Now that we’ve walked through the stages of the Sentient Enterprise journey, you can see how this approach is designed to apply broadly to many different kinds of companies scaling their analytic architectures to grow bigger, more autonomous, and more agile. The flip side of this broad applicability, however, is there’s no single way to go about implementing this capability maturity model. Nor is it completely sequential.

As we stressed way back in our Introduction, it’s possible to embrace different themes simultaneously, or even out of sequence, depending on your organizational structure and business needs. And however you organize it, the sheer scope of your to-do list can be daunting. In fact, after hearing us talk at a seminar about the Sentient Enterprise journey and all that’s involved, one panicked executive at a major corporation buttonholed us to say she felt like she was at sea level looking up at Mount Everest.

She made a good point, and our reply was that even Mount Everest is tackled incrementally, in phases. Everest has more than a dozen routes to the summit, in fact, with a base camp and four other major camps along the way. The Sentient Enterprise journey is also doable if we look at it the same way, in phases. Our rattled executive can also take heart from those who are already well on their way up the mountain. “Of course, when you look at the beginning, it looks very hard,” said Siemens Mobility’s Gerhard Kress. “The only thing you can do is dive in and get started somewhere.”

More than anything else, creating the Sentient Enterprise involves a tremendous amount of change management that goes far beyond anything you could capture in an employee handbook or the typical strategic plan. Your job is to put an entire culture shift into action at an established company, with fiefdoms to engage and IT policies to overcome. We’ve already shared with you the “change management on steroids” shorthand we use to describe the journey at conferences and companies around the world. There is simply no other way to describe it, and there’s just no single way to go about it.

We won’t prescribe a single path to sentience, because every company is different and must uniquely tailor the journey to its own unique history, culture, operations, challenges, and goals for the future. That said, we can show you some very common and practical considerations that most businesses will face along this journey.

To that end, we’ve structured this chapter as a kind of guided tour of some important and interrelated priorities to keep in mind as you implement. These are some of the major signposts as you drive your organization’s analytics capabilities along a path to sentience and maturity.

ASK THE RIGHT QUESTIONS, WARTS AND ALL

Making real progress starts with a candid assessment of where you are currently. Let’s be honest: these moments of recognition can be incredibly uncomfortable. When we finish talking about the Sentient Enterprise to a roomful of business executives, we invariably have people approach us afterward who are literally in shock, saying things like: “Oh my gosh, we are so behind the curve. Where do we start?” Our answer is that, by realizing the challenge, you’ve already started; the next step is to ask straightforward and honest questions of your business.

Where are the silos and hurdles to agility? What are the policies around copying data? Is our funding structured for innovation and adaptation to new opportunities? How do we recruit, train, manage, and retain our data professionals? These are just a few questions you need to include in a “warts and all” checklist that is a crucial first step toward changing things for the better.

As you might guess, the navel-gazing doesn’t usually go well when done solely from within the company. That’s why we recommend you invest in an agility audit. This kind of organizational review is best done by an outside party with no biases or political ties to your company to help establish an independent, baseline assessment of how agile your analytics operations and processes currently are. The audit should be thorough and company-wide, with resulting collateral like presentations, training, and a road map for change that includes future steps toward agility.

A good agility audit will feature an itemized and prioritized backlog of current issues and tasks that need addressing. The list should include some early wins—changes that can be made quickly and fairly easily—along with more far-reaching transitions that may require months or even years to implement. The entire process can help you socialize and expand on practices and areas where your company is agile, and remedy those areas where agility lags.

AGILE STRATEGIC PLANNING IS NOT AN OXYMORON

Many established strategic planning approaches you’ve learned about for years in business school are perfectly good ones that remain relevant today. It always makes sense, for instance, to define goals, set direction, and make decisions on allocating resources in pursuit of a strategy. It’s also a great idea to measure results and have flexibility for both planned and emergent events as you compete and adapt to market environments.

However, big data has changed one factor—timing—in a way that has upended the whole notion of strategic planning and how we should go about it. Technology has helped speed enterprise production to the point where some industries operate on a three-week product release cycle. In such a fast-paced and data-enhanced world of rapid resourcing and action, it doesn’t make sense to have all your strategic planning happening on a two- or three-year rhythm, with reliance on long-term waterfall methodologies that simply don’t apply in today’s world.

As we discussed in Chapter 3, waterfall methodologies are step-by-step approaches that may be fine in cases where you have a clear picture of what your final outcome should be, and time to completion is not a big deal. Those two conditions alone, however, are nonstarters when you consider the role analytics needs to play in today’s competitive markets. Most companies nonetheless stick with outdated waterfall playbooks from a less agile era. This leaves them stuck with micro-incremental product improvements and vulnerability to competitors that understand that bolder and more disruptive improvements are what now drive competitive advantage.

Your strategic planning processes must recognize this reality, and the Sentient Enterprise journey requires we update to a more agile and opportunity-driven framework. Think, for example, of how financial options work or how movie studios strategically bankroll many productions concurrently in hopes that a few efforts will pay off big. We need to pivot the enterprise to strategically invest in similarly agile ways.

At the same time, we can’t ignore the basic financial realities that come with being a large company. Especially for publicly traded firms, for example, strict annual and quarterly budgeting processes are a fact of life. No amount of cheerleading for agility can change that. But compromises are possible. Consider designating annual or quarterly-funded programs that, in turn, feature flexible allocation models internally that sponsor a handful of short-term projects happening concurrently or in quick succession.

These sprints can be pilot efforts where program or project managers can track progress in shorter increments—a week or two of work, say—using measures like net present value and other real-time snapshots on return on investment (ROI) and profitability. It’s the kind of agile, opportunity-driven approach that dovetails with our next suggestion: that you view your analytics team almost like a venture-backed entrepreneurial team within your own company.

ADOPT A START-UP MIND-SET AND DON’T BOIL THE OCEAN

It’s easy for a start-up to be agile; it’s almost a given when you’re new, small, and hungry for results. But larger enterprises can get caught up in bloated procedures, legacy systems, static financing, workforce issues, and other barriers to agility.

Our advice: don’t try to “boil the ocean” by taking on the whole company at once in your pursuit of agility. We have a tremendous challenge in front of us, but we can take it in steps. The way to do this is to start thinking of yourself and your analytics team as a kind of venture capital-driven start-up team within your own company.

Here’s what we mean by that: Instead of trying to fix the whole company at once, pick a key area or two where some business problem happens be a good candidate for a particular analytics solution you’re working on. Find partners who are close enough to the problem that they’re desperate for your solution and are willing to act as validators at staff meetings, budget sessions, or anywhere else you need buy-in. The more you can limit the scope of these early pockets of agility, the more quickly you can scale success within them to demonstrate what works best.

If these targeted and value-driven strategic engagements inside the company resemble how a start-up operates out in the real world, that’s exactly the point! In fact, try to look at internal partners as those “lighthouse customers” we mentioned in Chapter 6—internal divisions or departments whose business problems make up a kind of venture portfolio within the company. Focus on these pockets of agility and peg success to milestones. Use short-term performance indicators to spot early wins or pitfalls so you can know which projects to highlight and keep pursuing, tweak, or abandon.

As for time lines, condense them in a “fail fast, recover fast” approach where you focus like a hawk on one or two things for a month, maybe two, and then adjust your approach based on a granular understanding of what worked and what didn’t. This is much better than trying to monitor a bunch of outcomes over the course of a year or more. Wherever possible, in fact, look for business problems with natural limitations around scope and time frame. This is good advice, even for established analytics leaders looking to refine or test new approaches.

For instance, Verizon Wireless mounted a proof of concept for better predictive modeling around customer churn in its prepay division; the goal was to examine behaviors and decide which customers to target most aggressively for retention. Verizon focused on the 30- to 60-day window after a prepay balance hits zero, during which time the company holds the phone number in limbo just in case the customer decides to replenish the account. The built-in limitations—prepay customers only, and that built-in 30- to 60-day time box before releasing the number—served as natural contours to help focus and pilot a new (and ultimately successful) analytic approach for the company.

PICK THE RIGHT INTERNAL PARTNERS TO DEMONSTRATE VALUE

Choosing these intracompany partnerships is not an exact science, but you can follow a few guidelines. Certain operations, like marketing, tend to work quickly and incrementally, with feedback loops built in; marketing campaigns tend to run on short cycles and measure progress often through surveys, social media, and other channels.

Such operations can be good candidates for a pilot project, since they tend to match the agile cadence of your analytics approach. Other parts of your company, like finance, may be more tethered to long-term cycles and measures and are therefore going to be tougher prospects for early wins.

Sometimes the same kind of operation will vary in suitability, depending on the business context. The pace of supply chain operations, for example, may be ripe for agile analytics in the case of just-in-time parts provision manufacturing, or in e-commerce. Other contexts, like supply chain operations at a government agency, could take additional effort to apply the venture approach to analytics.

Throughout, make sure to respect your partner department’s expertise. Don’t come in and tell them how to do their work; your job is more as a solutions provider whose goal is to share the philosophy and framework around agile analytics and show how this approach can work for them. As you rack up successes, take them viral throughout the company by finding bigger projects and/or seeding team members from successful efforts onto other teams that are gearing up for another pilot project somewhere else in the company.

EMBRACE AGILE PROJECT MANAGEMENT STRATEGIES

We’ve talked at great length about the use of platforms as a kind of framework to help users be agile in accessing, visualizing, or making decisions around data. But your analytics teams also need an agile framework of their own to set up these architectures in the first place. Big data demands fundamental changes not only in how we conceive agility, but how we actually carry it out in practical, everyday terms as we structure and build our data architectures.

Many organizations talk about agility but don’t have the right tactical methodologies to follow through. Fortunately, as we mentioned in Chapter 3, agile project management platforms such as Scrum are already out there. And as we discussed in Chapter 6, the whole concept of DevOps is revolutionizing the agility and repeatability of analytics work flow and applications development.

Embrace iterative and incremental methodologies that challenge assumptions of waterfall and other traditional, sequential approaches to product development. Enable teams to self-organize through physical colocation or close online collaboration, with frequent (often daily) face-to-face communication among all team members.

Rely on user groups to share best practices and get training or certifications. One such group, the Scrum Alliance, summarizes the allure: “When was the last time you put ‘collaborative, sane and enjoyable’ in the same sentence with ‘business goals’? You may not remember unless you already use Scrum, but with Scrum you can, indeed, enjoy your work again!” Agile project management improves not only the team experience, but also the work product by maximizing the team’s ability to deliver quickly and respond to emerging requirements.

EMBRACE CONCURRENCY, ENSURE SCALABILITY

It’s easy to be agile when you’re a start-up with just a few employees and a vision. But what if you’re a large enterprise with many employees, legacy systems, and a hardened culture that may be anything but agile? We need to keep the insights coming in the face of mushrooming data volumes at scale.

Agile methodologies aren’t much good if you don’t have the analytic heft to run them at scale. This is true whether you’re talking about documentation, product management, supply chain, customer satisfaction, quality control, or any number of enterprise activities. Wherever we grow the business, we have both the techniques and infrastructure to support big data at scale. Otherwise we’re just investing in bigger data streams as the insights float by unnoticed.

One of the biggest scalability challenges by far involves concurrency: the need for today’s large enterprise to do many things at the same point in time. Consider how a major international bank today can have 18 billion events or more across all identified customers. Especially with complicated challenges like fraud prevention, you may need to deal with hundreds of thousands of events happening concurrently—things like teller interactions, online transfers, call center and e-mail traffic, ATM visits, account cancellations, and other occurrences. The payoff in a use case like this is huge if your systems and people can navigate all this information in real time. But, if the concurrency isn’t there, your discovery teams and tools are stuck in linear mode.

Even in cases where there’s adequate funding, too many organizational systems and processes continue to be built in uncoordinated ways. The resulting mismatch can feel a lot like driving your expensive new sports car on the Autobahn stuck in first gear: it’s very frustrating, you don’t get very far—and you still paid a lot for the experience.

DESIGN IN GOVERNANCE THAT’S SEAMLESS AND REPEATABLE

Scalability also drives our argument for making sure your governance is built early and seamlessly into your data analytics systems and architectures. Think back to our simple analogy in Chapter 6 about version control: When composing documents, Microsoft Word is a very good platform when writing a draft or sharing tracked changes with a colleague or two; but the process can get unwieldy and complicated if more people join the effort. Google Docs, on the other hand, manages version control automatically and effortlessly, even if you have many contributors.

When you’re building analytical applications, as we did in Chapter 6, the governance and documentation should be seamless. The ultimate test for governance is whether the experience feels seamless and effortless for your end users, and that means governance must hold up at scale by being repeatable and hassle-free.

As another example, consider taxi reimbursements in the pre- and post-Uber eras. Reimbursement offices at large companies can get swamped with paper receipts (or scans of those receipts); or that documentation can remain hassle-free via Uber’s automatic documentation online when people use that Internet-based ride-summoning service.

OPTIMIZE A WORKFORCE TO ACT FAST, FAIL FAST, AND SCALE FAST

Through the Layered Data Architecture, LinkedIn for Analytics, and related platforms, we’ve shown you ways to optimize collaboration around data. But ultimately, collaboration happens between people, and that means implementing the Sentient Enterprise can’t become reality in your organization without close attention to the people who work there.

When hiring new talent, look for gifted and self-driven engineers—and look for proof of this beyond what people may claim on a resume. Certain companies have reputations for progressive approaches to analytics, so pay special attention to applicants who are coming from those organizations. Successful cloud companies, in particular, are good sources for talent. As we learned in Chapter 6, the cloud is all about agility, new tooling, and finding ways to scale products and services without scaling people. So look for folks who come from that background.

Visit GitHub when vetting candidates, or some other code repositories, to see if your candidate has self-published interesting projects or technical challenges. If you don’t find anything, ask your candidate to go ahead and do something like that as part of the interview process. When someone asks, “What is GitHub?” or “Why would I do that?,” it’s time to move on to the next candidate.

That said, don’t focus excessively on new hires. Getting your workforce where it needs to be is less about new minds than a new mind-set. Focus on the kind of people and skills you need on hand to innovate. Chances are that many of those people are already in your company and just waiting for an opportunity to lead or support change.

“IT’S THE CULTURE, STUPID”

Any of the aforementioned strategies need to happen with an overall culture shift in mind. We’ve talked a lot about the collaborative spirit and how to promote and protect that. Employees must be trained not to rely excessively on intuition, emotion, and anecdote-based decisions and instead start trusting the data. Intuition is great for ideas, but data is actual proof.

“One of the biggest challenges we face in moving the needle as fast as we want is finding the right people who both understand the data and build the models; that’s a very unique skill set,” said Dell’s Jennifer Felch. “You can have the most incredible technology, but you also need people who are very comfortable with the data—the definitions, what math means, and how to handle distributions. Our challenge used to be: ‘We don’t have all the data.’ Now it’s: ‘We have the data, but we need people with the skills to understand and make sense of it.’”

Data-driven decisioning stops us from relying too much on what we feel or think may be happening. Workers also need to understand which metrics matter most for the business, what decisions need to be driven by the data, and how to harness algorithms to make the most strategic decisions possible. This involves understanding the business context within which data correlations occur.

It’s been said that agile is not something you do; it’s something you become. With that in mind, make sure you have a solid system to promote agility at every turn: from how you communicate to employees and conduct company meetings to how you build your teams and run your campaigns. Ignore corporate culture at your peril: as famed management consultant and theorist Peter Drucker reportedly said, “Culture eats strategy for breakfast.”