The CMO’s Guide to Big Data & Analytics shows how CMOs can use exploding digital social and transaction data from websites, enterprise systems, smartphones, tablets, cars, appliances, devices and the Internet of Things in surprising ways that have more to do with customer insight than technology.
The digital social world and modern cloud-based technologies have reinvented the practice of “analytics” so that it’s faster, cheaper, and more open. I’ll paint a broad picture of this new world before referring you to other resources where you can drill down.
Big data is usually discussed from a big-ticket I.T. perspective, but new technology enables marketers to practice “lean data,” which starts small and proves/iterates hypotheses before scaling with large investments. The biggest barriers are knowledge and imagination, not technology.
Big data elevates opportunities for marketers because they can know their customers and serve them better based on their actions, not only what they say. Conversely, big data is increasing customers’ expectations for being treated as individuals, not “consumers.”
Big Data & Analytics Overview
“Big data” is in full hype mode as of this writing, but there’s serious meat behind it and exceptional opportunity for chief marketers. Most readers are probably somewhat familiar with the rather painful legacy way that “analytics” has been practiced up to now. Several things are changing the situation in ways that favor marketers:
- Modern technologies like cloud make processing, storage, and other computing resources far less costly, and start-ups are providing a dizzying array of tools to make analytics easier. Massive computing power is available by the second or the transaction.
- Digital social technologies have made it exceedingly easy and fun for people to interact digitally about everything. For the first time ever, the whole spectrum of people’s fears, dreams, desires, frustrations and exhilarations are online, searchable, and forever. Social data is the key to understanding the context of customer satisfaction.
- Open data is exploding because governments, universities, people, and enlightened firms are making their data available for free.
- Mobile technologies are creating digital data that fuse transactions (what people are doing) with where and when they are doing it, creating an unprecedented view of behavior.
The above points make big data a new phenomenon. Its official definition refers to “the four Vs:” volume, velocity, variety and veracity. More types of data are available than ever (“variety”), and the volume of data is growing exponentially (“velocity” and “volume”). Veracity refers to the degree to which the data can be trusted.
Here are some examples of big data sources: social network data, email response rates, geosocial data (i.e. Google, Yelp, Foursquare checkins) ecommerce transactions, chemical composition of sewage, shopping cart abandonment, call center logs, warranty data, weather, sports schedules, disease outbreaks, how many Parisians ran sub six-minute miles between 6-9 a.m., natural disaster data, school closings, import/export data, city license renewals. The main barrier is imagination, which is a great place to be.
Big data will be huge data next week because mobile devices follow people everywhere, and apps encourage people to interact with their surroundings or as they pass through their surroundings. When you rent a bike, pay for something, text someone, even walk down the street, you are creating digital information that can be gold to certain firms or people.
Big Data: How to Get Started
Get Grounded in a Customer Context
To use big data most effectively, companies and marketers need to reorient themselves to their customers (clients, employees and other stakeholders). All the types of data mentioned here are rapidly changing customer expectations: Because people are sharing a lot, they increasingly expect to be treated as people, not “segments” or “demographics.”
The customer context also means recognizing that people are less interested in products and services than in the outcomes they can attain by using products and services.
The new customer context represents a profound shift for marketers. Up to now, it hasn’t been practical for companies to know their customers individually, so no company did and customers couldn’t expect it. Now, digital social and big data make it efficient and practical to “know” customers individually.
It is now possible for companies to treat their customers personally, at scale.
So focus your big data initiatives on learning about and supporting the outcomes your customers want. Don’t let the product focus be a barrier between your company and the customer. The current state is “brands are from Mars and customers are from Venus.” You don’t have to do this anymore. Align with customers by focusing on what they really want: outcomes of using your product.
For example, if your company or brand involves camping equipment, focus on helping people have better camping experiences. This has to be specific: family camping is different from cross-country cycling, college expeditions…In another example, women wear hats in various situations, so learn about what outcomes they want from wearing hats in situations that are most relevant to the hats you offer.
Don’t Get Blinded by Technology
The overriding promise behind big data is seductive: machine derived learning about customers or other stakeholders. Big data enables better decision making by selectively replacing managers’ “impressions” or “gut feel” with “data” that are presented using amazing visualization tools. This is all true, but it’s still a long way from producing a profitable return on investment.
All organizations are sitting on data, and cloud and modern tools make it easier than ever to work with structured data (often from internal data bases) and unstructured data (social network data, other external data). The question is not whether an organization can “create value;” rather, can it create enough business impact to justify the investment within a certain time frame?
Be Creative with Data
Most teams start from the organization perspective, “What do we have, and how can we use it?” I’ll suggest that a more direct path to success is to approach from the customer point of view by asking, “What do our best customers want to accomplish by using our products/services?” What do they need to know to attain the outcomes they want? To really get this, the company must put customer outcome first and product sales second. Companies that do this will sell more at higher margins.
Most big data thought leadership emphasizes creating and testing hypotheses. Take this further by focusing on outcomes like these:
- The camping equipment brand thinks, “If we could analyze transaction data, we could learn when customers go on extended family camping trips and help them to have rewarding experiences. By helping them to have more fun, a greater portion of them will become more avid campers. We could help campers by analyzing weather patterns, festivals, sporting events, and other events that influence traffic and business at camping areas.”
- The hatmaker team hypothesizes, “By analyzing transaction and returns data, we can discover that an important portion of women who buy our exotic hats do so for weddings and graduations. In certain geographies, intermittent rainfall can ruin certain models, so we can analyze geographic and weather data to make suggestions for them to look smashing without having their hats ruined.”
- A commercial bank team asserts, “By analyzing small business and personal accounts, we could explore how the success of a business affects customers’ uses of more of our services. We could look for patterns and triggers in their businesses’ websites and offer to help them grow their businesses, say, by making introductions to other clients in complementary areas.”
The Social Angle
The conventional wisdom says that the teams need to test their hypotheses by piloting big data, and there are elegant ways to do that. However, in most cases, teams can use digital social venues to test the value proposition embedded in the hypotheses, to pre-validate the project at a lower cost. For example:
- The camping team obtains preexisting reports that show what kind of events produce the largest camping equipment purchases and returns. It analyzes the data to deduce events with extended families. For example, families camping with grandparents probably buy certain things. The team analyzes the digital social ecosystem to find venues in which people are talking about camping trips. It learns, fast, what’s important, what information has the most impact on outcomes.
- The hatmaker and bank teams can do the same thing. It’s important to realize that the traditional method of surveying people is far less useful than observing people talking among themselves. When a business asks an individual what kind of information would have made a difference, that is an explicit question, and few people have the awareness to give a complete response. However, when people are interacting among themselves, far more implicit information emerges.
Recommendations from Big Data Expert Kaiser Fung
- Start small. Often the biggest impact is achieved with combining very simple data sets. Big data often has diminishing returns to scale.
- You must invest (big data not a quick fix); Netflix prize to beat its Cinematch; diminishing returns, the simplest (coarsest) data usually has the biggest impact. Winner “107 Ensemble” 10% better than Cinematch, but other contestants were close.
- Think long run. Email opens is an example. One of his charts showed a marked dip in opens in June, but if you have data from other years, you note the same thing other years; history gives context.
- Network data helps, too; for example, the more Facebook friends you have, the better Facebook can make recommendations (it has more data on your social graph). So don’t quit before the fruit ripens.
- Don’t be “data-driven,” be “results-driven.”- Cisco example: their (response) strategy is based on network data from their observations of and interactions in digital social networks.- Showed Harrah’s casino flowchart: takes data at various checkpoints in a workstream so each checkpoint generates data and they can measure the workstream.
– Target uses data for customer acquisition, but most of their 25 metrics can only be measured on existing customers, so it’s more about customer retention. They know that pregnancy is a life-changing event, so people are far more likely to change their permanent shopping habits. Therefore, Target wants to retain existing customers and acquire new ones. They offer promotions to existing customers.
- Bad data is worse than no data, and it’s hard to detect. Example of data table with three columns: date/time, user ID and “upload video” data. One day it went all haywire. Had to deconstruct the user workstream and checkpoints to find where the deviation was happening. Besides, what precise user action populated the “upload video?”
- Keep it simple. Quick audience poll asked us to predict, within a sample of 100 study participants and 30 were engineers and 70 attorneys. Gave a bio description of a 45 year old man, hobbies, political interests, family. Then multiple choice: the probability that the sample was an engineer. a) 10-40%, b) 40-60%, c) 60-80%, d) 80-100%. Answer was a) not because of any of the bio, but because you knew 30 were engineers. Showed that it’s too easy to get distracted by extraneous details like the bio.
Working with big data requires an unusual combination of expertise and skills, and its relatively sudden growth means that there is a serious people shortage. However, you can mitigate this risk by being creative with how you access the expertise you need.
- If you ground your big data initiatives in helping customers attain desirable outcomes, you will find that internal data may be less relevant than you might have assumed at first. You will waste less time chasing skills you may not need.
- In most cases, creating a cocktail of internal and external data will add the most value. However, you can use a small core of data science experts and build customer focused people around them. Contracting might make more sense than you think, especially during the pilot phase.
- Kaiser emphasized that creativity and hard data science is a very rare combination. It will be easier to create that combination at the team level rather than the individual level.
- Include social business people on your team; they are people that specialize in deep interaction more than promotions and marketing.
The Privacy Angle
Too many companies don’t take privacy seriously, and they are sitting on a time bomb because it’s only a matter of time before they are outed by customers, the government, or geeks.
- “Dollar on the sidewalk” mentality by firms: they think, well, it’s there, use it.
- “Because it’s free” syndrome: social networks, because they provide a free service (Facebook, Twitter), they feel little compunction to safeguard privacy.
- “Forced consent:” so many websites force the user to agree to terms before they can even interact with the website (and virtually no one reads the consent form or understands it).
- “Left hand, right hand problem:” the marketers in the firm make privacy promises to users, but the data scientists are rarely informed about it, so they use the data in ways the firm has promised they won’t. Example, telecoms promise they aren’t using location data.
- We need strong self-regulation, or the government will step in.
- How ‘Big Data’ Is Different, MIT Sloan Management Review.
- Big Data: The Next Frontier for Innovation, Competition, and Productivity, McKinsey.
- Big Data & Analytics Competency Center, CSRA.
- Real-time Big Data Feed, curated for marketers.
- Personal Individualized Experience, the DNA of Digital Transformation, CSRA.
MENG is the indispensable community of executive level marketers who share their passion and expertise to ensure each member’s success.