Segmenting Segmentation

The problem with the word "segmentation" is that it has lost its meaning. Like "analytics", a marketer saying "segmentation" can mean virtually anything. This isn't necessarily a problem, if we are simply using "segmentation" to mean "the parsing of customers in some way", but often people mean something very specific when they say the word, while the people listening interpret something very specific as well, but something specifically totally different.

I've often tried to come up with ways to segment the word segmentation. And yes, I am guilty of the same sin the "Two Bobs" made in Office Space when they wrote "plan to plan" on the whiteboard, but I'm still going down this tautological road.

"He's a real straight shooter, that Andy Hasselwander. He segments his segmentations"

So, here we go. In my view, any segmentation starts out with business goals. What is the company itself trying to do with its customers? Are we trying to grow profit? Grow revenue? Are we trying to do it in a specific area? Do we have multiple goals? What are their priorities? Ignoring these questions and just going down the path of segmentation without context is a big problem. So start by understanding business goals. The two Bobs would be proud of you, even just doing this.

Then, we get to the actual source insight. This is before we do any segmentation. What I mean by insight is pulling all of the stuff we do and can know about customers that would make meeting the business goals defined above easier, and putting them in a giant bin. This can mean old and new market research, qualitative or quantitative, data mining, management hypotheses... it's everything.

Only then do we get to segmentation. Now here I think there's an important thing to consider, which is that "value" segmentation kind of supercedes all the others. The reason is that if we want to do anything to customers, we first need to understand how valuable these customers are. So, I'd argue that segmenting customers first by value--whether customer lifetime value or something less sophisticated--has to happen first before you do anything else. This is probably somewhat controversial, among the 150 people who regularly read this blog. So, I encourage people to engage in a heated, uncivil debate on the topic.

Then, I'll put forth four key "types" of segmentation, in addition to value-based:

• Product. What customers want and need from a product or service. Sometimes called "needs-based segmentation"
  
• Creative. How customers respond to different visuals, animation, etc. Some of the cool things going on in biometric research are very useful here.


• Promotion. How customers respond to different types of incentives... price, points, free shipping, etc.


• Channel. Which customers use which channels, and which they prefer.


• Targeting / Finding. Where to actually find customers wherever they are
  

This is represented in my little "segmentation tree" as a continuum, and that's deliberate. This is because the left side of the tree--needs-based segmentation--is really hard to combine with the right side of the tree. Also, making a segmentation scheme do all five things at once is almost impossible. There are two reasons for this:

1. A statistical problem called "assignment" which means that you can't make one segmentation based on needs, generated from a survey, fully predict or "find" individual customers


2. A scope problem that in any one study, you can't possibly combine all elements of "segmentation" into one uber-model. It's just impossible.
   

One way around this is with behavioral segmentation, or "deterministic" segmentation. This means you just start out segmenting your customers into groups 100% defined by their behavior, and then define the six attributes in the tree post-facto. This way, you're always certain you can target a customer. It might not be perfectively descriptive, but it's guaranteed to be perfectly actionable.

Later everyone.

Verticalizing Internet Marketing

The hottest thing in direct marketing today is targeting individuals online vs. targeting via publishers or content. The idea that we can "know" a user and target them frees up the 90% of online inventory currently referred to as "remnant" space-- e.g. the space left over after GM, Microsoft and Coke buy the top page banners at Washington Post and NY Times. The remnant space can be just as valuable as the prime space if we only knew who the users were. A user interested in gourmet cooking is still a user interested in gourmet cooking when she leaves the NY Times food page and goes to her hotmail account or to her son's preschool website.

Verticalizing the internet is a potential solution. Or rather, verticalizing internet users. The internet is already as verticalized as it's ever going to be. So how would one verticalize internet users? First of all, it's only going to happen on specific networks of sites. You can't, for privacy and chinese wall reasons, put a universal cookie a browser that decodes them perfectly everywhere. I guess you could, and maybe that's an interesting business model. But for now, you need to work within the context of a network or a publisher. Let's use Yahoo! as an example.

So, say Yahoo! decides that they really want to make their inventory appealing for technology marketers, specifically B2B technology marketers. One option would be to put up pages that are really appealing to B2B technology buyers and influencers. These pages, assuming they attract a lot of volume (another key issue for networks / publishers), would get high CPM / CPC from technology marketers. However, once the user leaves this site, they're still valuable, right? So here's the rub, and how you can actually verticalize the internet. You've got the site; you've got the interested users. So you're missing two steps: (1) follow them around, and, (2) identifying the same types of users regardless of where they are on the property. I'd call this "200-level" and "300-level" verticalizing.

200-level verticalizing. Following them around is pretty easy. This basically means finding the really good vertical targeting-type pages, tagging engaged customers, and making sure we follow them around. This is pure behavioral targeting. We can get quite fancy doing this, by adjusting content on the verticalized page to correlate to segmentation dimensions and then adjusting people's segments based on what they click on. The trick seems to be in the details on this. For as long as I've been a member of Microsoft "Live", for example, I still seem to only get display ads for Hannah Montana movies. I view this as low-hanging fruit.

300-level verticalizing. The real nugget will be identifying all traffic according to vertical. There are two ways to do this. The first way would be doing traditional market research, such as a quantitative instrument, that would be distributed to a broad cross-section of a network or publisher's traffic. All respondents would be pre-linked to as much "targetable" behavioral data as possible. For example, Microsoft could use all the cookie data it aggregates for Windows Live ID users. Then, an analyst does a latent class analysis to derive segments relevant to advertisers; does a multinomial logit model to assign; and hands it off to the ad sales team. A/B testing to determine incrementality, and you're off to the races.

The second way would be to use only behavioral data to create "mock survey data", specifically from the vertical-relevant sites. So, we'd define our orthogonal dimensions that would be relevant to advertisers, such as "self integrators vs. needs help" and "windows vs. open source" etc. for a B2B technology scenario. Then we'd create content on our verticalized site that reflected these dimensions, and start measuring who clicked on and was otherwise engaged with what. We then use the same exact technique I outlined above to define and target segments.

The promise of these approaches is, I think, pretty huge, specifically for B2B where it's much more important to be much more targeted online.

Build vs. Buy to Land Marketing Analytics

I break analytical marketing into two families: performance optimization and performance improvement. Both families are equally popular these days. However, I'm still seeing significant frustration landing "analytical marketing", particularly in large companies. Smaller "born on the web companies" have no problem at all with this. Generally, the founders of these companies built the core strategy around analytics, hired people comfortable with these concepts, and have woven analytics into the core IT systems of the company as it has grown.

However, for larger "digital immigrant" companies, weaving analytics into marketing can be a daunting task. It's no accident that large global consultancies like IBM are investing heavily in analytics consultants. They've realized a simple truth: that no matter how many fancy new airplanes an airline has, they won't get off the ground without skilled pilots and mechanics. So, why not just book a ticket on United?

I can't take credit for the pilot / mechanic / airplane /airline analogy, but it's a good one so I'll use it shamelessly. To be clear:

• Pilot = analytics marketing professional. A dork like me. There are different kinds of pilots. There are the ones that can only fly the plane, and then there are the ones that can also probably start their own airline.
• Mechanic = data warehouse, SAS server, web analytics etc. software guru. These are the people that actually understand how the plane is put together and fix it when it breaks.
• Airplane = the data warehouse, SAS, etc. software itself, or "cloud", doesn't matter. These are the $100,000 + CAPEX investments that make analytics possible.
• Airline = IBM Global Services etc.--the firms that will just do this stuff for you. You just get on the airplane and it flies you where you want to go.

Looking at it from the company's perspective that just wants to get better at customer insight, targeting, ROMI, web ad optimization, etc., it's a very difficult strategic question: Where should I start? One could make an argument for starting at any of the four bullet points above, and I can think of case examples of companies who have gone down each path. But, I'll give you what I think the answer is now.

• Don't even bother hiring mechanics if you are less than $500 M in revenue, unless you were born on the web. It's extremely expensive; the people that can truly build great systems are few and far between, and the pain will be enormous.
• Everyone who wants to do analytics needs at least one good "senior pilot". All the other stuff might be outsourced, or you might fly United, but you need one person who really understands the stuff and can direct vendors, etc.
• Whether to buy airplanes or not depends on business context. I'm a huge fan of the cloud and renting airplanes, and this is getting more and more feasible. Companies that provide managed ROMI environment that still house your data like MSights are a powerful option. However, for large companies, buying other components selectively can make sense. And it certainly makes sense if you were born on the web.
• Full-service airlines can also make a lot of sense. Totally outsourcing all analytics to IBM might be the way to go if your culture is simply not built around digital and it's not a core focus. Over time, the company may get enough digital natives on board to make analytics a core competency.

"Barrier Removal" Marketing Strategy

"Barrier Removal" is a tactic I've used to great effect many times in strategic situations where customers don't seem to be behaving the way companies want them to. In many cases, to a customer, there are hurdles galore when getting information or making a purchase. To companies, these barriers may not be at all apparent.

Barriers can be physical, rational or emotional. Physical barriers might include:

• Long load times on a web page
• Credit disapprovals at a car dealership
• Small type size in an ad
• Long distances to a dealership

Rational barriers might include:

• Higher prices than competitors
• Feature sets mismatched to needs
• Confusing steps to buy

Emotonal barriers might include:

• Fear of the unknown
• The brand doesn't match my identity
• Distrust of the product / brand

When a marketer adds up all the barriers, you get something similar to a Markov chain. Each barrier can be thought of as "erected" or "non-erected." A non-erect barrier means the probability of a customer passing through that barrier is 1--because the barrier doesn't exist. But an erect barrier will cause some fall-out, and will effect all downstream barriers, too.

Doing a buying analysis based on barriers can be a powerful tool. One gets the sense that a company like Apple has done them in their retail stores, but maybe not for their B2B business. Looking at an Apple store, there are very few barriers.

Physical:

• Everything is in stock
• The stores are close to target populations
• There are products available for demo all the time

Rational:

• Prices are clear and intuitive
• Features meet consumers' needs
• Buying is simple--just talk to an associate and you walk out with your mac / ipod.

Emotional:

• The store is appealing and comforting
• The brand is trusted

However, when you look at B2B, there are barriers galore.

Physical:

• Where does one go to get Mac for business? The store? A distributor?
• What if my battery runs out? Why are the batteries internal? (some insulting me emotion here.)

Rational:

• Apple doesn't work seemlessly with Office. Or does it (some confusion mixed in here.)
• Apple doesn't make servers.
• There isn't enough business software available for Apple.

Emotional:

• I've never thought of the Apple brand as being built for businesses.
• Apple spends all its time marketing iPods.

Anyway, it's a simple example for illustrative purposes. But I love doing these barrier / hurdle projects. Maybe I'll do another post on the simple math... which gets to the upside business case of barrier removal.

Creating a Dynamic View of Customer Status

When trying to manage a population of customers in a high frequency-of-purchase business where unique customer data are obtainable, it can be helpful to look at a grid comparing:

• Frequency of purchase in the last "active" 30 day period, and;
• Last purchase date.
  

This grid will look like this, when the report is run against the current state of customers:

Another way to look at these dimensions is as continuous dimensions where:

• Freq of Purchase {0-->∞}
• Last Purchase {0-->∞}

And each customer is plotted along these dimensions.

But, to make things convenient, we cut off "last purchase", at, say, 90 days, and frequency at, say, 4 X for our table. We can then bucketize these scalar dimensions into a neat 4X4 (or nXn) grid like the one above.

A simple customer count in each bucket lets one understand current state. This is a perfect application for an OLAP reporting tool. It's also interesting, however, to understand dynamics: How customer status has changed from previous periods--the real point of this post.

To do this, I use the concept of vectors. A table such as the following can be created:

We can then simply average all of the movements for each unique starting cell location. For example, for the starting cell 1,1, which would correspond to Frequency of Purchase = 1 and Last Purchase <= 30 days, we might get {1.283, 1.321} for movement. These two numbers are the sides of a right triangle, the hypotenuse of which is the vector of customer movement.

There is an even more precise way to do this, which is to measure actual change in frequency of purchase and last purchase for each customer. In this way, we can actually measure customers who are going "beyond the edges" of our table--for example purchasing more than 4 times per 30 day period.

We can then plot our vectors in our table, which might look like this:

Each vector "direction" has a meaning. For example, "North" means more recent; "East" means more frequent / valuable, and so on.

We can also get to the velocity of change--the second derivative--by simply differencing two vectors, say, a month apart. These would tell you how a dynamic trend like the one above is changing. This is a very useful indicator of how marketing programs are impacting customer status.

This approach can also be used with virtually any customer attribute replacing these two dimensions, for example:

• Total value
• Customer Lifetime Value
• Category-Specific Activity
• Etc.

This is a neat approach that can be done with basically any pivot table / OLAP tool and a database of customer purchases.

Scott Cross (Office Depot) Talks Efficient Customer Response, Customer-Centric Marketing, and ROI

I recently interviewed Scott Cross for an upcoming issue of MarketBridge's client email newsletter (Minds Over Markets). Scott is Director of Strategic Campaigns at Office Depot. Scott's a super smart guy who's tried a lot of innovative things and has a firm grasp on the big picture of B2B Marketing.

---------------------------------------------------

AH: Talk a little about how customer data is being used to bring together vendors, manufacturers, retailers and distributors in a more customer focused strategy, and where that’s headed from a B2B perspective.



SC: The manufacturers don’t really understand who is making the decisions and what information is available within the network of their distributors and retailers. A lot of them aren’t set up to understand the information, and don’t have the capabilities to analyze it. So it’s really a green field, and that offers businesses the greatest upside today in the marketplace.



How can you get that data and marry it with the manufacturer? It’s critical because the manufacturer has the expertise when it comes to their products, customers’ behaviors and reactions to those products, and even customer trends when it comes to the features, benefits and requirements. What distributors and retailers have are customer data -- transactional and market basket data -- that completes the picture. So there is definitely room to grow there. For us, that’s where MarketBridge comes in: bringing the manufacturer together with the retailer. 



AH: How has customer insight evolved at Office Depot over the past few years?



SC: Obviously, the retail sector is important for us, but there has been a big insurgence in B2B. But when you look at our customer base, about 80 to 90 percent of our customers are actually businesses now. So we really started trying to understand what drives their buying behavior, and who the decision makers are. In the past we did a lot of qualitative research. Any type of quantitative research was based on surveys, but we didn’t do a lot with customer analytics, database mining, and market basket analysis on the delivery side of the business with our larger contract customers. Most of the market basket analysis was done with the retail customer data. 


So we’ve taken it to the next level by looking at the market basket information such as buying trends and behaviors, and looking at high value customers. Now much of our primary research is done with business customers rather than consumers. MarketBridge has played a big role in helping us understand that, and doing a lot of the modeling on our customer data set.



AH: As the manager around customer insight, what are the three most valuable reports that you receive or would like to receive? Which ones help you make better decisions about customers?



SC: I think ideally there are three reports managers want to see. First, a daily overall sales report that shows how you’re doing by product category and by channel is important, because every morning I can see whether we had a good day or a bad day and why, and the most effective channels from a product perspective. We’re also working to understand that at a customer level.

The ideal state is overall sales by product category and channel, and the second layer would be to look at that by customer segment. It’s the ability to drill into your sales vs. plan vs. last year, find out which categories are up and down, and in which customer segments. 



The third report would be how we are doing against campaigns: how a campaign is performing by product category, customer segment, and by channel. Then, taking it even further, how a campaign is doing at the sales rep level. It’s really measuring the performance at a high level down to product, channel, customer and sales rep levels.



And the next step would be to get this information in real-time, and I think we’re headed in that direction.

AH: Given that, we often talk about the three dimensions of account structure, product market basket and marketing effectiveness reporting. How much importance do you put on those three when you’re managing your business?



SC: Marketing effectiveness is certainly important. You want to understand what vehicles are working and which customer segments are responding to which vehicles. In today’s economy, knowing which customer segments are doing financially better and have a little more money to invest in our products is important. For example, the public sector is definitely a growth opportunity in the current state of the economy. 



Understanding the effectiveness of marketing to those segments is important, but just as critical is the product marketing basket and the account structure. For example, the product marketing basket allows us to understand that customers are buying ink, toner and paper, but they are no longer buying filing products, because budgets have been cut and they can reuse their file folders. So to understand how the product marketing basket changes by segment, by the cyclical nature of the economy and by the overall macroeconomic environment is important. 



And the account structure is equally important. Knowing the difference between a small single-office customer down the street and a multinational corporation with small branches all over the country is key. At the same time, how do these different accounts make buying decisions? Is it the owner of a small business vs. the administrator within a large company? When you are able to marry those three together, it’s really powerful information to use in effectively marketing to your customers.



AH: Who understands customers better, the marketing organization or the sales force, and why?
SC: The sales force understands the customer better when it comes to interacting and knowing the way they customers think. When it comes down to one-to-one relationships, the sales force is number one. That said, the marketing organization is more effective in taking that information and using it to segment the customers into similar groupings, so you can have the same type of effect as with the one-to-one sales effort. So in looking at the overall customer base, how to segment them and market to them, and also which product or offering is most relevant to the entire set of customers, marketing is the most effective.



AH: Switching gears, you’ve got a market research person on your team and plenty of analytics people that are dealing with your core systems. How can market researchers and data analytics people better integrate to drive customer insight? How do you take those two separate tools and merge them together?



SC: Our philosophy is that research and analytics should be in the same group, because data is just another view of the customers, so if you can understand the customer through market research, you are getting a better picture of what the customer looks like. So market research and analytics go hand in hand.



AH: With your CPG background, you saw efficient customer response take off with retail scanner data in the supermarket sector, for example. Now it’s finally happening for B2B, and Office Depot is a pioneer in that.



SC: I do think we’re on the cutting edge. You always feel like it can’t go fast enough, and you’re always feeling behind when you look at retailers and all the information they’ve shared for many years. But it makes you realize there’s a lot of opportunity.

Sterling Cooper What Happened to You?

As a shameless tag along to last night's Mad Men premiere, I want to develop a hypothesis that I've been kicking around about how agencies, companies, and other elements of the marketing value chain have evolved since 1963.

Back in the 1960s, I've heard from some reputable sources, agencies were much more "do it all marketing companies." The distinctions we make today between a digital agency, an old line agency, an experiential marketing agency, and a marketing strategy firm would be foreign and meaningless back then, because agencies, in many cases, handled all elements of a client's customer facing presence. This was marketing in it purest form. The agency handled research, ideas, creation of message, art, and getting it out to the market. In many cases, agency folks would do things they had no idea how to do, but succeeded anyway. From what I've heard, it was (even removing all the booze, womanizing, and other crap,) a much more exciting time.

I work in a marketing consultancy. We do things that agencies would have done in the 1960s without knowing they were doing them. Did they do those things as well as a specialist would do today? Probably not--but I'm also willing to bet that marketing has lost something to over-specialization and fragmentation. Having all of the people thinking about a brand under one roof has huge benefits. Network effects drove huge creative outbursts and incredible innovation in marketing in the 1960s and 1970s. This is similar to a Google today, at least what I imagine Google to be. You have a company trying to do a whole bunch of cool stuff; specialization has not yet occurred, and innovation happens.

Much more of marketing has been brought in house by companies trying to save money. A lot of this was definitely good, but I think it can also be stagnating. Having people working on your customer facing presence who were working on, say, London Fog last year has benefits. I'm not saying that companies should look to "re outsource" the marketing function, but it's an interesting idea. H1: A company that came along today looking to trade on network effects of multiple creatives handling all elements of a company's customer facing presence... from sales to TV to database to you name it... and really executed on it (and not just said it) might actually be able to make some serious hay.

This is why it would be tough to make a Mad Men today about a company in the "marketing business". If you were going to make a show that good about a company today, it would probably take place at Google. Actually, let's try "company shows" most emblematic of their decade:

• 1960s: Mad Men (Advertising)
• 1980s: L.A. Law (Legal)
• 1990s: Office Space (IT.. ok, it's a movie, but give me a better one)
• 2000s: Project Runway (Fashion)

Seems like a Google show needs to get made.

P.S. As far as Mad Men last night, while initially underwhelmed, I've been thinking about the episode all morning, which probably means it was actually really good. I didn't like the birth flashback at first, but I now think it was really clever. I loved the London Fog sales call, and I guess the thing that struck me there was that Don was essentially out of ideas and looked pretty impotent. The Sal scene with the bellhop was so awkward but great... and finally Pete's behavior as the junior executive we love to hate reached new levels. Lots of great aesthetic pieces too--the Japanese porn, the Stolichnaya as a cuban cigar equivalent, the use of raincoats as metaphor for (staying in the closet; hiding from the past)... By the way I'm headed to Baltimore tonight. Should I be worried?