Large organizations still struggle to maximize the value they capture from data analytics projects. The blame is spread widely – a shortage of data scientists, failure of executive leadership, disaggregated data and interoperability problems and so on. But what if the problem is too much data science, and not enough visualization of operational models, based on the available data?
Failing to Capture the Value of Analytics
A 2011 McKinsey Global Institute report (“Big data: The next frontier for innovation, competition and productivity”), predicted, among other things, that the use of big data will be the key basis of growth and competition for business enterprises. “The use of big data will underpin new waves of productivity growth and consumer surplus,” the report opined.
McKinsey studied the potential impact of the use of big data across five sectors: U.S. health care, the European public sector, global personal location data, U.S. retail, and manufacturing, making bold predictions about the cost savings, revenue increases and other benefits attributable to big data.
Recently, McKinsey released a follow-on report (“The Age of Analytics: Competing in a Data-Driven World”) which examines industry’s progress against the forecasts of five years ago. A key finding in the report is that companies are capturing only a fraction of the potential value available from big data and analytics.
Retail and Manufacturing
Of particular interest is are the conclusions that the US retail and global manufacturing sectors have been making only grudging progress in capitalizing on analytics.
McKinsey predicted that the retail sector would see a more than 60% increase in net margins and 0.5% to 1.0% annual productive growth. Manufacturers could expect to see operating costs reduced by 25%, gross margins increased by 30%, and product development costs halved.
Today, McKinsey, while still bullish, estimates that retailers have captured only 30% to 40% of the value potential, hindered by the lack of analytics talent and disaggregation of corporate data. Likewise, manufacturers are hampered by disaggregated data in legacy IT systems, and leadership teams skeptical of the potential of analytics; these firms have captured perhaps 20% to 30% of potential value, McKinsey said.
Some of the benefits for retailers have been passed along to consumers, as with the location-based services sector. In manufacturing, the distribution of benefits is lopsided, with most of the value captured by a few industry leaders. Apart from the well-documented need for organization cultural shifts, and a bigger pool of a data analytics talent, what might help large enterprises maximize the value they capture from their data, and sooner?
The von Neumann Connection
The answer may come from an unlikely source – Hollywood. In the 1964 dark political satire Dr. Strangelove, a nuclear crisis – a nuclear first strike on the Soviet Union launched without approval by an insane U.S. general – slowly escalates.
As events reach a crescendo, the U.S. president, in the name of transparency, orders that the Russian ambassador be admitted to the war room, where a massive display wall plots the course of nuclear armed B-52 bombers streaming towards the Soviet Union.
Seated at the war room board table, General Turgidson, brilliantly played by George C. Scott, and modeled after Curtis LeMay, who, after successful Pacific theater commands in World War II, went on to head the U.S. Strategic Air Command, is aghast.
Addressing the president, played by the extraordinary Peter Sellers, Turgidson stammers, “I… I don’t know exactly how to put this, sir, but are you aware of what a serious breach of security that would be? I mean, he’ll see everything, he’ll…” gathering briefing books and looking over his shoulder, pointing to the display wall, “he’ll see the Big Board!”
The slowness with which large enterprises have realized the value from data analytics is attributable in part to the absence of their own “big board.” The enterprise big board is the single visual display that crystalizes operations to speed and simplify executive decision making.
(Mathematician John von Neumann is credited with having created the nuclear deterrence concept of Mutually Assured Destruction. von Neumann, along with Wernher von Braun and Edward Teller, has been floated as Kubrick’s inspiration for the Dr. Strangelove character.)
The Business Operations Center
Many of us are familiar with the Network Operations Center, the activity hub for telephony and network operators that ensures uptime and network services. Or the Emergency Operations Center, the hopefully never used disaster response center for public safety and government agencies.
What’s needed for large enterprises to fully capitalize on the promise of data analytics is the business operations center. The one place that consolidates the carpeted, non-carpeted, and raised-floor activities of the business into a single “big board” display.
Technologists, comfortable with the futuristic visionary role, tend to want to use the USS Enterprise bridge command center from Star Trek as the model for the modern business operations center, but that’s too futuristic.
Stanley Kubrick had it right in Dr. Strangelove. The big board captures, in a concise, single presentation, not only charts and graphs and KPIs, but a visualization of actual business operations, just as Kubrick’s war room big board plotted the military operations of the strategic bombers.
From the Abstract to the Physical
By and large, analytics solutions to date tend to focus on the abstract – tree maps, histograms, box plots, word clouds, etc. The user is locked in the abstract world of their data and their graphs.
And despite well-intentioned efforts by various vendors to democratize analysis through self-service capabilities and “citizen data science,” a great deal of analytics remains strictly in the purview of expensive, and heady, data science teams.
But business executives, at least, those within the discipline, know that “operations” is where the rubber hits the enterprise road. Operations is how stuff gets done. It’s one thing to visualize a dataset, it’s quite another to depict how the business operates. It’s one thing to look at a chart or table summarizing the progress-to-target of the B-52 bombers; it’s quite another to see them moving in real time on the big board.
McKinsey surveyed more than 500 executives across a range of industries and found that many blamed the shortcomings of analytics efforts on the failure to translate insights from analytics experiments into operational model changes for the entire organization.
According to Nicolaus Henke, a co-author of the McKinsey follow-on report, “Many companies invested in analytics systems without fully appreciating that turning data into real value requires a profound reshaping of their day-to-day workflow.” McKinsey argues that process redesign and workflow integration are an integral element of enterprise transformation predicated on data analytics.
Charts and graphs, whether human-made or the product of machine learning, are not enough. For enterprise leaders to change operational models, they need to be able to see them. They need to be able to glide from the abstract – charts or graphs, or visual representations of functional areas – to the physical and the logical, to the plant floor, to the production process, to a piece of equipment streaming data, or to the live video stream monitoring the production line.
And they need to be able to do it in real time, to support situational decision making. The long lead times associated with centralized data science teams are well-documented, and are the impetus for increased adoption of hybrid models that add decentralized analysis. Real-time visualizations of actual operational processes facilitate fast response to exception events.
Beyond situational decision making, executive teams need to be able to perform what-if prototyping of operational changes, to mitigate risk and reduce uncertainty. Swashbuckling executives in large organizations are a rarity. More executives are more likely to much-needed changes if they could better see, in advance, the operational impact of those changes.
The promise of data science all too often is that the answer to improved business metrics is hidden in the torrent of data, it just needs to be surfaced. And that may be true. But it shouldn’t be the starting point. The starting point should be understanding how the business gets stuff done, and visually modeling that. The data and the insights will follow. The big board approach puts operations first, and then leads the user to insights through the logical and physical views, within operational context.
One can readily see how this might apply to manufacturing and the retail supply chain. Intelligent automation is increasingly enabling lights-out production lines on factory floors and robot pickers in distribution centers. Nonetheless, operation of the manufacturing floor and distribution center still requires oversight.
McKinsey, and others, have noted that “the biggest barriers companies face in extracting value from data and analytics are organizational; many struggle to incorporate data-driven insights into day-to-day business processes.” Surfacing insights from the torrent of data is fine, but what to do with those insights?
A single-presentation “big board” view of operations puts what needs to get done at the forefront for business managers. Drilldowns from that top-level operational view to the data and supporting workflows, within the context of operations, will surface the insights that provoke operational improvements.
Implementing the Big Board
It is universally understood that senior executive sponsorship is a prerequisite for large organizations to maximize the value they capture from the use of data analytics, by driving programs and leading cultural shifts. Although data is now, unquestionably, a key corporate asset, skeptical leadership teams would be well-served to shift their perspective on analytics projects from the context of doing something with the data to the context of their business operations.
A simple first step to implementing a big board for your business is to map key operational processes on a whiteboard, from the top down. Beneath that, drill down on each function to map supporting processes, equipment, data sources and their repositories. Solicit input from the operational teams who execute the processes and workflows associated with functional areas – where are the trouble spots that might be addressed through an ongoing examination of the underlying data? Determine if the objective for improvements is reduced operating costs or increased revenue.
Identify two to three processes that would be most important for the executive team to have in its “war room.” (There is increasing evidence that success with data analytics requires investing in multiple use cases.) Determine the metrics for those processes that will measure improvement – lot yields, throughput, and so on – and set objectives that will have a material impact.
Finally, eschew solutions premised on a “data first” approach in favor of an “operations first” modality. A solutions based approach contrasts with the analytical, data science, approach which has been the cause of so many analytics projects failures.
The McKinsey survey heard one quarter of queried executives report that their data analytics programs were ineffective. More than 85% said their programs had been, at best, only somewhat effective. Clearly, if large organizations are to capture the value promised by analytics – value that is measured in hundreds of billions of dollars and which few dispute – new approaches to analytics programs must be considered.
Centering analytics projects on the concept of the business operations center, and a solutions-based, what do we have to get done each day? implementation is a fresh angle of attack on extracting value from data.
Isn’t it time your executive team had a big board in its war room?