Wednesday, April 19, 2017

Here's Why Airlines Treat Customers Poorly

Last week’s passenger-dragging incident at United Airlines left many marketers (and other humans) aghast that any company could purposely assault its own customer. As it happens, airline technology vendor Sabre published a survey of airline executives just before the event. It confirms what you probably suspected: airline managers think differently from other business people.  And not in a good way.


The chief finding of the study is that the executives rated technology as by far their largest obstacle to improving customer experience. This is very unusual: as I wrote in a recent post, most surveys place organizational and measurement issues at the top of the list, with technology much less of an issue. By contrast, the airline executives in the survey– who were about 1/3 from operations, 1/3 from marketing, sales, and service, and 1/3 from other areas including IT and finance – placed human resources in the middle and organizational structure, consensus, and lack of vision at the bottom.  The chart below compares the two sets of answers, matching categories as best I can.



It would be a cheap shot to point out that the low weight given to “lack of vision” actually illustrates airline managers’ lack of vision. Then again, like everyone else who flies, I’ve been on the receiving end of many cheap shots from the airlines. So I’ll say it anyway.

But I’ll also argue that the answers reflect a more objective reality: airlines are immensely complicated machines whose managers are inevitably dominated by operational challenges. This is not an excuse for treating customers poorly but it does explain how easily airline leaders can focus on other concerns. Indeed, when the survey explicitly asked about priorities, 51% rated improving operations as the top priority, compared with just 39% for aligning operations, marketing and IT, and only 35% for building customer loyalty.

There’s a brutal utilitarian logic in this: after all, planes that don’t run on time inconvenience everyone. The study quotes Muhammad Ali Albakri, a former executive vice president at Saudi Arabian Airlines, as saying, “Two aspects generally take precedence when we recover irregular operations [such as bad weather]: namely crew schedules and legality and aircraft serviceability. Passengers’ conveniences and connecting passengers are also taken into consideration, depending on the situation.” In context , it’s clear that by “situation” he means whether the affected passengers are high-revenue customers.

But as you may remember from that college philosophy course, most people reject pure utilitarianism because it ignores the worth of humans as individuals. Even if you believe businesses have no ethical obligations beyond seeking maximum profit, it’s bad practice to be perceived as heartless beasts because customers won’t want to do business with you. So airlines do need to make customer dignity a priority, even at the occasional cost of operational efficiency. Otherwise, as the United incident so clearly illustrates, the brand (and stock price) will suffer.

If you’re a truly world-class cynic, you might argue that airlines are an oligopoly, so customers will fly them regardless of treatment. But it’s interesting to note that the Sabre paper makes several references to government regulations that penalize airlines for late arrivals and long tarmac waits. These factors clearly influence airline behavior. There's even a (pitifully slim) chance that Congress will respond to United's behavior. So the balance between operational efficiency and customer experience isn’t fixed. Airlines will react to political pressures, social media, and even passenger behaviors. The fierce loyalty of customers to airlines that have prioritized customer experience, such as Southwest and Virgin America, should be lesson to the others about what’s possible. That those airlines have had very strong leaders who focused on creating customer-centric cultures highlights the critical importance of “vision” in producing these results.

In short, the operational challenges of the airline industry are extreme but they’re not an excuse for treating customers poorly. Visionary leaders have shown airlines can do better. Non-visionary leaders will follow only when consumers demand better service and citizens demand governments protect them.


Thursday, April 13, 2017

Monetate Adds Machine-Learning Based Real Time Ecommerce Personalization

Monetate is one of the oldest and largest Web testing and personalization vendors, founded in 2008 and now serving more than 350 brands. Its core clients have been mid-to-large ecommerce companies, originally in the U.S. and now also in Europe. I’ve been meaning to write about them for some time but when we finally connected late last year they had a major launch coming this April, so it made sense to hold off a little longer.

That day has come. Monetate last week announced its latest enhancement, a machine-learning-powered “intelligent personalization engine” that supplements its older, rules-based approach. Machine learning by itself isn’t very exciting today: pretty much everybody seems to have it in some form. What makes the launch so important for Monetate is they had to rebuild their system to support the kind of machine learning they’re doing, which is real-time learning that reacts to each visitor’s behaviors as they happen,

Montetate now holds its data in a “key-value store” (meaning, instead of placing data into predefined tables and fields, it stores each piece of information with one or more identifiers that specify its nature). This is a “big data” approach that lets the system add new types of information without creating a new table or field. In practical terms, it means Monetate can give each client a unique data structure, can rapidly add new data types and individual pieces of data, and can maintain a complete, up-to-the-moment profile for each customer. These are all essential for real-time machine learning. (Of course, the system still has some standard events shared by all clients, such as orders and customer service calls. These are needed to allow standard system functions.)

Important as these changes are, the basic operation of Monetate is still the same. First, it builds a database of customer information. Then, it draws on that database to help test and personalize customer experiences.

The database is built using Monetate’s own Javascript tags to capture behavior on the client’s ecommerce site. Users can also add other first- and third-party data through file uploads, by monitoring real-time data streams, or by querying external sources on demand. Monetate stitches together customer identities across sources and devices to create a complete profile. It can also build a product catalog either by scraping product information directly from the Web site or by importing batch files. Customer browsing and purchase behavior are matched against this catalog.

Testing and personalization rely on Monetate’s ability to modify each visitor’s Web experience without changing the underlying Web site. It achieves this magic through the previously-mentioned Javascript tag, which can superimpose Monetate-created components such as hero images, product blocks, and sign-up forms. Users manage this process by creating campaigns, each of which contains a user-specified target audience, actions to take, schedule, and metrics. Users can designate one metric as the campaign goal; this is what the system will target in testing and optimization. They can track additional metrics for reporting purposes.

The campaign audience can be based on Monetate’s 150 standard segments or draw on Web site behaviors, visitor demographics, local weather, imported lists, customer value, or other information derived from the database. Actions can virtually insert new objects on a Web page, or hide or edit existing objects. Users can build content with Monetate’s own tools or import content created in other systems. The content itself is dynamic so it can be personalized for each visitor. Actions can be reused across campaigns and campaigns can contain rules to select different actions in different situations. The new intelligent personalization engine automatically picks the best available content for each customer, drawing on both individual and group behaviors. Users can also embed split or multivariate tests within a campaign. The system will reallocate traffic to better-performing options while the test is running and switch all traffic to the winner when enough information is available.

In other words, this is a very powerful system.  The user interface is also remarkably, well, usable: some training is certainly required but no deep technical skills are needed.

Monetate’s intelligent personalization is currently limited selecting content for Web interactions. The company plans to add product recommendations later this year (finding the best product among thousands is a different challenge from finding the best content among dozens or hundreds). It will add support for other channels next year.

Pricing for Monetate has also changed with the new product. It was previously based on page views but is now based on unique visitors and number of channels. This reflects a desire to stress customer value over individual decisions. Fees start around $100,000 per year for a small to mid-size company.

Wednesday, March 29, 2017

Do CMOs Really Spend More on MarTech Than CIOs? A New Study Says No.

Like many people in the marketing technology industry, I was tickled in 2011 when Gartner predicted that CMOs would soon have bigger tech budgets than CIOs, and even more tickled when Gartner said in 2016 that it had happened.  But my recent pondering of the relationship of marketing and IT departments had me rethinking the question. On an anecdotal level, I’ve never seen or heard of a company where the marketing technology group was anywhere near the size of the IT department. And from a revenue perspective, there’s no way that marketing technology companies make up half the total revenue of the software industry.

But just as I was working myself up for some back-of-the-envelope calculations, the good people at International Data Corporation (IDC) announced a report with authoritative figures on the topic. Actually, the study estimates spending on 20 technologies and 12 corporate functional areas across 16 enterprise industries in eight regions and 53 countries, comparing the amounts funded by IT departments and by business departments. They haven’t published the figures for marketing in particular but did graciously provide them to me with permission to reprint them here. Without further ado, they are:



As you see, marketing technology expenses for 2016 are estimated at $82.3 billion, which is just 6.7% of the $1,235.3 billion for all categories. Slightly more than half of the marketing spend is business-funded, which presumably means it’s spent by CMOs. But that wasn't what Gartner had in mind: they were definitely comparing corporate IT budgets against marketing IT budgets. 

I understand Gartner's logic but I find the IDC figures more plausible. One reference point is the known revenues of martech vendors. Adobe, which may well be the largest, just reported $1.6 billion in 2016 revenue for its marketing cloud (apparently including analytics and advertising products).  Even if there are ten other vendors as large as Adobe, the top ten would have just a 20% share of the $82 billion. It’s hard to imagine the market is really that fragmented, even allowing for expenses that are unrelated to software.

Another reason I prefer the IDC figures is that surveys consistently show that marketing technology is far down the priority list of IT managers.  That wouldn’t be the case if martech spend were equal to all other tech spending combined. Indeed, one of the main reasons that marketers have been eager to take control of their technology has been the neglect, benign or otherwise, shown by corporate IT.

So let's assume the IDC figures are much closer to correct.  Does it matter?  I'd answer it does because understanding the real relationship between martech and other systems is important.  Marketers need to recognize that their systems are a small part of a big picture and can’t work independently of the rest of the company. Yes, marketers should control their internal systems. But the IDC figures show that sales and customer service spend more on tech than marketing. So, when it comes to customer-facing systems, marketers shouldn’t expect other departments to simply adopt marketing systems as a new core.

More likely, all departments will need to coordinate their existing systems with a shared, enterprise-level resource. This suggests that the common core / edge model of marketing systems needs to modified to distinguish an enterprise-wide core from a marketing department core.  In some ways, this isn't a huge change because marketers have always co-existed with enterprise-core systems such as human resources and accounting.  But marketers are much more likely to want control of customer-related systems like the Web site and Customer Data Platform.  In this model, those are also part of the enterprise core.

Of course, having enterprise core systems leads right back to having the IT department manage those systems and ensure that departmental systems are compatible.  IT departments are not necessarily eager to take this on. They have their hands full with things like security, cloud migration, and digital transformation. Nor are enterprise IT teams usually experts at the finer points of customer data management. They’ll certainly need help from marketing and other customer-facing tech teams. But, ultimately, managing the customer experience is a job for the whole enterprise, and enterprise  IT is the logical team to manage enterprise-wide technology.






Thursday, March 23, 2017

Wondering How Customer Data Platforms Relate to Other Marketing Systems? Here's a Picture


I was asked the other day about the distinction between Customer Data Platforms and Journey Orchestration Engines. My immediate answer was “Some CDPs are JOEs and some JOEs are CDPs. A CDP is a JOE if has journey orchestration. A JOE is a CDP if its data is accessible to other systems. Think Venn diagram with two intersecting circles.”  It's not clear the answer helped, but it did get me thinking about clarifying with a Venn diagram.  The diagram I originally had in mind was this one, showing that CDPs unify customer data and make it available, while JOEs unify customer data and select messages. Systems that do all three are both a CDP and a JOE.


On reflection, that’s not the right way to draw a Venn diagram. Each circle should represent one set of traits. So the picture should really look like this:


That's fine, but it seems odd that “unify customer data” has no system associated with it. Is there a type of system that just unifies customer data without making it accessible or selecting messages? Come to think of it, there is.  Systems that just do customer matching used to be called Customer Data Integration but I don’t hear that much any more.  Sometimes people talk about Identity Resolution but mostly it seems that Customer Data Integration has been absorbed by the larger category of Master Data Management (MDM) systems, which integrate all kinds of data. So let’s add MDM as the label for that.    
But why stop there?  Let's see how other systems would fit into the diagram. First to come to mind was marketing automation platforms (MAPs), which also select messages (like a JOE) but don’t build a unified customer database or offer open data access. The diagram with MAPs included looks like this:
The next is a Data Lake. It provides open data access like a CDP, but doesn’t build a unified view of the data.  Adding that to the diagram gives us:


Hmm, what about CRM? In many ways its out there with MAP: another system that selects messages but doesn’t build a unified database. So we need to introduce a new split, of marketer-controlled vs. sales controlled. I'll give control a different color for clarity.  Apologies to CRM people that your circle is so tiny; I'm not suggesting anything about the importance of your systems.

Still thinking about control, Data Management Platforms (DMPs) look a lot like Marketing Automation Platforms: they’re marketer-controlled systems that select messages (sort of) but don’t unify data from all sources or provide open access. So unless we want to further subdivide the marketer controlled space, they share the same location as MAPs.
Since control has its own color, Data Lake and MDM jump out as not having an owner. In fact, they’re both typically owned by corporate IT, so we can easily add that circle.
This raises one more question: is there an IT-controlled equivalent of a CDP?  That would be a system that unifies customer data and provides open access but is owned by IT not marketing.  You betcha.  It might be an Enterprise Data Warehouse (EDW) if that has all the access features of a CDP (high speed, flexibility, etc.). But most EDWs don’t meet that standard. So let’s just call it an Enterprise-controlled CDP, or ECDP, if you’re wild and crazy enough to accept a four letter acronym. You’ll remember there’s some debate about whether marketing or IT should really own the CDP.  This doesn't provide an answer but it does give a clearer picture of the question.
I've summarized this information in a table below.  Still confused?  We just posted a answers to Frequently Asked CDP Questions on the CDP Institute blog.  Maybe that will help.  


Thursday, March 16, 2017

Is MarTech Too Important To Leave To The Marketers?

I’m still pondering the relationship between marketing and IT: what it is, will be, and should be. A few new ingredients have kept the pot boiling:

- a chat with Abhi Yadav, founder of Zylotech, a MIT-bred, artificial intelligence-driven Customer Data Platform and message selection engine.  Those roots made it seem a likely candidate for IT-driven purchases, but Yadav told me his primary buyers are marketing operations staff.  In fact, he hasn’t even run into those marketing technology managers everyone (including me) keeps talking about. On reflection, it makes sense that marketers would be the buyers since Zylotech includes analytical and message selection features only used in marketing.  A system that only did data unification would appeal more to IT as a shared resource. Still, Yaday's comments are one point for the marketer-control team.

- a survey from the Association of National Advertisers that found marketers who control their technology strategy, vendors, and enterprise standards are more likely to have a strong return on martech investment. (The study is only available to ANA members but they gave permission to publish the table below. You can see a public infographic here).  That’s two points for Team MarTech.


- a study by IT staffing and services provider TEKsystems that found senior marketers with more advanced strategy were more likely to control their own technology.  The difference wasn’t terribly pronounced but it’s still the same pattern. MarTech is now ahead 3-0.  (I was actually more impressed that 65% of departments with no strategy were in charge. Yikes!)


So far, the game’s a blow out. Marketing is usually in charge of its technology and does better when it is.  A doubter might question if marketers really make better choices or are just happier when they’re in control. I do suspect that IT people would be less confident that marketers are making optimal decisions. Still, there’s no real reason to doubt that marketers are the best judges of what they need.

But the game’s not over. Let's call in a recent Ad Week article about global tech consultancies buying marketing agencies. The article cites Accenture, Deloitte, IBM, KPMG, McKinsey and PricewaterhouseCoopers and notes that each already has huge agency operations.  To the extent that these firms are working with marketing departments, it’s still more evidence of marketing being in charge. But the real story, at least as I read it, is these firms are getting involved because they see a need to integrate marketing technology with over-all corporate technology, just as marketing strategy needs to support corporate strategy.

“The consultants’ bread and butter has traditionally been large IT and business-transformation projects,” says Julie Langley, a partner at fundraising, merger and acquisitions advisor Results International, in the article. “But, increasingly, these types of projects have ‘customer experience’ at their center.”

To me, this is the key. As every aspect of customer experience becomes technology-driven, technology must be integrated across the corporation to deliver a satisfactory experience. Marketing may be the captain, but it’s still part of a larger team. If marketing can be a true team player, it gets to call the plays. But if marketing is selfish, then a coach needs to step in for the good of the whole.

I’ll spare you the extended sports analogy. In concrete terms, if marketing picks systems that only meet marketing needs, then the integrated customer experience will suffer. Worse still, some new tech-driven offerings may be impossible. This could be fatal if other, nimbler competitors deliver them instead. Tech-based disruption is a real threat in many industries. Companies can’t just hope that each department working on its own will yield an optimal solution for the business as a whole.  In fact, they can be quite sure it won't.

That’s why I’m not convinced by surveys showing marketers are happier or get better return on investment when they control their own technology. It’s possible for that to be true and for the corporate to miss larger opportunities that require cooperation across departments. If marketing can take that broader perspective, there’s no problem. If it can’t, IT or another department with enterprise-wide perspective will need to enter the game.

Tuesday, March 14, 2017

CrossEngage Orchestrates Customer Journeys Using Events

It feels like forever since I first wrote about Journey Orchestration Engines (JOEs), although it is just one year. Orchestration was already a hot term when I started, so I take neither credit nor blame for its continued popularity. I will say that I’ve now seen enough orchestration systems to start making subtle distinctions among them.

Subtle distinctions are needed because the systems are basically similar. They all ingest data from multiple sources; convert it into unified customer profiles; apply rules and analytics to find the best message for each customer in each situation; and, send those messages to external systems for delivery. Unified customer profiles make these products look like Customer Data Platforms. JOEs that expose their profiles for external access really are CDPs; JOEs that keep the profiles for their own use, are not. In theory, a JOE could connect to an external customer database rather than building its own, but I haven’t seen that configuration in practice.

The main ways that JOEs differ include:
  • Channel scope. Some systems are largely limited to online interactions, while others are built to combine online and offline channels. Some systems that look like JOEs work with only Web or email. But orchestration pretty much implies multiple channels so I’d probably exclude those from the JOE tribe.
  • Decision methods. JOEs can work with conventional, rule-driven campaign structures or use automated techniques to customize the path followed by each customer. There’s also considerable variation in exactly what gets automated: some automate campaign assignments but use static content; some automatically run a/b tests and pick the winners; some automatically create customer segments that receive different content; some use machine learning to dynamically generate custom content. 
  • Journey framework. My original definition of JOE was quite rigorous: journey orchestration meant all campaigns were defined relative to a master model of the customer journey. This really means that stages in the journey are “states” that customers flow between, and campaigns are chosen in part based on each customer’s current state. I still think of JOEs that way and definitely see some systems organized along those lines. But when you start looking at some of the more automated decision methods, it’s harder to apply concepts of fixed states or journey flows. So I still check whether a system has a journey framework but don’t necessarily require a JOE to use it. I realize this means you could have a journey orchestration system without journeys. If that’s the silliest thing you’ve been asked to accept recently, you haven’t been watching the news.
This is all a very long-winded introduction to CrossEngage, a Berlin-based firm that released its product about six months ago. CrossEngage works in online channels, using its own tags to capture Web interactions and API connections to ingest data from email providers, mobile apps, and other sources. It can also load CSV files if necessary.

CrossEngage treats most data as either a customer attribute or event, using big data technologies that store inputs and to allow data access with minimal schema design. The system also stores some information that’s neither attribute nor event, such as products and locations. The vendor maps new sources into the system and can define logic to create custom events. (A self-service event builder is planned by July.) Customer data from different sources is stitched together using deterministic matching only (that is, CrossEngage will only connect different identifiers to the same person if an external source provides the relationship).

A dashboard lets users see Web site events as they stream into the system. Users can apply filters to see only certain events. Campaigns also make heavy use of events, referencing them as entry and exclusion conditions, in combination with user-defined segments; as campaign goals (which may be one or several events); and, as campaign steps (each step being a different event). Event definitions can reference other events and can include brain-bending logic such as checking whether a second train fare request specified the same departure city as the first request and happened within ten minutes. In that example, the first request would be first event in the campaign. This is tremendously powerful and, as the vendor points out with some understatement, poses a substantial technical challenge to do in real time.

Each event in a campaign can be assigned a message, which will be delivered by an external system such as an email vendor. CrossEngage can map its data to delivery systems so they can use the data in their own message templates. Alternatively, messages can be created in CrossEngage’s own templates, which can include conditional scripts for dynamic content generation. The system has external integrations for email, direct mail, mobile push, text messages, and Facebook Customer Audiences, with more on the way. It has its own connectors for Web site and Web browser messages, Web hooks, and file extracts. Users can also attach discount coupons to messages.

Campaigns can be assigned frequency caps that limit the number of messages each person receives in different time periods (per minute, per hour, per day, per week, or per month). Caps are defined separately for each campaign. Another set of caps applies across all campaigns on a per channel basis. Campaigns that generate transactional messages can be exempted from the frequency caps to ensure their messages are always sent. People can also be excluded from campaigns based on whether they were recently in that same campaign or a different one.

CrossEngage also has user journeys, which involve a set of related events. Journeys can exist within a campaign or be used outside a campaign to analyze customer behavior. If you’re keeping track, this is a different use of the term “journey” from the one I described earlier.


This is a pretty mature set of features, especially for such a young system. But nuances also include noticing what CrossEngage doesn’t do. There is no machine learning to recommend the right campaign, right message, or right message timing, although the system does support a/b tests. There’s also no visual flow chart to lay out campaigns. This is a choice made by CrossEngage based on its designers’ previous experience that flow charts quickly become too complicated to be understood or maintained over time.

Speaking of nuance, CrossEngage also has a mature approach to user rights management, allowing administrators to specify which users can perform which actions on each object. Team-based rights are on the roadmap.

CrossEngage currently has about fifteen major clients, spread across travel, ecommerce, fashion, dataing, and other industries. Pricing is based on the number of events tracked in the system, not the number of messages sent. It starts as low as $2,500 per month although average client pays about twice that. 

Monday, March 13, 2017

Forecast: Self-Assembling Application Bundles Will Manage Customer Experience

I recently described a Deloitte paper on technology trends, focusing on their descriptions of IT management methods. The paper also covered broader trends including:
  • Unstructured data, which they saw as a potentially bottomless source of insight. What’s interesting is they didn’t suggest many operational uses for it.  By contrast, traditional corporate data management is almost exclusively about business operations.
  • Machine intelligence, which they described as broader than artificial intelligence. They saw deployment moving from offering insights, to interacting with people, to acting autonomously. They also described it as controlling internal business processes as well as customer interactions. That's not the way marketers tend to think but they're right: the bulk of company processes are not customer-facing.
  • Mixed reality, which is a combination of virtual reality, augmented reality, and Internet of Things. They focused less on game-like immersive experiences than on new types of interfaces, such as gesture- and voice-based, and on remote experiences such as collaborative work. They also listed some requirements that aren't usually part of this discussion, including machines that can understand human expressions and emotions and security to ensure hackers don’t falsify identities or inject harmful elements into the remote experience (such as, telling you to make a repair incorrectly).
  • Blockchain, which they presented as mostly in terms of easing security issues by verifying identities and allowing for selective sharing of information.
Those are intriguing thoughts but don't present a specific vision of the future. A recent paper from Juniper Networks rushes in where Deloitte fears to tread.

Juniper's term is "digital cohesion", which they desfine as "an era in which multiple applications self-assemble to provide autonomous and predictive services that continually adapt to personal behaviors.”  It somewhat resembles the ideas I offered in this post about RoseColoredGlasses.me
and further elaborated here.  I guess that’s why I like it.

Beyond having excellent taste to agree me, Juniper fills in quite a few details about how this will happen. Key points include:
  • Disruptive competitors can use high speed networks, local sensor data, and centralized cloud processing to offer new services with compelling economics (e.g. Airbnb vs. Hilton).
  • Smartphones provide pre-built mass distribution, removing a traditional barrier to entry by disruptive competitors.
  • Consumers are increasingly open to trying new things, having been trained to do so and seen benefits from previous new things.
  • Natural interfaces will eliminate learning curves as systems adopt to users rather than the other way around, removing another barrier to adoption.
  • Autonomous services will self-initiate based on observing past behavior and current context.  Users won't need to purposely select them.  More barriers down.
  • Services will be bundled into mega-services, simplifying user choice.
  • Open APIs and interoperability will make it easy to add new services to the bundles. This is a key enabling technology.
  • Better security and trust are essential for users to grant device access and share information with new services.
  • Business relationships need to be worked out between the individual services and the mega-bundles.
I’m sure you see the overlap between the Deloitte and Juniper pieces. Machine intelligence and insights from unstructured data will be critical in building services smart enough to make the right choices. Machine intelligence will also create an underlying infrastructure that’s elastic and powerful enough to deliver services reliably regardless of user location or aggregate demand. Mixed reality will be key for gathering information as well as delivering interactive user experiences. Loosely coupled systems and disaggregated services will make it easy to inject new services into a bundle. Blockchain could play a critical role in solving the security and trust issues.

I’m also sure you see how this relates to ideas that neither vendor mentioned directly, such as the increasing value of rich customer data, importance of accurate identity resolution, role of brands in creating trust, and natural tendency of consumers to do everything through a single mega-service.

Of course, there’s no guarantee this vision will come to pass. But it’s an interesting working hypothesis to shape your thinking. More than anything else, it should help you look beyond optimizing your current stack to ensure that you’ll be able to adapt if radical changes are needed.