Mind the Apps

GPS was originally designed to deliver missiles and military payloads to their targets. It proved to be highly effective for this purpose, and was later modified and released for civilian commercial purposes – navigation, finding fellow travelers in the theme park or ski resort, and more recently, for routing us effectively through traffic.

Several years ago, the GPS technology that was developed to deliver agents of destruction to military targets was repurposed to deliver agents of marketing to new targets – consumers.

That means us. We are now the targets.

Our smartphones (are there any other kind of phones these days?) and our tablets all have GPS capability. When you turn on the location services on your device, you enable many of the apps on your phone to know where you are. This awareness of your geo-location provides marketers with a significant opportunity to provide consumers with offers, promotions, and services specific to the context of their environment.

Your phone or tablet may also provide a continuous stream of data that helps shape marketers perception of what you may be interested in so that they may offer you something of interest.  In a recent article, Thomas Davenport in the WSJ  (subscription required) described the many apps that monitor his daily behavior. These apps and associated services then send info and offers to him based upon the stream of data that these apps provide.

If your Social Media and smartphone app log-ins are tied to your Facebook, LinkedIn, or Google accounts, then the marketing profile about you just became much richer.

Add to this landscape of information the cookie trackers that are ubiquitous on your desktop or laptop, and you have a combination of three significant streams of data – Social, Local, and Mobile, known to marketers as SoLoMo. To marketers, the convergence of this data provides a window into your consumer psyche, which marketers often define via a practice called segmentation.  There are too many segmentation methods for us to get into for this post, but let’s just say that nearly all of them require the application of advanced analytics and modeling techniques to be successful.

For hospitality, travel, gaming and leisure marketers, the possibilities are limited only by the imagination:

Gaming – Player’s club member who lives in Kentucky is discovered to have arrived in Las Vegas, but no record of a reservation or casino host interaction appears in the casino system for the customer’s arrival during the next three days. Should the casino ping its customer via a special offer designed to bring the customer over to its slot machines, or is the casino prepared to let another casino in Vegas host the customer’s $5k credit line? Exactly what offer should the casino make to attract this customer? What is the optimal customer reinvestment strategy for this scenario?

Hotels – most large chains have decent mobile apps for loyalty members, but few provide customer-centric offers tied to a customer’s location. Simple matter to check centralized reservations records against a customer’s location and make a last-minute offer to fill an unsold room. Today, no major hotel group does this, so instead this app-only company has grown from spunky start-up to well-funded early growth company with a large collection of hotels that use its service to fill unsold rooms. And they do fill at lot of rooms.

Retailers are already using convergence marketing to tailor offers to customers. This technology offers one method for in-store offers, and many others are available or under development.

For all convergence and individualized offer strategies, it is necessary to employ advanced analytics.  The analytic models are either embedded in off-the-shelf tools or can be developed and customized for each company and its specific market niche.

Unlike off-the-shelf tools, customized models become the proprietary IP of the retailer, hotelier, casino, or travel company, and these models can be operationalized into virtually any IT infrastructure or human processes – think front desk, casino hosts, or Customer Service or Call Centers.

The proliferation of well-designed apps enables the delivery of highly targeted offers to customers and the continuous refinement of marketing offers to accommodate changing customer needs and buying behavior. In rapid time, offers can actually be individualized to specific customer preferences. This is trendspotting, in real-time, that supplements or replaces panels, surveys, and those godawful focus groups.

The trade-off in all of this? Privacy. Anonymity. And the value of privacy vs. the convenience or economic benefits of personalized offers that may be of interest to consumers.

So mind your apps. They are most certainly minding you.

Lipstick on a Pig – Avon blows $125MM on a failed ERP

$125,000,000 is a lot of lipstick. The last few weeks could not have been fun for Avon’s IT team. But I suspect the real blame lies with a failure of senior management to use proven Analytics techniques to help Sales force leadership, the CIO and the IT team determine the viability of this initiative.

Last week the Wall Street Journal reported that Avon would take a $125MM charge for the failed roll-out of a new order management solution provided by SAP (registration required). Gigaom also has its take on the situation.

At Avon, the field representative are independent contractors who sell products and recruit downline sales reps. These Avon representatives have to be stroked, motivated, and inspired to sell lipstick, beauty products, and gifts. You cannot impose anything on them – they have to be persuaded, convinced, and sold on an idea or a product before they will take action.

In many respects the field users of Avon’s IT apps and services should be viewed as customers to be cultivated via marketing efforts rather than compelled, instructed or trained in a classic hierarchical implementation model. Somebody, somewhere at Avon needed to do an analysis of what the Avon reps used, what they needed, and create some grassroots enthusiasm for a new ‘product’ that might help the Avon reps sell, order, track and fulfill their orders.

Avon is data rich. Surely Avon knew, or could determine, which devices, software, apps, and productivity tools were most used by their hundreds of thousands of North American Avon reps. They could have, or should have, determined where there was a gap in services and then filled that gap with a technology that would inspire them to opt-in rather than incite them to quit Avon.

There are multiple analytics techniques that can be applied to solve for customer need, propensity for change, and adoption rate projections. In Avon’s case, a $200k-$300k expenditure on the appropriate analytics might have saved them $125MM in quarterly charges.

That’s a lot of lipstick.

Even if SAP’s product was flawless, inspiring Avon’s field reps to use new technology would require many million of dollars more in marketing efforts and development of network effects within the assertive and independent culture of Avon’s reps. The demographic distribution of Avon field reps is varied – think of your grandmother confronting a new remote control or a 20-something socially networked rep who is fluent in smartphone and accustomed to quick, simple and easy apps.

BYOD is here to stay – and the hundreds of thousands of users of Avon’s IT should’ve been polled, assessed via focus groups and Conjoint analysis, and then beta-tested this “productivity tool” before committing $125M to a new ordering and fulfillment platform.

Predictive Analytics vs. Prescriptive Analytics

Grappling with incomplete data? Trying to forecast future customer activity, future demand, or score your customers for certain levels of treatment for your Loyalty program?

The team at Decide, a Spanish consulting and analytics firm, have made it easy to understand the difference – http://bit.ly/18CQxUE

Put another way, let’s say you are driving a curving highway on a foggy night. You can drive slowly and safely without “perfect” information; your view is partially obscured and you cannot see everything, but you can infer the condition of the road ahead based on your experience and your brain’s rapid estimation of the absent information. You can also determine the relative speed, distance, and rate of change (acceleration/deceleration) of the other cars around you. As the fog lifts, you can then drive with greater confidence, at faster speeds, and with greater confidence. Why? Because as the fog lifts, the increase in useful information has enabled you to make smarter, better driving decisions.

Congratulations! You have just used a form of Predictive Analytics.

Now let’s say you are a pilot flying an aircraft loaded with passengers on a dark and foggy night. The safety of your passengers and crew is paramount. You cannot see a thing – you must rely on your instruments. Your instruments have been tested, verified, and tested again, as have the instruments of the other pilots in the sky around you. The relative speed, distance, and rate of change (acceleration/deceleration) is further complicated by varying weather conditions, but nevertheless you can safely determine your altitude, ground speed, rate of descent, and position of all of the nearby aircraft. You can do this, and land your aircraft safely, because your instruments and the data flowing through all of these complex systems, in concert, are assisting you with the hundreds of decisions required to fly the aircraft – even though no visual data is available to you via your own eyeballs.

You then land the aircraft safely, smoothly, as the ground and the runway lights appear below you.

Congratulations! You have just used Prescriptive Analytics.

Q: Was it the you – the pilot – or was it the sophisticated autopilot technology that landed the aircraft?

A: It was both. The autopilot still has work to do, to communicate and set a course and make decisions based on the available human and systems information. Meanwhile, all of the instruments and technology have to operate, and the pilot has to trust the technology. Training, trust, and reliance on the models behind the technology are required to fly the aircraft safely.

In the case of the aircraft, smarter decisions were enabled by predictive analytics.

The choice of which analytics framework to employ – Predictive and/or Prescriptive – is determined by the outcome you wish to achieve and the data and tools available to you. The same decisions about techniques must be made by organizations wishing to take advantage of today’s advanced analytics technologies, available data and analytics resources, both human and machine.

Of course, if this ever catches on, we’ll all be able to take our hands off the wheel.Drive safely, or have a good flight, whichever the case may be.

Value is the Biggest V – Making Big Data Solutions work for Hospitality

AbsolutData’s Tim Brooks talks about predictive analytics — and how it’s improving revenues and customer experience in hospitality right now

Summary

What: We’ve moved from “lagging measures that describe where we were” to “extracted insights and prescribed action”
 that impact future outcomes.

How: Understanding data in context—and exploring huge quantities of data where it resides—provides powerful new insights that enhance real-time decisions about customer treatment, channel management, marketing spend, sales effectiveness and pricing.

Tim Brooks has heard it repeatedly from hotel execs, asset managers, GM’s, and marketing leaders wrestling with the explosive growth of business data: “I know there must be value in there, but how do I get it out?”
 
Brooks has answers for hospitality leaders.

Leading AbsolutData’s enterprise solutions group in hospitality, his focus on advanced analytics strategies lets him help enterprises obtain insights from data that has gone untapped in their repositories
“The strategic goal of many hospitality and travel enterprises today is to get smart enough to apply new analytics technology in ways that create substantial value,” he says.  “Many hospitality companies want to optimize their human and technology resources so that they can connect data into their decision structures and create continual streams of intelligence that can guide strategy and frontline management decisions.”
 
We recently talked with the AbsolutData senior executive about Big Data and predictive analytics across different industries, and the conversation turned to hospitality and travel. Brooks expressed a keen interest in this business and its Big Data ecosystem, as he spent the first 20 years of his career in sales, marketing, asset management, software development and revenue management roles with luxury hotel companies Ritz-Carlton, Four Seasons, Le Meridien, and Rafael Group, a European hotelier acquired by Mandarin Oriental. During his hospitality career, Brooks opened 4 luxury hotels, led brand transformations for two others, and later founded two companies that he grew to manage more than $500MM in hotel revenue. We learned that he also developed patents and mobile apps for online transactional solutions used by hotels and their customers.

Here, the conversation touches on current approaches to data analytics and extracting value from insights, including what hospitality or gaming CEO’s or CMO’s should be thinking about when choosing tools or developing a sustainable analytics capability.

Q: All this talk about applying advanced analytics to make sense of huge volumes of data—what should hospitality leaders be looking for?

Tim Brooks: It used to be that when we talked with hotels, it was all about storing and reporting: “ ’How do I store all this data cheaply?’ or ‘I am inundated with reports but they don’t tell me anything about next week or next month, much less next quarter. What good is it?’ ” Today, it’s more about: “I’ve got all this customer-created data, channel data – there are hundreds of channels – competitor data, dynamic pricing data. I know there’s great value in there, but how do I get it out?”

The challenge is twofold – much of this information is unstructured and most of it moves at very high velocity. The pace at which customers discover, research, and buy hotels and cruises creates a tremendous amount of valuable information. Some of it is online and it’s trackable – even Social Media feeds and travel review sites are trackable via multiple tools. Some of it is offline or captured via call centers or property interaction, so frequently the data is unstructured or what we call semi-structured. One method of analyzing this data flow is to hire a busload of people to sift through, read, and analyze everything.  At the end of this project, you might know where you were and what happened, you may even now why. But given the 3 V’s that apply to the massive river of data – Volume, Velocity, and Variety – it is highly impractical to do that, and it’s costly, too.

 

Q: We’re familiar with the 3 V’s of Big Data. Are they a challenge?

TB: Three years ago, yes. Today – not so much. Those three factors are input factors, and now there are tools, technology and resources that can handle anything thrown at us. Currently we are tackling a project that evaluates 1.5 trillion bits of data generated by machine sensors that transmit 10,000 different log measures every second, then adds human performance data and ecosystem data at regular intervals. It’s not easy, by any means, but it’s doable and there are tools that we used to set up the ETL for this solution.

For hospitality, we are more interested in what we call the 3V’s of Insights – Veracity, Viability, and Value. These are outputs or results of the analytics. Veracity – can we trust the data? Are we using the proper modeling methods and asking the right questions? This calls for domain expertise in travel, hospitality, or gaming.

Viability – is the information important and are the resulting insights actionable? By whom? In what context? Is action required in real-time or near real-time – people sometimes call this fast-twitch data – or should the insights be considered or tested longitudinally? We can call this slow-twitch data.

The last of the output V’s is Value. What economic difference do these insights create for the company? How do they impact revenues, margins, service levels, customer retention or strategic advantage?

Q: Do you have a tool that provides these insights?

TB: No, we create customized solutions as a service. Some of these solutions are nearly customer-ready off-the-shelf, we have done them so many times, but they do require configuration or tailoring to client needs.

An added level of complexity is that the travel and hospitality ecosystem is deeply intertwined and highly reflexive. Consumers have a multitude of choices, channels, and convenient access to a wealth of information. The information advantage lies with hospitality customers, so travel and hospitality strategists need to optimize each customer touch point to encourage customer stickiness, engagement, and a smooth path-to-purchase.

Computers and data ingestion tools, combined with powerful analytics can now help do this work for us. It’s now possible to infer and validate patterns and automatically see nuances in unstructured data, the way a human would. These technological and analytical capabilities now make it possible so that senior execs and frontline managers can understand information in context and in time to do something about it. The technology and the analytics help us to rapidly sift through the noise to find the precursor signals that correlate with or ‘cause’ a future action. An additional benefit is that these capabilities frequently identify new opportunities or answer questions we might not previously have thought to ask. Hospitality marketers are realizing that this is not only possible, it’s happening right now.  We see real examples. Let’s say you want to look at what customers are saying on YouTube, Twitter, TripAdvisor, and in customer survey responses, and then correlate that with geospatial data from iOS or Android apps. You can get insight from stuff like this in real time, and you can act on it in time to make a difference.

Q: What do you mean by ‘understanding data in context?

TB: Let’s look at an example. If we want to look at the performance of mobile apps on tablets or phones and how they correlate with guest usage, we can now provide a geo-location and behavioral model for customer usage. Some customers research via an app or via an OTA, but book directly online at a Brand.com website. Others use call centers or an agent to make their purchase. When we then add their SM data, guest loyalty data and transactional data, we can micro-segment customers to refine our future offers, pricing, or service enhancements to build customer lifetime value. We can even give them easy ways to become our brand ambassadors and help our happy guests help us market to other guests.

We did this analysis recently for one major hotel company, and among the many insights that emerged was that “less is more” when it comes to offers, campaigns and promotional campaigns. The customer micro-segmentation, developed in the context of all of the other customer touchpoints, enabled the hotel company to achieve greater customer satisfaction and more customer reservations activity through fewer offers. Instead of a brand-centric campaign or offer, this new capability empowers their hotels to offer each customer the sort of offer that most appeals to them – almost on an individual level.

 

Q: There is a lot of hype about the volume of data that large enterprises struggle with today, also the incredible variety of types and sources. How much of a challenge is this?

TB: Unstructured data provides a vast haystack of human information, and it is growing at a phenomenal rate. However, to most humans as well as to most computers it’s just noise. Most hospitality and travel IT systems just weren’t designed to deal with things like tweets, video, email text and attachments, audio, geo-spatial data, or image info like Pinterest. Our brains regularly sort out and make meaning of the complexity of slang, irony, sentiment, and new patterns and relationships. We do this rapidly and without pausing to think about it. But this becomes far more challenging for us when learning a foreign language, for example. Likewise, absent a powerful analytics capability that can understand all types of data and which can adapt to that data as the meaning changes, our only choice is to go back and start over with the data warehouse and find a way to include the new data sources.

That’s a costly and temporary fix. It’s untenable. New sources of unstructured data emerge every year. Instagram and Pinterest did not exist until a few years ago, and we don’t know what new sources will emerge in the coming 2-3 years. Right now there are millions of new apps under development globally and the data from some of them are going to become very valuable to hospitality marketers – we just don’t yet know which ones will make a difference.

 

Q: So how are leaders from other industries approaching this hurdle?

TB: You can look at the success of Amazon, Netflix, and many financial services companies as validation of their investment in predictive analytics. We need look no further than the online travel marketplace to recognize winners and losers. OTA’s continue to capture market share and customer wallet. One of the reasons why is that historically they have done a far better job of leveraging data to win and retain customers. They were Innovators in Big Data and first movers in realizing the value of Big Data for travel and hospitality.

The hotel industry is now at the stage where Early Adopters are also seeing significant ROI. A handful of cruise lines, too.

What these companies have learned is that it’s important is to separate the relevant signals from all of the noise. You don’t need to capture and store all the unstructured data. First what you need to do is identify the data sources, patterns, and insights that impact your business. Some of the relevant data, and in many cases most of the important data, is external to the hotel, casino, cruise line, or travel organization. This data often resides outside the walls of the business but can be captured to enrich the signals. Then you need to test, learn and validate the effectiveness of those signals that strongly correlate with future business performance. This is a Big Data and analytics challenge, and it’s highly solvable.

The greater challenge is in extracting real business impact from those signals. This calls for domain knowledge, expertise in cross-functional best practices that can be applied to each company, and transformation to a culture that becomes data-driven.

There is no substitute for experience. Big Data and analytics are no panacea. But it’s useful to think of Big Data and analytics as what I call “Insights Solutions.” Predictive analytics should serve as a prosthetic for the mind, to help decision makers make smarter decisions more quickly and with greater confidence.

Q: What does this mean for the day-to-day operations for hotels, cruise lines, and casinos? How does the CEO or CMO change internal thinking and processes?

TB:  Think about all the places in your business where you are forecasting future outcomes. Demand forecasting can be improved, customer engagement, loyalty, CLV and resulting reinvestment in existing customers and new customer acquisition efforts. The entire travel industry has a large opportunity to make progress with abandoned carts, and retargeting is not the only answer.

Marketing is where lots of opportunity can be readily seized to get quick and visible wins. Simply getting a better understanding of the ROI from multiple channels, the incrementality effects of campaigns and promotions, the real value of affinity relationships, and local market pricing reflexivity vs. brand premium – these are all areas that hotels, cruise lines, and casinos are looking at.

Other opportunities to apply decision science or insights solutions would be areas where you are currently throwing lots of people or outside consultants at problems to get the answers you need. Let me give you an example – until 5-6 years ago, hotels did not have the position now known as “lead-catcher.” This is a sales function in which a full-time employee at a hotel or small cluster level receives eRFP’s for groups or events, then logs each eRFP into the hotel CRM, distributes it to an actual salesperson for follow-up, and responds to each customer eRFP with a summary of hotel availability and rates. Problem is that eRFP’s are often sent indiscriminately by event planners to dozens of hotels in multiple destinations as a means of shopping. Hotels that we’ve talked to reported sales conversion rates of less than 1.5% of all eRFP’s received – in some cases hundreds received each week. That’s a 98.5% failure rate. In many cases the annual cost of employing a lead-catcher at hotels exceeds the profits from the conventions that are actually booked. But because hotels are in the service industry, their standards require that each eRFP must be responded to within 24 hours of receipt.

Using hotel data combined with external unstructured data, predictive analytics techniques, and validation over time, a solution was created that effectively performed triage on each eRFP and scored it before it was ever reviewed by lead catchers or salespeople. Essentially the solution reads the eRFP and tells the hotel if it’s highly valuable and should be acted on immediately by a salesperson, if it’s a gray area eRFP and requires further evaluation, or if it is irrelevant for that hotel, in which case the solution provides an auto-response to the customer which may also include suggesting sister properties or new event dates.

When it encounters a gray area eRFP, it “learns” from prior experience and refines itself. So this solution increases sales conversions, decreases staffing costs, relieves sales frustration, and provides immediate customer response to each eRFP. All for less than the cost of 6 lead-catcher positions.

So now an analytics engine is reading, evaluating and summarizing what it finds in tens of thousands of documents against millions of variables before you can ask any question at all—and it does this very fast. Salespeople don’t need to guess at the veracity of the eRFP, because the data is now talking to you via a scored eRFP.

There are other areas where machines can enhance human experience, making customer responses more informed, marketers more effectives and companies more agile.

Q: That’s great. Can you give us a few more examples?

TB: Of course. We now have the ability to train high-speed computers to categorize data as humans would, based on context and concepts. Take retail marketing, which has significant crossover application to hospitality. We are now using technology that quickly and accurately segments customers based on their open source social media reviews, preferences and conversations and we can analyze that content—in addition to the action of buying shoes, abandoning an online shopping cart or clicking a “like” button.

In gaming, the real-time data from a slot machine session-in-progress can be used to determine the next game to propose to the customer, or to optimize a casino’s reinvestment in that customer – an offer of a comp room vs. show tickets vs. buffet. Converged with mobile apps, a hotel or casino can prompt a traveling customer to visit a local property for dining or gaming, or offer a last-minute special to the customer to book their stay in the area.

Q: What about other travel sectors?

TB: A cruise line can determine the optimal mix of pre-board and onboard offers to make to its passengers, thereby enriching the guest experience and prompting rebooking or redemption of benefits on a future cruise. These are not experiments in predictive analytics, these are real business solutions deployed and creating value right now.

For travel distribution channels and destination marketers like CVB’s there are significant opportunities. Most of these organizations do some BI and report on lagging measures, looking at their rear-view mirror to see where they’ve been. These organizations are rich in data that can reveal insights about customer path-to-purchase, media or channel effectiveness, and consumer attachment rates or purchase behavior. Those insights can help destinations to optimize limited marketing budgets, figure out emerging markets or customer segments to target, and also provide their CVB and travel partners with sharper demand forecasts.

Q: For organizations struggling with huge volumes of data, where do you start?

TB: Start small and get some early wins to move the organization toward a data-driven culture. Connect and analyze your current data sets, obtain insights, use sophisticated data exploration and modeling techniques to test and validate the insights. After that, you can move to live data. This process enables you to learn what insights you might develop from the data while training the analytics models to target the precise clusters of information that you most care about. This is the process of separating signal from the noise, and most of the data is noise.

Of course, to do this you first have to clean and normalize incomplete or disparate data. This will maximize your chances for success and reduce the cost of exploration and modeling. And this process will also enrich your understanding of which sources of data to keep, enabling the business to see the value that a decision science, Big Data and predictive analytics solution can provide.