to add yet another artifact adorning web sites, that like the ubiquitous privacy policy statements, few people ever bother to review.

Posted at 09:59AM Dec 15, 2009 by David King in Data  |  Comments[0]

An Emerging Digital Divide

In a recent article, I noted that social networking and mobile devices were creating a shift in how people interact "online". The two phenomena seem to be converging in a way that supplants the desktop computer and traditional applications, such as email, with social interaction centered around the handset.

Another aspect of this trend was highlighted on a segment during the December 1, 2009 broadcast of NPR's Morning edition. The story is based on  research from the Pew Research Center that shows young Hispanic and African American mobile users are using their handheld devices more – and for more activities – than their white counterparts. A big driver for this higher usage is that applications such as texting are preferred for keeping in touch with friends and family. Another driver: adding additional mobile services is seen as more affordable than adding a home ISP connection and purchasing a computer.

A few years ago, one of the debates was the "digital divide" that separated whites and minorities based on their access to the Internet. Mobility seems to have flipped the divide, with whites now trailing behind blacks and especially Hispanics in their consumption of mobile services.

Speaking of digital divides, a well-known study from 2006 suggested that we are getting more isolated from each other, and that technology either was not stemming this trend or was actually accelerating it. A new research study from Pew challenges this popular notion that technology is leading to increased isolation.

Posted at 03:18PM Dec 01, 2009 by David King in Technology  |  Comments[0]

Behavioral Economics at the Wharton School

A short article – just a reference, really – to an article in today's issue of Knowledge@Wharton. We've been writing a lot about the importance of behavioral economics in marketing: this article presents an interesting overview of the debate going on the policy levels of government and industry about economic theory. It's worth a read.

By the way, the article also mentions the fact that Elinor Ostrom was one of this year's Nobel laureates. I commented on her work in an October 12 entry and its applications to Internet marketing and social networking.

Here's the link to the Wharton article: Efficient Markets or Herd Mentality? The Future of Economic Forecasting

Enjoy.

 

Posted at 05:51AM Nov 12, 2009 by David King in General  |  Comments[0]

Variety Is The Spice of Marketing

One common theme in current marketing punditry goes something like this: "consumers are overwhelmed by choices and are seeking simplified offerings." As an example, these writers point out that thousands of new consumer products are introduced onto grocery shelves each year, but only a handful will survive.

While there may be some merit in the core argument that at least some consumers find the array of choices overwhelming, I think it ignores the fact that a more complex process is at work. Here are some things not considered in the simple statement of the problem:

The assortment of items within categories has been going up, because it is a winning strategy for retailers and manufacturers. Imagine a grocery store that took an approach to merchandising that only permitted two products in each category: regular Coke vs. regular Pepsi; Diet Coke vs. Diet Pepsi.  Sales would be restricted to those consumers who like one of these choices. Walk down an actual aisle in an average grocery store, and you will see up to a half-dozen different formulations each of Coke and Pepsi, plus many smaller brands. The incremental cost of producing a new selection is relatively low given modern manufacturing processes, and therefore all a new product or flavor needs to do is to attract a small minority away from competing lines in order to be profitable.

Variety is not just a variable independent of any other factor. In fact, the availability of product choices can heavily influence purchasing behavior. For example, let's say we have two competing companies offering dish detergents: one sells for $2.25 a bottle and the other for $2.50 for the same-sized container. The introduction of a premium brand selling for $3.00/bottle by the second manufacturer might cause some interesting changes. First, the premium brand may take relatively more market share from the more expensive choice. However, it may also stimulate its sales, as the previously more expensive selection may now be perceived as a relative bargain.

There is good evidence that a greater number of choices can spur consumption directly. For example, researchers in obesity know that increasing the number of food choices increases caloric intake. For example, one well known research study showed an average 14% increase in caloric intake when subjects were offered two types of pasta shapes instead of one in a meal. There is also a sequential aspect to this phenomenon: when consumers are offered a choice of items and have them served on a single plate, consumption is lower than when consumers are offered the same items, but each time as a single course.

All of these aspects of assortment suggest that data-driven marketers would do well to pay attention to ways that variety may be influencing customer behavior. Here are a few things to consider:

• Many marketing tests use simple methodologies. For example, one set of customers will be shown Product A, and another set, Product B. In the real world, products and services compete in a wider ecosystem where decisions are unlikely to be binary.
• The phenomenon of "early adopters" is well accepted in product marketing. Early adopters are those customers who purchase soon after a product's introduction. A good percentage of these customers may be "variety seekers," buyers that like to try new technologies or products, but who will move on when new products appear. Basing product roll-out primarily on such customers can lead to misleading approaches
• Variety seeking may also be higher among a company's "best customer" segment. Ironically, for some companies, attrition is quite high among segments that spend heavily and might be considered "engaged" or "brand loyal." There may be sub-segments consisting of variety seekers who need a steady stream of new product or feature introductions in order to remain loyal.
• Much of marketing is still planned around discrete campaigns that try to maximize the sales of a single product or service. Here, too, there may be segments that can be grown more effectively by a longitudinal approach that offers a "menu" of items in sequence.

Posted at 03:31AM Nov 10, 2009 by David King in Analytics  |  Comments[0]

I'll Go To The Gym -- Tomorrow

One of the wider debates about the economy is whether the severe recession will alter consumption patterns. From the perspective of classical economics, one might suggest that the answer is "yes." Consumers, businesses, and financial institutions will have learned from the experience and will return to a more balanced, cautious approach to consumption, borrowing, and lending. The traditional view that markets consist of parties acting rationally and in their best interests promotes the notion that we will learn from the experience and will practice better self-control.

On the other hand, behavioral economics suggests that we largely may revert to the same consumption patterns that led to the crisis. As we've discussed in earlier arguments, behavioral economics holds that many of the decision-making processes we use are not rational and that we often make decisions that are not in our best long-term interest.

An interesting field of study helped by this branch of economics has been the field of weight loss. There are some good analogies between losing weight and trying to manage money more responsibly. Despite the billions Americans spend each year on dieting, the introduction of hundreds of "healthy" foods each year, and the constant drumbeat of how to lead a healthier lifestyle by government, health advocates, and news organizations, obesity rates continue to climb. Similarly, despite a great deal of advice about financial planning and the benefits of long-term saving, the rate of savings had continued to decline steadily until the recent downturn.

One powerful insight of behavioral economics is the mechanisms that people use to make a choice over time. Dieting and exercise have long-term benefits that most people recognize. Yet, when presented with a choice – let's say, watching a favorite show now or going to the gym now – we tend to take the choice that offers instant gratification. We get an immediate and tangible benefit from watching our program or eating a piece of chocolate cake and a deferred benefit from eating an apple or hitting the gym. The gym may be embraced with enthusiasm, especially if it is a choice that can be deferred to the future.

Similarly, researchers have learned that emotions usually trump rationality in the choices we make. The sight, smell, and taste of chocolate evoke an emotional response that can win out over the desire to diet. The same types of emotional responses accompany any desirable object, and we are thus prone to buying the dream house or luxury car, even though its cost exceeds our budget. In this area, there is also plenty of evidence to suggest that a significant minority of people who have made a considered choice will abandon it in light of a compelling emotional option. For example, in experiments where people have pre-selected a meal from a menu of healthy choices, up to half will end up actually choosing unhealthy items when presented with an assortment of items that includes both health and unhealthy foods.

As marketers design new marketing programs for the recovering economy, one area that will need to be considered is the short-term economic realities that will continue to hold consumption in check (unemployment, consumer anxiety, less available credit, etc.) and the fundamental dynamics of how we consume. Once some of the immediate concerns abate, it is likely that many of us will revert to pre-recession patterns. Which means we'll once again be savoring our chocolate cake while contemplating our visit to the gym tomorrow.

Posted at 06:38AM Nov 09, 2009 by David King in General  |  Comments[0]

The New Social Mobility

Every marketer is trying to devise a smart social marketing strategy. One key driver: Facebook has seen the highest increase in traffic of any Internet site over the past three years.  

At the same time, marketers are trying to work out good mobile strategies. A key driver: the iPhone (and more recently Android-based handsets) have become the primary browsing device for a growing number of consumers.

Yet, companies that view these strategies as separate efforts may be missing an even important phenomenon: social media and mobile are evolving together in ways that reinforce each other.  Social media by itself is partially displacing other, more traditional, interactive forms of communication. For example, such activities as emailing one's friends is being supplanted in some segments by messaging through Facebook and Twitter. Mobile devices, meanwhile, are being used by consumers as an alternative to the desktop, particularly for such activities as email and social networking. Companies in this space have seen the potential for some time: Apple's iPhone was visionary in this regard, Google's heavy investment in technologies like Android and Waves represent efforts to stake a claim to this emerging space.

One of the best overviews of the trend that I've seen is a recent presentation by Mary Meeker of Morgan Stanley.  She has assembled a wealth of useful statistics that illustrate the convergence of mobile and social media.

From a data perspective, this convergence presents unique challenges. Most companies have a wide gap in the information and understanding that they have about their customers' mobile usage and social networking. As companies seek to develop strategies in these areas (and, one hopes, have them well-integrated with other customer interaction channels), mastering the information flowing from these sources will be a critical element to success.

Posted at 02:34AM Oct 29, 2009 by David King in General  |  Comments[0]

Elinor Ostrom Wins Nobel Prize in Economics

The Royal Swedish Academy of Sciences announced today that it awarded The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel for 2009 to Elinor Ostrom and Oliver E. Williamson.

Dr. Ostrom's work has challenged traditional ideas of how common property can be managed. In general terms, economists have approached the management of common property in one of two ways: government bodies are best suited to managing resources for the common good; or private markets are more efficient at managing commons. The ongoing political battles over the environment provide a clear example of these two schools of thought. On the one hand, most environmental groups support government management of natural resources, such as forest and fisheries. Their arguments are countered by commercial interests and property-rights advocates who maintain that stewardship of such resources is best accomplished by private interests.

Dr. Ostrom's work, based on many specific studies, examines how users of resources can cooperate in managing resources, with outcomes that generally exceed the predictions of standard theories. She highlights how groups of users, sometimes with competing interests, can work together to develop quite complex management schemes that help the group make decisions and resolve conflicts. As we pointed out in other articles, some of the most interesting work in economics in recent decades has been around how many decision-making processes operate outside of the boundaries of traditional economic theory, whether it be such areas as behavioral economics or the areas that Dr. Ostrom has brought to light.

By now, you must be wondering what her work has in common with marketing. Well, for one, social media represent a good analogy in which a community of users develop sophisticated rules and mechanisms for managing a common resource.

On Wikipedia, users with a shared interest in a topic often bring divergent viewpoints about a subject. For example, on an article about a musical group, some contributors may be ardent fans, while others are vehement detractors. In some cases, there are intense battles between authors over  content, but there are community rules and norms that eventually result in more balanced and complete articles. In fact, the more participants there are, the more likely the article will be balanced.

Chris Anderson described the Wikipedia phenomenon this way in The Long Tail:

"In the popular entries with many eyes watching, Wikipedia shows a remarkable resistance to vandalism and ideological battles. On study by IBM found that the mean repair time for damage in high-profile Wikipedia entries such as 'Islam' is less than four minutes. This is not the work of the professional encyclopedia police. It is simply the emergent behavior of a Pro-Am swarm of self-appointed curators. Against all expectations, the system works brilliantly well." [The Long Tail, revised edition. 2008. p. 70]

Wikipedia's arbitration council intervenes when authors cannot resolve disputes or blatantly partisan in their entries, as it did recently around articles concerning Scientology, which in turn set off a debate among bloggers about whether Wikipedia was being too heavy-handed.

Technology has enabled many other examples, ranging from open-source software development...to Facebook and Twitter...even to Google. All of them share the same elements: users acting in both their own interest and that of the community (although not always at the same time) and a basic set of rules and shared etiquette that govern behavior and mediate disputes. Once such phenomena reach critical mass, their influence is often surprising: think about the Linux operating system's presence in corporate IT, the success of Google's IPO, or the sheer number of users on Facebook and Twitter.

Marketers are both excited and apprehensive about such self-organizing communities. On the one hand, each community is seen as a potential marketing channel, but on the other, unfavorable opinions about a company can now propagate rapidly. As a consequence, marketers are still trying to feel their way through this social order.

Dr. Ostrom's work in another field, as well as that of other researchers, suggests that user-organized groups are far less chaotic than they appear. As a consequence, companies should work to understand the underlying mechanisms of social behavior before wading into social marketing under the assumption that it is like any other channel.

To view a representative bibliography of Dr. Ostrom's work, visit her page at Indiana University's web site.

Posted at 07:14AM Oct 12, 2009 by David King in General  |  Comments[0]

Why Counting Clicks May Be OK

A lot of web analytics are based on simple counting – the number visitors, the number of new visitors, the domains visitors came from, etc. This sort of simple analysis may not provide a great deal of useful insight, and analysts in other areas of marketing (most with richer data) often criticize web analytics for its basic nature. But search and clickstream data may turn out to be more valuable than one might think.

We recently had some of our thoughts on this topic published at iMedia Connections.

 

Posted at 03:59AM Sep 14, 2009 by David King in Analytics  |  Comments[0]

Intelligence Through Meta Data

In my last article, Using Hierarchy in Data, I wrote about the importance of hierarchical data and how it often is neglected due to the challenges in brings. My argument presupposes that such data has been captured and formatted in a way that preserves such information. Moreover, there are opportunities to enhance the existing data with additional attributes that strengthen our understanding of customer behavior.

Here is a simple thought-problem that illustrates the point:

Your web site's home page devotes about a third of its real estate to three featured products, which are presented as part of merchandising agreements with manufacturers. At any given time, you have about a dozen of these in "inventory" and your web server offers these up in rotation. You track clicks for each featured product and can trace the clickstream all the way through to purchase.

So far so good. But can you actually retrieve the three offers that each customer received on the home page? There is much evidence that the context in which an offer is made – in this case the assortment of three items – can have a significant influence on the acceptance of the offer. Could you systematically identify which combinations of offers generated the highest interest and purchase rates? Most companies discard this data, and retain only the actual log file data (alas, some companies fail to track even this information). Admittedly, on large sites, the amount of data generated would be quite large, but here is information that is central to generating revenue.

Let's take the example one step further: do you have sufficient meta data about these featured products to make meaningful decisions about what to feature in the future? Because merchandising arrangements and SKUs change often, even if you retained the combinations of offers made, the information may be too specific to be useful over time. Rather, in addition to the product-specific information (e.g. price, color, etc.), it may be important to maintain some additional meta data. For instance, it may be that products that boast the newest features (but that are more expensive) do best when offered in combination with a product with a standard set of features, but also perform poorly when offered with an obsolete, but very inexpensive, item.

Such information would also yield valuable insights on a customer level. For example, what if some customers always buy the newest products, regardless of what other items are offered in combination? That suggests opportunities for constructing a more tailored set of "featured" products to those customers that seem to be technology-oriented. If such offers can be constructed from the merchandising inventory, then the retailer, manufacturer, and customer may all benefit from more relevant offers.

In this example, we could create a meta data element that describes the product as "new," "mainstream," and "obsolete."  And here, we probably need to keep some time-series history, as today's leading edge  product will be obsolete some day.

I recognize that creating such meta data represents a challenge, both from the perspective of developing appropriate attributes, as well as creating and managing tags. For a site that may have several hundred thousand SKUs, it may not be practical to do this for every product. On the other hand, there could be tremendous value in doing this just for the featured products, since their importance is elevated due to the merchandising dollars supporting their placement and (presumably) the higher purchase rates that promotion generates.

Here are a few additional areas, in which these concepts can be applied:

• Catalogs: Could you recreate the essence of a catalog that you produced six months ago? [Retrieving a PDF of the catalog does not count.] 
• Weekly circulars: For many retailers, circulars are an important marketing vehicle, particularly since they are supported by merchandising. Similar to the web and catalog examples: do you have the ability to reconstruct the circular from stored information and meta data?
  
• Digital marketing: Emails, PURLS, and digitally printed mailers may feature variable content and offers. Can these be recreated to reflect accurately what each customer received?
• Content sites: Can you identify what types of information users are seeking? In other words, do you have meta data that categorizes content so that you identify patterns that may cut across some of your content areas (for instance, I may not really be interested in Sports or Entertainment, but I will read content in these sections when there is some scandal involved).
  

The full list of applications would encompass many more areas. Stated more broadly, the real goal is to create a complete picture of the company's interactions with its customers across all channels. And in order to accomplish that, we need to retain and store both the operational data and  enrich it with contextual information that generates more meaningful insights.

Feel free to share any experiences or success stories you have around this topic in the comments section.

Posted at 05:44AM Sep 01, 2009 by David King in Data  |  Comments[0]

Using Hierarchy in Data

Hierarchical data is all around us. Think of any business and you are bound to find a large number of hierarchies: management structures, product categories, geographical organizations, and many others. Hierarchies are not just convenient ways to organize information, but are themselves a form of information.

Yet in many statistical methods, dealing with hierarchical data is difficult. Often, such data have high cardinality and the hierarchical structure itself presents a challenge, since many methods need a "flattened," or denormalized, analytical view. Here are a few common examples of such business data.

• Sales offices and retail locations: a company may have several hundred physical locations that sell and service customers, with great variation in their size, composition, and trade areas. Moreover, at each location, we are likely to have additional hierarchies, such as sales management/sales force or department/category/sub-category.
• Content categories and sub-categories: Web sites, printed directories, and similar properties may have hundreds or thousands of categories, each attractive to certain types of advertisers, although some advertisers also list in multiple categories. Content itself is often arranged hierarchically, so that users can start with general information and drill down to specific items;
• Markets and trade areas: A company may operate in hundreds of markets...think telephone directories, retailers, and home-based services, to name but a few. Depending on how granular the analysis needs to be, markets potentially number in the tens of thousands (e.g. census tracts, of which there are over 66,000).
  

Let's use as an example, a publisher that serves 800 markets with directories that contain 2,000 category headings. We want to have a model that prioritizes leads for a field sales force, who call on local businesses.

Plumbers are good advertisers nationally, but the markets they operate in are also a strong determinant: in some markets, plumbers are prospering and advertising; in others, they are not, perhaps due to local economic conditions. So we know there are interactions between the advertising category, geographical markets, and spending on advertising.

If we just look at the plumbing category, we may see that on a national level, the differences between plumbers in different markets tend to cancel each other out. If this is the case, then the plumbing category will not be useful in the model, and even if it is, we then have a variable that gives the average for plumbers nationally, even though their spending seems quite polarized. So we know that there is variation that would be helpful in modeling, but it is difficult to use (especially considering that we still have another 1,999 categories to consider!). 

One strategy might be to build a clustering model that lets us group similar markets and categories together; these groupings would then be used in the model. Hypothetically, plumbers might be grouped with other building trades and car dealers, and trial lawyers with chiropractors. In the same way, we might group markets together, so that Detroit and Las Vegas are in one bucket, and Dallas and New York in another.  But deciding what is similar is subjective, and the categories may not remain similar. For example, we currently have a mini-boom in auto sales, due to the government's cash-for-clunkers program, which has injected $3 billion into that market. Construction, however, continues to be depressed nationally, and we can expect plumbers in many areas to be tight with their spending.

There are other many other data transformation methods that might be employed to try to reduce the number of categories and markets into a compact form suitable for modeling. Again most will either require us to make some subjective decisions, or flatten some of the data and thereby lose information.  There are also some analysts that might advocate a particular technology – neural networks, for example – in the belief that the data can just be dumped into software application, which will figure out how to make sense of the data.

Fortunately, there are well-established ways in which such hierarchical data can be incorporated. One elegant method we employ often is hierarchical Bayes modeling, which, as the name suggests, explicitly accommodates the use of hierarchical information. While the methodology is technically challenging (many marketing analysts may not be wholly comfortable even with simpler Bayesian methods), we have found the resulting models to be both robust and stable. In some cases, they have also been much more compact, as direct use of hierarchy reduces the number of variables in the model, compared with more traditional approaches.

We feel that this is a rich area for growth in marketing analytics. Hierarchical data is so ubiquitous, yet so under-used, that we in essence have significant new data sources for analysis. This has been an active area of research for Fulcrum: beyond hierarchical Bayes, our R&D team has found several other useful methods for using hierarchical data.

For those interested in learning more, feel free to contact us. You can also catch a presentation on this topic by our Chief Scientist, Hongjie Wang, at the National Center for Database Marketing conference, in December 2009.

 

Posted at 03:43AM Aug 28, 2009 by David King in Analytics  |  Comments[0]

Estimating Customer Share-of-Wallet

Market share is one of the fundamental metrics that companies monitor to see how they are faring against their competitors. But companies often want a more granular look at what share they command of their customers' spending in their category. A particularly attraction of this data is the potential to find customers that could consolidate their business with the company and generate significant incremental value. It is also seen as a good indicator of loyalty and customer engagement.

Before You Make the Leap...

While the the concept of being able to understand share at the customer level is a seductive one, companies should consider a few things before embarking on such an initiative. First, one needs to recognize that many markets are mature and stationary, which means that market share at the company level is stable. Absent some disruptive event (poor quality from one of the competitors, a new entrant into the market, etc.), market share is expensive to change solely through marketing. (By the way, we wrote an article about market share dynamics a while ago; you can find it here.

The "macro" share is a function of the "micro" actions of customers in the market. Merely knowing how customers are allocating their purchases does not imply that marketing alone can persuade low-share customers to switch their business. Again, it is likely to take some larger set of changes – new or differentiated products, pricing schemes, or some other significant factor – to change the dynamics in a meaningful way.

In fact, I believe that one of the best reasons to consider investing in understanding customer-level share is precisely when a company is planning such a major change, because for it to be successful, it must either change the dynamics of market share or represent an opportunity to become more profitable with the same market share.

One other compelling reason to contemplate understanding customer-level share is as a tool to monitor the health of the customer base. Often, overall market share erodes slowly, but it manifests itself first among certain customer segments. If we start to observe declining share in valuable segments, we need to address the problems before they lead to a real decline across the whole base.

Common (and One Uncommon) Methodologies

Unlike the market share of companies and brands, for which there commonly are reliable public and private sources of data available, estimating the share of wallet for customers is much more challenging.  In certain categories, there are data compilers and resellers that track customer spending; some of this comes from survey-based data, some from transactional sources, and some of the information is modeled. For example, in the travel and leisure area, several vendors market overlay data that provides estimated category spending (air travel, hotels, etc.) for consumers.

Such data sources may be valuable in some cases. But in general we have observed that they often to not provide as much insight as hoped. For example, a hotel company may have several brands that service different segments – budget, business, luxury – yet the spending data may not provide sufficient detail on customers' spending within these subcategories. The data may also be heavily modeled, which is not necessarily a bad thing, but this is not always how it is presented. Finally, if such overlays are needed as a permanent data source, then another structural operating expense is introduced.

Another traditional way to develop this data is through quantitative research. We have used this approach many times in the past, and it brings with it all the challenges that research studies must overcome. Survey design and sampling are especially important, because customers often do not have good recall about their activity in a category. For example, if someone asked me to give the number of nights I have stayed in a hotel in the last six months, I could probably provide a reasonable estimate. But the share-of-wallet questions, which  ask me to allocate stays to various brands, would be much more difficult for me to recall and subject to more bias. Nonetheless, a well-designed study can reveal competitive choices that customers are making, as well as the reasons behind these choices, and the resulting data can be used to build a model to estimate share for the entire base. Developing models to accurately score the customer database using research data presents some additional challenges, and in order to keep the models up-to-date, ongoing tracking research needs to be conducted.

A more direct approach, one that we have used successfully, is to use transactional data from your customers to develop a model. While the methodology is a bit involved, and beyond the scope of what we can cover here, the basic premise is first to build a model that estimates spending for your customers. We then apply this model to find customers where the estimates are wrong, on the assumption that these customers must be either giving us a lower or higher share than we expected.  We can then build additional models that estimate share of wallet. This method has been surprisingly robust and validates well against data gathered from other sources.

The summary: make sure that market share estimates at the customer level will help drive business results. If they will, then choose the methodology that yields the best information at the lowest cost.

Posted at 03:43AM Aug 20, 2009 by David King in Analytics  |  Comments[0]