This article summarizes the business modeling features of IBM Planning Analytics, and how they differ with competing products.
Adopting IBM Planning Analytics (PA) brings many benefits to planning, analysis and reporting processes. Anyone trying to make-do with Excel spreadsheets was already struggling with growing data volumes and faster forecasting cycles before the COVID-19 pandemic. Conversely, anyone who has built a model of their business on a platform like IBM Planning Analytics (powered by TM1!) is in much better shape, and able to run what-if scenarios to prepare for a range of possible futures as frequently as they need to.
There are a growing number of competing planning software platforms that offer some of the same benefits as PA. This “minimal viable product” (MVP) list of features includes offering a single version of the truth for your data, reporting and dashboarding capabilities, and collaborative interactive planning. Prospective buyers tend to take these features at face value and are unlikely to be able to appreciate the subtle advantages of one product over another without personally spending extensive time in all of them.
What about building models of your business? There is surprisingly little discussion of model building as a feature, considering:
- This is where the value is, beyond the MVP features
- This is technically difficult for a platform to support well, and can therefore be a real differentiator
At time of writing, PA still has – by far – the best tools for modeling your business. By “best” we mean most interactive, fastest time to calculate the answers and (in TM1 Rules) easiest (relative to calculation complexity) to maintain by non-technical people.
The mystery is why PA is not appreciated more for its robust modeling capabilities:
- Business/finance users who only have experience with PA don’t have a basis for comparison, and may assume that what they experience is “standard”
- Technical developers who don’t understand the dynamic nature of business requirements may under-appreciate the complexity of the task at hand, and end up over-engineering a solution as the layers of the onion are revealed
- Modeling power, and the cost of not being able to do something, is hard to quantify to decision makers
Let’s begin by describing PA’s business modeling features. Think of business modeling as the creation of derived or calculated values, as distinct from data modeling which is the process of designing your cube and dimension structures. They are both very important and highly interrelated. When building a complex model you need to structure the data in a way that will support your calculations, analyses and reporting needs. Keep in mind that TM1 is the engine technology that powers PA.
|
Business modeling feature
|
PA Calculation approach
|
Alternative approaches used by competing products
|
|
Dimension aggregation or C: elements
|
Can be used for addition, subtraction and weighting (for example weighted averages, when the weights are constant).
Immediate results.
Calculated on-demand.
Performance (calculation speed) is enabled by TM1’s superior sparsity management.
Results are cached to speed up future retrievals.
|
Some tools need to pre-calculate aggregated values before they are available. In large models this can take minutes or even hours.
Some perform clever partial pre-calculation steps to optimize the pre-calculation process.
|
|
Dimension Hierarchies
|
The same functionality that enables cross-tab analysis using two hierarchies within the same dimension, also allows TM1 Rules logic (more below) to be hierarchy specific, introducing a huge range of business modeling capabilities.
Examples include using separate Years and Months hierarchies on a single Periods dimension, and exploring the impact of restructuring an organization using separate “Current Structure” and “New Structure” hierarchies.
Hierarchies are easy to add and remove, and minimizes the need to add extra dimensions.
|
Dimension hierarchy support is not unusual in the multi-dimensional analysis world, but it’s generally restricted to read-only analytics (Business Intelligence) tools, which are not direct competitors to PA.
Some tools allow for new dimensions to be added dynamically.
|
|
TM1 Rules
|
Immediate results.
Calculated on-demand.
Performance (calculation speed) is enabled by TM1’s superior sparsity management.
Rich expression-based language allowing for linear, one-to-many, and many-to-many relationships to be defined.
Feeders are required for performance and can be difficult to understand and write. This is a disadvantage of using TM1 Rules, but once mastered, the ability to perform immediate calculations of complex requirements is a very powerful differentiator for TM1 and PA.
Results are cached to speed up future retrievals.
|
Many tools only support simple linear expressions such as [Units] * [Price].
Some tools only support expression-based calculations using whatever data is visible on the current screen.
Many tools require batch calculation scripts to be written to accomplish things that PA can do in rules. These need to be "run" and may take time to do so.
Most other tools do not have a feeder equivalent, but neither can they meet the same kinds of requirements without using slower batch scripts.
|
|
TM1 Turbo Integrator (TI)
|
Scripting language and tool that provides TM1’s ETL (“Extract, Transform and Load”) functionality.
TI scripts are run, either on a schedule or on-demand (for example via a button press). In some applications the script can run so quickly that it appears immediate.
TI can be used to perform batch calculations, as well as to move data into, out of, and within TM1. Many data sources are supported including flat files, and direct connections to ODBC, OLAP and other data sources. TI can also be used to call external programs.
|
All robust tools have a version of a scripting language for batch calculation processing.
Their capabilities will be similar to TM1 TI, but some of them may not support the equivalent of a parameterized TI process that can be run quickly by an end-user from a web page or Excel form.
Some tools only support flat file data loads, unless additional software license fees are paid to enable direct connections to data sources.
|
This is not intended to be an exhaustive list of PA features, tools and APIs nor does it provide a full competitive analysis. The hope is that it will be helpful is when a narrow evaluation of business modeling requirements is called for.
Real-world examples will be provided in our next post: “IBM Planning Analytics Rules vs TurboIntegrator: which should I use?”
#pa-administration#pa-home#PlanningAnalyticswithWatson#PlanningAnalyticsWorkspace