–

Problem of transportation logistics

MM: Not just trucks, but what’s on the pallet and how many pallets get organized by what truck.

MB: That’s right. And how many stops it takes and so forth.

This brings me back to what we mean by a “Diagonal,” BI application.

To build an application that really helps address the problem of transportation logistics, or the truck shipping of goods, you have to embed a lot of industry understanding and knowledge of trucking into the application. So it requires information specific to the business problem of shipping goods by truck, but it’s not specific to any particular industry.

You don’t really care whether you’re shipping machinery or consumer packaged goods or clothing. These applications cut across industries, but not all industries. Obviously, financial services people aren’t shipping goods around by truck, and for the most part, shipping is just not a part of their primary value proposition. Similarly, higher education is not a truck-oriented industry. But any manufacturing company, whether in the food segment, the clothing segment, the toy segment, the industrial products segment, etc., all have a similar trucking problem to solve.

Another example is any company that makes or sells something that typically has sales margin and profitability issues. The companies really want to understand what products are selling at good profit margins. They want to be assured that the inventory they carry, relative to sales rate, is in balance.

Sales margins and profitability issues cut across industries that have goods to buy and sell—but obviously these aren’t applicable to government or higher education. It’s not like a database system because it doesn’t apply across all industries.

These diagonal types of applications are important because they add high value for their customers. They typically save companies thousands and thousands of dollars all the time, or even millions, for large companies. So they are applications that can command high price points, because they really deliver great savings and a very attractive return.

But also, they’re applications that—because they can be sold across many industries—have a pretty large base of prospective customers—larger than vertical-market applications that are targeting a very narrow perspective. They are very attractive from a business standpoint.

Diagonal applications also work very synergistically with SaaS deployments. That was one of the things that I emphasized in the talk I gave at SaaScon. The reason there are companies like Oco and obviously other new market entrants in this space is because of this synergy.

When you build a system for a particular business problem, transportation logistics, let’s say, then the structure of the database of information that’s needed to support it is not specific to that particular customer. It’s a database that’s designed to support transportation logistics.

As a result, you can get great economy of scale in the deployment of that system by creating a SaaS multi-tenant deployment of that database. All the customers sharing that infrastructure are trying to solve the same kind of transportation and logistics problem against a database of similar structure.

This works a lot better than the ASP models of a decade ago. Back then, custom data warehouses would be designed for each business. If you tried to aggregate those together, you’d get a whole bunch of totally different databases. In some sense, they were too customized. You’re not going to get common behavior by putting them together.

–

–

Decision making versus predictive analysis

MM: Mike—traditionally, again, I come from a background of database marketing, specifically, dealing with really large data stores. The fact is that most relational databases and most business intelligence tools are not really good about drilling down on a what-if heuristic. Then, modeling—if I have these demographic or psychographic factors in my database, how many people does that represent?

The notion of being able to drill down into very specific sets, and almost do a little simulation in terms of the ability to access that group by e-mail, direct mail or whatever? That typically gave rise to specialized analytic databases, to specifically a deal with that kind of ability to drill down.

Could you just give us a quick reprise of both of those database strategies—and then get into the compare-and-contrast of them?

MB: I think the simplest way to understand it is that the relational databases were originally designed around transactional processing—the kinds of things that an ERP system needs to do. They are row-oriented. You’ll have a customer table and a customer buys something. So you put an order into the order table and there’s one row for each order, or probably each line in each order.

The whole approach is organized around the notion that there are entries—which are rows. They’re created in response to transactions. Transaction processing is the primary activity.

Then people started trying to use this to do data warehousing, where the workload is much more analytical, answering questions like ”find all the customers with these characteristics,” and so forth. And this required organizing the data in certain ways to support analysis and decision making versus transaction processing.

At first, people took the databases that were organized around transaction processing, and started trying to use them in different ways—to index them differently and so forth.

Then, many databases that centered around decision support entered the market. Teradata was really the first one. But there have been many since then that have entered with decision-support workloads in mind. There continues to be all kinds of interesting innovations in the database market.

The relational database market is around 30 years old. It should be mature by now, but every year there seems to be new innovations in the relational database space. I’m always astounded that there continues to be new entrants. There’s a whole slugfest among new entrants for who will have the crown of the TPC— the Transaction Processing Council benchmarks. The TPC publishes benchmarks, and they label them Benchmark A, Benchmark B. They’re down to Benchmark H now. This is a decision-support benchmark. It’s really quite a good benchmark, because it measures not only how good the database is, and how fast the database can answer the question—but how expensive it is at doing that based on a cost and a cost-performance type of metric. It’s interesting that these small vendors are continually displacing each other on the top of the heap for that benchmark and it illustrates the continuing innovation that is occurring.

The columnar or column-oriented databases are databases that are organized in order to be able to handle these decision-support workloads optimally. There are quite a few of them on the market now.

But Michael, you asked the question about “what if,” analysis, as well. The columnar databases and the decision-support optimized databases are very good at answering questions like, “find people that have these interesting characteristics.” One needs to feed a very complex SQL query to find them, and these databases can very rapidly extract and produce that result set.

But there’s a big difference between this type of query and predictive behavior addressing questions like, “If this was true, what would happen then?” These kinds of things get you into the world of data mining and predictive modeling. These types of things are—to my knowledge— not yet embedded in the databases.

–

Marketing performance

PvT: You just mentioned analytics. How important is it to integrate marketing and customer data across the organization versus at the local level?

MM: Well, analytics is generally a can of worms that once you open it you never find a can large enough to get all the worms back in.

Analytics has become central and critical to success in the always-on, 24-by-7 integrated, online-offline brand theater.

When we start talking about analytics, we discover that 80-90 percent of the data that a marketer needs does not reside, or exist at all, in their CRM systems.

So many marketing organizations spent the last six to ten years getting organized around what I’ll call tactical CRM – your sales force automation platform. I am astounded how many firms still struggle with CRM as operational capability.

Many firms have separate or loosely connected operational CRM used by the customer service or call center. I am also amazed with the number of these system that contain little more than a transaction record about previous purchases and logged complaints.

A number of firms that have not yet integrated tactical CRM from sales operations with the operational CRM of their call centers and customer interaction centers.

I just chalk that up to the penalties of execution—everyone’s heads down hitting their numbers with little extra time or incentives to innovate something better.

Integration of multiple CRM systems represents a major undertaking for most firms, and it requires developing huge data model by which to specify – in very concrete table-to-table or data-element-to-data-element level—specifically how to transform data into high-level business information that supports specific business decisions.

Most companies that I’ve run across have incomplete or just simply wrong data maintenance procedures in place. So, as a function of that they end up with glorified mailing lists with very little useful analytic data beyond who bought what and why.

One version of the customer truth

Often first major initiative in data integration entails creation of customer master.

While simple in name, the development of a customer master represents a Herculean accomplishing: one-version-of-the-customer-truth.

As this starts by developing a data model of what constitutes a customer relationship—and I stress the relational aspects of the customer and way beyond basic name and address—we often discover that multiple individuals with multiple roles and responsibilities within a single customer object.

In this data-centric view of the world, a household or business entity constitutes the cornerstone of a customer relationship—to which you can associate a number of individual buyers and influencers by context.

Right there, many CRM implementations fall down: they make no meaningful distinction between an account, an contact, and customer object—the business entity or household—that represents the economic context for many buyers, transactions, interactions, and influencers.

So, let’s say we have a customer master—one version of the customer truth expressed in clean, uniform data!

This invokes 90/90 rule which state after you have completed 90 percent of the work (i.e., building your customer master), then you another 90 percent more to complete—the second 90 percent!

That almost always requires the purchase of external enriched data overlays to your customer master.

This will take to you companies such as Acxiom, D&B, Experian, Epsilon, InfoUSA, Merkel, etc.

Enriched data overlays of households might include credit histories and scores, the model and year of cars in the household, names of other members of the household, marital status, plus things like educational levels, current job position, annual income, total credit available and the equivalent of a business profile.

By the way, one of the most interesting developments as it relates to the customer data master, relates to the emergence of an XML standard from business reporting called XBRL (XML Business Reporting Language) that mandates that all public firms must publish their annual reports, 10-Ks, and 10-Qs in explicit 2000-element XML schema. While just a side show for now, XBRL will transform database marketing into true one-to-one engagement. Gosh, we take another hour unpacking that idea. But here’s the seed of a big idea: Every system of record in the next 5 years will adopt XBRL for all its publishing and reporting functions, creating a level of hyper transparency within business operations that will boggle the mind.

Segmenting for profit

So let’s get back to customer masters and enriched data overlays. Now you have the ability to really start to slice, and dice, segmenting customers and markets.

However, you can’t slice and dice your customer database using the relational database or the tools of a CRM system. You can start there. But, soon enough you will need more speed and better visualization.

At this point you need to bring in specialized, analytic databases—wicked fast columnar databases—for plowing through 5 or 50 million customer records with a response time of several seconds; as opposed to using a relational database that might take hours or all night to complete one complex query.

So specialized analytic databases with train-of-thought visualization tools use the enriched overlay data to begin understanding things like price sensitivity, unmet needs, and other sorts of buying criteria within dozens or hundreds of micro-markets—what analysts call consumption cohorts.

This fast-cycle analysis enables a practitioner to think in terms of predicting long-term value of individual or small clusters of customers.

With time and practice, a good analyst can profile the ideal or most profitable customer sets, specifically identify them by name, engagement criteria, and media consumption preferences..

Now, everything we have discussed to this point deals with database analytics. Four more analytic disciplines now come into play: Web analytics, messaging or email analytics, social media analytics and content analytics (or semantic analysis of one’s inventory of content and advertising)

Web analytics, site performance, and customer experience management will continue to evolve into an integrated suite—all good but fairly narrow sets of data.

Closing the loop with messaging

Messaging or email analytics really start to validate with quick call-and-response or probe-and-validate procedures of newsletters and emails specifically targeted to those segments that your predictive modeling identified.

In practical terms, this means that you need to have something far more than just the mail list manager or a newsletter system. You need to have really powerful analytics process driving each newsletter.

A creative and analytics team starts by building newsletters with Lego-blocks of content and data that correspond to a specific set of segmentation and targeting criteria.

So as I send out 15,494 emails to those individuals that I know are interested in Mexican cruises with Salsa dancing lessons, I want also want to see the response level to other recreational ideas, venues, and offers.

This will require that each email embeds personal URLs, sometimes called ‘Purls’, so that each click through takes the recipient to an individualized landing page—built just in time, just for them—that validates the messaging effectiveness or lift and associates that event’s data to a preexisting user database record.

This closes the loop in terms of my analytic profile, engagement criteria, and consumption of the media.

Now, most of the time that kind of closed-loop feedback information remains locked up in the newsletter or messaging system, and very rarely, if at all, comes back into the customer master or the creative teams driving other media creation processes.

So, as Website, database, and messaging analytics come together, guess what happens: Gee, given all these really fresh insights that our multi-channel analytics has developed, how then we inform the strategic communications teams in our agencies and our tactical content teams pushing content into the various websites—brand touchpoints that passively activate engagement as visitors land.

I have met hundreds of executives who struggle with breach: how do we get the advertising, web, direct response, and field marketing teams on the same page, using a common set of analytic insights to create effective engagement? How do make creative briefs more interactive and driven by same-day analytic insights.

Part of the underlying problem, we have discovered, lies in the very structure of what most creative and marketing professionals call content—the process of creating content and the operational capabilities of managing multimodal content.

But, I skipped to the end of my argument about the evolving integration of five analytic disciplines: Web, database, messaging, social media, and content analytics.

Getting social with analytics

So let’s pick up with social media analytics.

How do you use technology to quantify three really important dimensions of the Web 2.0 mediaspace (blogs, tweets, forums, and social networks).

How can you track, in near real-time, the mood of the market, the voice of the customer, and their individual patterns of engagement?

Social media analytics takes you further upstream into the buying process—much further up in the buying process where people are still developing awareness and consideration.

For that you need to have a really effective voice-of-the-customer program coupled with social media monitoring.

A good voice-of-the-customer program entails long-form interviews with 50 to 300 customers a month, transcribing exactly what they said about the process on discovering, considering, buying, using, and disposing (where applicable) a featured product or service—what we call the ‘customer journey’.

Of course we now see powerful new systems coming to market that automatically transcribe call-center interactions with customers—requests for information or service—all social content to feed a voice-of-the-customer content analytics process.

With semantic tagging of voice-of-the-customer content and mapping that against segmentation and engagement profile, something quite amazing emerges: each step of the customer journey.

So as a customer transits from no awareness to awareness, consideration, trial, purchase, commitment, repurchase, loyalty, and advocacy—as they transit customer engagement lifecycle—you will have actual dialog of real interviews with people at each of those stages, and, more powerfully, how they transit each stage of the customer engagement life cycle.

With just a few hundred of long-form interviews, a team will use a text mining engine map keywords and phrases of the voice-of-the-customer content and develop a taxonomy of desire: awareness, consideration, trial, preference, as well as things like dissatisfaction and satisfaction, wow, or disgust.

And as you develop this library, this taxonomy of engagement supports all kinds of goodness, including which AdWords to buy, how to optimize content for search engine discovery, and the structure of engagement.

This taxonomy of engagement also supports what practitioners call the basis of conversation—the details of how your customers talk about themselves, their lives, and what makes a contribution, including your products and services.

This all syncs up with social media analytics, usually the work of social agencies or monitoring services with specialized spidering tools that crawl through the 50 to 100 million blogs and forums and hundreds of millions of social network profiles and billions of tweets.

Social media monitoring then mines them these sources for keywords and phrases that correlate to your markets and competition, generating a dashboard with statistics on awareness, consideration, trial, etcetera, by your various customer segments, and more specifically what your customers are saying about your brand, what it means to be in a relationship at various stages of the brand lifecycle.

The voice-of-the-customer basically mines interviews about how customers talk about being in a relationship with you, and then the social media monitoring tells you how to validate which brand stories connect brands and consumers.

Agile methods for content creation

This all brings to the last analytic discipline in my rant: content and how marketers will have to reengineer their processes of creating content and and manage multimodal content.

First, it starts with the principles of digital asset management: systematic reuse, do it once, get it right up front, tag and classify everything for speed discovery and retrieval, optimize media components for database publishing and content transformation processes, build and use templates, etc.

Second, adopt the principles of agile software development. A bit much to go further here…