Formerly known as Wikibon
Close this search box.

IBM's Mission: Remake Customers Like The Weather Company


IBM has always sold industry-specific, transformative solutions.  Now these solutions are moving from “sense and respond” to “predict and act”.  The message to customers is that it is a journey, not a destination.  As soon as they near one objective, another appears on the horizon.  The Insight 2015 conference detailed the current journey toward making every company a decision engine.

  • Systems of Record are the foundation because they are accessible to other applications, specifically digital omni-channel customer experiences.
  • As digital customer experiences become pervasive and no longer differentiating, back-end applications need to be more adaptable in order to support rapidly evolving products, go to market models, and business models.
  • IBM’s Compose-able Business Solutions are far more adaptable and customizable for this purpose than packaged software apps can be without rendering them un-upgradeable.
  • Then sophisticated analytics, such as that built with Watson Analytics, has to infuse insight into “every action, interaction, decision, and business process” to improve outcomes.
  • Adding additional, external data, such as Weather data and social media sentiment that IBM delivers via APIs, allows predictive and prescriptive analytics to keep improving.
  • Building on data and analytics, enterprises can be decision engines to their customers just the way the Weather Company’s output can automatically feed into retailers’ shelf space, inventory, and staffing plans.

A snapshot of IBM’s progress in delivering industry-specific analytic solutions

IBM has always seen itself as a trusted adviser to large enterprises that can deliver transformative business outcomes.  Occasionally, the technologies behind those transformative outcomes go through a major change and IBM itself has to go through a transformation.

Roughly twenty years ago IBM went through a wrenching change from what had become a conglomerate of hardware, software, and integration services to a services-led solutions vendor.  Today, AWS and Azure are commoditizing hardware and integration services.  Similarly, open source and cloud metered pricing are commoditizing infrastructure software.  As a result, IBM is in the midst of another transition.

At Insight 2015 IBM gave an update about how it was pulling more of the puzzle pieces together.  The highlights were well above the Hadoop and Spark infrastructure layers that much of the rest of the industry is focused on.  Instead, IBM was showing how the Internet of Things and external analytic services such as The Weather Company could improve business processes with predictive analytics.

This wave of analytic solutions, which Wikibon calls Systems of Intelligence, IBM calls the Insight Economy.

This research article isn’t a detailed review of the announcements but rather shows how the pieces fit together. IBM’s ideal customer seeks a solution that provides more differentiation than packaged applications can support at one of the spectrum.  At the same time, IBM’s offerings don’t require the skills of a consumer Internet company such as Uber that requires a collection of open source components from independent vendors that need to be built, integrated, tested, and operated at the other end of the spectrum.

Adding value to customer solutions over time

How we got where we are: starting with Systems of Record that automate business processes

Investments in Systems of Record delivered over several decades have provided the foundation for analytic business solutions.  Only by codifying and automating business processes could they become accessible by other programs.

The most widely deployed applications were packaged solutions primarily from SAP and Oracle.  IBM’s Compose-able Business Solutions were always far more customizable than their packaged competitors.  That’s what allowed customers to chain them together and tailor them to support business processes that provided them with differentiation.  It will be no different in the analytic era.

Figure 2: IBM’s industry solutions matrix for Compose-able Business Solutions
Source: IBM, Wikibon 2015

With SoR in place, customers accumulated enough internal operational data for basic analytics

SoR mostly collected only enough data to describe what happened for historical reporting, for example for audit and financial reporting purposes. These types of reporting qualify as descriptive analysis (see figure 3 below) – some forecasting was done but, again, mostly based on historical data.  The data that drives the analytics, however, is growing ever more important.

Figure 3: The different categories of analytics starting with backward-looking reporting all the way to forward-looking prescriptive recommendations.
Source: IBM, Wikibon 2015

With Systems of Record automating back-end operations, business processes were codified and accessible to other applications rather than manual process hand-offs.  That’s what allowed companies to create digital omni-channel experiences.  But in the not too distant future, these digital customer experiences will be so pervasive that they will no longer provide differentiation.

Delivering transformative outcomes composed of industry-specific analytic solutions

IBM is now positioning itself to play at the next frontier, which is to enable companies to achieve more fluid and adaptable back-end operations.

That adaptability would make it much faster and easier for companies to offer new products and services or even change business models.  An insurance company might offer greater pricing flexibility for auto policies.  They could do a better job assessing risk by taking into account the way individuals drive using telemetry data from the car itself.  Packaged software applications such as those from SAP and Oracle can’t yet accommodate that type of customization without rendering them un-upgradeable.

Over time, IBM will build a catalog of such analytic business processes like figure 2 above.  At a high level, figure 4 below shows how analytics are beginning to fit into industry-specific solutions.

First, the solutions make existing processes work better and faster by automating operational decisions with more sophisticated predictive or prescriptive analytics (see figure 3 above).  IBM has begun infusing its industry-specific solutions with more sophisticated analytics, with the goal of infusing insight into “every action, interaction, decision, and business process”.  IBM’s industry solutions matrix shows analytics beginning to permeate them more pervasively. See figure X below.

Second, the new analytic industry solutions are compose-able just like the System of Record ones that preceded them (see Figure 4 above).  That means IBM and customers can create new business processes across the value chain by chaining together different building blocks.

Figure 4: IBM’s matrix of capabilities by industry solution shows analytics becoming more pervasive.
Source: IBM, Wikibon 2015

IBM seems to be a little sketchy about the role Watson’s cognitive computing plays in this process.  Supposedly it processes the “dark” data that is not accessible to conventional analytics.  Wikibon describes cognitive computing as parsing structure and meaning from complex information such as raw data, documents, or images so that the elements can be understood and compared.  IBM has published some API’s to Watson that are accessible through its Bluemix PaaS that give some idea of how far it reaches beyond conventional analytic products (See figure 5 below).

Figure 5: IBM has started publishing API’s to Watson capabilities.
Source: IBM, Wikibon 2015

Developers can use these API’s to extract structure from “dark” data in order to work with it more easily with conventional analysis tools.  Conventional predictive and prescriptive analytics (in Figure 3 above) can drive decisions in analytic industry solutions.

Analytic data feeds are becoming a critical new ingredient.

If analytics is like a recipe, it’s the data that makes up the ingredients.  And when you’re dealing with sophisticated, forward-looking predictive and prescriptive analytics, you can never have enough data.

Figure 6: IBM’s Watson Analytics predictive intelligence tool, introduced in early 2015, guides users through steps that might otherwise require a data scientist.
Source: IBM, Wikibon 2015

Figure 6 above shows Watson Analytics helping a business analyst for a communication service provider.  The analysis is displaying what are the most important drivers of churn, such as age, number of services they subscribe to, whether they stream movies, etc.  The tool is using internal operational data to show that it can predict churn with an 80.2% probability of being correct.  This analysis step helps create the model that determines the best course of action.

The next step in developing the analysis would be to enable the model to automatically prescribe the best action to prevent losing the customer.  It’s that action that the application would automatically present to the consumer if they are online or to the customer service rep if the consumer is talking to customer service.

Things start to become very different if the customer is another business.  The Weather Company’s system can talk to a logistic company’s system to advise which roads will experience the most extreme weather so in order to reroute the trucks.

External analytic data feeds like this are becoming a new, “shadow” supply chain, joining traditional suppliers of physical components and subassemblies as a critical source of value add.  A more accurate name is decision engines.

Weather data from The Weather Company, which IBM announced its intention of acquiring at Insight 2015, and the integration of Twitter analytics, which it announced last year, are two early examples.

Weather Company CEO David Kenny on theCUBE at IBM Insight 2015

Weather has $0.5 Trillion in economic affects per year in the U.S.  And The Weather Company pivoted to address weather-related decisions more directly than it had been doing.  The Weather Company is a remarkable example of a company that in several short years transformed itself from a weather data provider to a weather-based decision supplier.  Their primary customers used to be TV and radio stations and newspapers who relayed the local weather forecast.  Their forecasts were accurate at roughly the county level and they updated them four times per day.

When they saw the transition to mobile devices getting underway several years ago, they remade their predictive models and infrastructure to be as accurate and responsive as the smartphone in everyone’s hands.  They went from updating their forecast four times a day to once every fifteen minutes.  It’s as accurate at 5 days now as it was at one day.  And the weather model is now precise to the level of 4 decimals on latitude and longitude, which works out to 10 meters in each direction!

With near real-time updates and GPS-level accuracy, they are able to serve a new class of customers with a much broader set of services.  Now it’s travel services, insurance companies, airline pilots, retailers, and myriad others.

Just the way Google Maps provides turn by turn directions from a road map, The Weather Company provides all sorts of recommendations from an extremely intricate atmospheric map.  For example, pilots have long used their weather forecasts.  Now the pilot’s iPad application can provide Waze-like mapping of turbulence and predict runway closures before they happen.

By driving these types of decisions they can link their pricing closer to the value delivered by their customer.  With flight planning, they can charge by the plane.  By helping insurers make better underwriting decisions, they can charge a percentage of claims.  Ultimately, products and services from many companies will incorporate the weather decision engine as if it were just another internal application (see figure 7 below).

This diagram from a Harvard Business Review article from October 2014 on IoT shows how multiple decision engines, including one for weather, can feed a farm management system of systems that orchestrates many smart, connected products to deliver higher yields.

Figure 7: The Weather Company has become a decision engine driving more informed processes in many industries. This figure shows weather data helping to optimize farm operations.
Source: Harvard Business Review, Wikibon 2015

The Weather Company believes they can do still better for the most transformative outcomes around life safety.  Currently their weather models are most accurate at the upper and lower parts of the atmosphere.  But tornadoes and lightning form in the clouds, in the mid-atmosphere, and then descend to ground level.  These natural disasters are hardest to predict with anywhere near the same precision as the rest of the weather.

But now that they are going to be part of IBM, The Weather Company wants to leverage Watson’s ability to analyze pictures of the sky to see extreme weather conditions forming farther in advance.  That way everyone would have more advance evacuation advice and disaster relief organizations can preposition supplies more precisely.

The Weather Company is not only a great example of a company that remade itself from a data provider to a decision engine.  It did this by building an infrastructure around the Internet of Things that can serve as a model for enterprises of all types.

They ingest 40TB of data every data to feed their learning and prediction algorithms.  They collect barometric pressure from 40 million smartphones.  They collect turbulence data for airlines from pilots’ iPads by measuring their gyroscopes as they vibrate on 50,000 flights per day.  They collect readings from the windshield wipers on cars.  140,000 individual weather stations feed data continually.  And then, of course, there are the traditional radar and satellite feeds.  Precision and accuracy will only grow over time as The Weather Company’s network of observation sources continues to grow.

Ingesting all this data and turning it into decisions is one example of IBM’s intention of building what they call Systems of Insight around everything in the Internet of Things.

At Insight 2015 IBM showed new predictive API’s that build on the analytics from the Weather Company and sentiment data from Twitter.  As IBM signs more partnerships or acquires more providers, the number of API’s should grow.

Figure 8: IBM has started creating prediction API’s that use external data, starting with weather predictions and social sentiment, to inform key business processes.
Source: IBM, Wikibon 2015

IBM is beginning to roll out predictive analytic API’s to augment its industry solutions.   For example, retailers and consumer product companies will be able to use weather or Twitter-based sentiment data to optimize key decisions (see figure 8 above).  A simple list of API’s can be somewhat abstract.  We found a more concrete example of what “Demand Insights” listed under “Consumer Products” in the figure above might look like.

IBM conducted a contest among partners and ISV’s to mock-up solutions based on the API’s and tools that are still maturing.  A company called SmarterData created an interactive workspace for a retail demand planner trying to adjust shelf space, inventory, and staffing based on the weather forecast (see Figure 9 below).

Figure 9: IBM is making it easier for customers and partners to combine external analytic engines such as The Weather Company with predictions based on internal operational data using spreadsheet-like Expert Storybooks.
Source: IBM, Wikibon 2015

Using historical detail about sales, seasonality, weather, and other factors, the planner can see at what temperature sales of particular products spike.  The idea is that predictive analytics would separate out each of the contributing factors.  Soup spikes when temperatures get below 45 degrees.  Lawn & garden products start to move when spring-like temperatures arrive.  These examples don’t represent any great surprises but they do demonstrate how businesses work when they become more analytically driven.

IBM is making these analytics much more accessible through a product call Watson Analytics Expert Storybooks.  Expert Storybooks are kind of like spreadsheets except that they have much more built-in intelligence about how to find correlation and causation in data.  In addition, partners and customers can extend them with data and analytics targeted at specific business processes.

Figure 9: IBM has a rapidly expanding catalog of Expert Storybooks that span horizontal and vertical processes.
Source: IBM, Wikibon 2015

Action Item

Enterprises trying to assess whether IBM’s solutions are right for them should see themselves as under-served by the limited customizability of packaged applications and over-served by greenfield development of applications built on a custom, multi-vendor data platform.

When evaluating IBM as a vendor, customers shouldn’t be looking to compare “speeds and feeds” of individual products or end-to-end coherence of its product lines.  Rather, the customer should ask if IBM’s analytics can support the business processes they are trying to transform.

You may also be interested in

Book A Briefing

Fill out the form , and our team will be in touch shortly.
Skip to content