QueBIT Blog

QueBIT Blog: How IBM Cloud Pak for Data supports xP&A

Written by Ann-Grete Tan | Feb 23, 2021 3:00:00 PM

What is xP&A?

QueBIT’s typical client is a Vice-President of Finance, CFO, or an operational executive - say an Inventory manager or Vice-President of Supply Chain. Our typical conversation with them involves an aspect of digital transformation in their planning, analysis or reporting processes; and how technology may help them increase business agility, improve forecast accuracy, speed up planning cycles and make better data-driven business decisions.

At QueBIT, we LOVE the business of Business. We see technology as a way to make business better by letting machines take care of mundane details, while liberating humans to focus on high-value activities like gaining insights, solving problems and innovating.

Recently, a frequent topic of conversation has been xP&A or “eXtended Planning and Analysis”. xP&A represents technology-enabled integrated business planning across various functions of a business such as Finance, Operations, and Sales. In most companies these functions plan in siloes with the Financial Planning and Analysis (FP&A) folks taking care of the annual budget, and quarterly financial forecast while the Sales and Operations Planning (S&OP) people do rolling demand forecasts, and capacity planning (for example). FP&A and S&OP are complicated time-sensitive processes that require extensive coordination and are hard enough to complete separately. They are also on different schedules – FP&A is most often on a monthly cadence, while S&OP may be working by week or day.

Even so, a CEO who takes a top-down view usually doesn’t care about process complexity. The CEO expects the revenue forecast in the financials to be consistent with the sales forecast as well as the demand and capacity forecast. After all, if the sales team claims they can sell much more than capacity constraints would allow, there may be a problem!

Many companies solve this cross-silo consistency problem through extensive, and expensive, reconciliation meetings. Besides taking up precious time and energy, this approach can reduce forecast accuracy because each group has built their plan based on different assumptions, and the final compromise may have more to do with negotiation skills than factual data.

The alternative approach embraced by an increasing number of forward-thinking companies is to use technology to set common data and modeling foundations for ALL the plans. All planning processes start with historical data. If this historical data is sufficiently granular, you can roll it up in different ways to serve different constituencies. For example, if you start with daily data, you can easily get to weekly data, monthly data or quarterly data simply by adding it up in different ways. Similarly, if you start with detailed product codes (or SKUs), you can easily get to Product Categories, or Brands, and maybe even Business Units (depending on how your organization defines a “Business Unit”).

Once you have a shared base of historical data that has been validated and that everyone trusts, you can build models of your business to construct your plan. The models will likely combine Artificial Intelligence and Machine Learning (AI/ML) powered predictive analytics with constructive driver-based techniques. For example, you may use predictive analytics to generate a detailed demand forecast based on historical data, and then combine this with a pricing scenario to arrive at your revenue forecast. The operational team may then look at the demand plan by week to drive capacity planning, while the finance team may look at the revenue forecast by month, for the rolling financial forecast. The starting points are the same, but the models, the reports and the downstream analyses can serve a range of business needs. And if any assumptions change, or if you need to play what-if games with different assumptions, you can easily do it across all parts of the plan.

This is the power of building models, and it is the promise of xP&A.

How do you get to xP&A?

While you certainly can’t get there with spreadsheets (as we explained here), there is in fact no single technology solution that can do it. What it takes is a combination of business process change, technology infrastructure to support the changes, and a practical incremental execution plan that builds on a sequence of small successes.

Even though business process change is a key ingredient and technology alone cannot solve the problem, the right choice of technology significantly decreases costs, time and effort – which contributes to the probability of success. IBM Cloud Pak for Data is an exciting technology platform for you to consider that checks all the boxes needed to support xP&A (and many other business) requirements in today’s world.

What is IBM Cloud Pak for Data?

Let us begin by breaking down the tasks that are part of an xP&A deployment:

  1. Accessing historical data which may come from many different systems, running in different places.
    • Some systems may reside in your own data centers (for example your Oracle or SAP Enterprise Resource Planning (ERP) system), while others may reside in the Cloud (for example your Salesforce Customer Resource Management (CRM) system or your Workday Human Resource Management (HRM) system). Companies that have grown by acquisition may have multiple ERP systems. There may also be data warehouses, or special industry-specific applications.
  2. Combining and collating all this data in a coherent way, to make it useful.
    • Each individual system has its own “language” and rules for describing the data. Data from different systems must be translated to the same “language” before they are put together.
    • Accessing, combining and collating data should not necessarily mean making a copy all the data and then finding a place to put it – which can get expensive!
  3. Validating the data so that it is trustworthy.
    • Once you translate the data, the people who know it from the original system may not recognize it anymore! Validation and control processes must be put in place to restore trust in its provenance and accuracy.
  4. Designing and building models to be used for planning.
    • These can be AI/ML-powered predictive models or constructive driver-based models like the units x price example we described earlier.
  5. Enabling reporting and analysis to drive better decision making!
    • This can range from management and statutory financial reports, to self-service analytics and interactive dashboards.

Think of IBM Cloud Pak for Data (CP4D) as a Lego base plate that comes with an unlimited supply of Lego pieces called “services”. Not only are there services to perform each of these individual tasks, they work seamlessly together in a highly scalable and performant way. Furthermore, just as you can carry a Lego base plate around, you can deploy the CP4D – with its services anywhere! You can put it in your own data center, or in AWS, Microsoft or IBM’s Cloud. You can even scale it outwards, which is like connecting many rectangular Lego base plates to form a larger rectangle!

The CP4D services come in two flavors: “Base Services” are free with the CP4D platform while “Extended Services” cost extra and are purchased as premium add-on “cartridges”. The services that are most useful for xP&A use cases include:

  • Data Virtualization, which integrates data sources across multiple types and locations, and turns it all into one logical data view
  • IBM Planning Analytics, which is an AI-infused integrated planning, forecasting, budgeting, analysis and reporting solution
  • SPSS Modeler, allows you to build AI/ML-powered predictive models
  • IBM Cognos Analytics, which provides self-service analytics and enterprise reporting infused with AI/ML

Click here to read IBM’s official answer to the question “What is CP4D for Data?”.

Is CP4D a good fit for my organization’s xP&A initiatives?

Whether CP4D is the right choice to support your xP&A initiative depends on several factors including:

  • The maturity of your data strategy and infrastructure.
    • If you already have access to all the data you need to meet your xP&A goals, in exactly the form and frequency that you need, then you may not need CP4D.
  • If you only have one data source, you probably do not need CP4D.
    • The one data source would typically be a single ERP system.
    • That said, it would be unusual for a single ERP system to suffice for a full-blown xP&A initiative aimed at integrating business planning across finance, operations and sales.
  • The volatility of your data strategy and infrastructure.
    • If yours is a fast-growing company, or acquisitive, or undergoing a lot of change and transformation, new system deployments may be happening frequently.
    • CP4D would give you an elastic and scalable way of easily accessing additional data sources, regardless of whether they reside in on-premises data centers, or in the Cloud.
    • The value of CP4D increases with the number of data sources, the variety of data sources, and the rate at which they are being added.
    • There may also be other economies of scale: remember that CP4D can be used for data, analytics and AI/ML use-cases beyond just xP&A!

Contact us at info@quebit.com to find out more about whether CP4D is the right choice for your xP&A data strategy.

Thank you to Mike Cowie, Walter Coffen, Dat Nguyen and the rest of the QueBIT CP4D Guild for their help and contributions to this blog post.