How Much Supply Chain Optimization Do We Really Need? - Part 2

Part 1 of this blog series introduced the concept of (Rapid) Response Management in the realm of supply chain management (SCM) via a software category pioneer, Kinaxis. The currently bullish Kinaxis has a number of customers that are SAP ERP customers too, and for a long time SAP was at first dismissive (or at least ambivalent) regarding the need for Response Management, as the company had its own well-known SAP Advanced Planner and Optimizer (SAP APO) product. In addition, Kinaxis has had to compete with other advanced planning and scheduling (APS) providers such as JDA Software (former i2 and Manugistics), Logility, and Oracle.

Since November 2010, SAP has been distributing a supply chain solution by a lesser-known German software company ICON-SCM as SAP Supply Chain Response Management (SAP SCRM) by ICON-SCM, a solution extension to its own SCM suite, SAP SCM (see TEC’s article entitled SAP SCM – Stepping out of Obscurity). SAP had investigated several options to satisfy this role, and presumably one of those options might have been Kinaxis. For its part, Oracle released its internally designed standalone product in 2009 called Oracle Rapid Planner, which can be layered on top of Oracle’s enterprise resource planning (ERP) products and other ERP products.

Now, Kinaxis feels vindicated by Oracle and SAP's endorsement of the market at long last, but is slighted by the IKON-SCM partnership, plus, it is a fierce competition now. In addition, isn’t "optimization" part of the SAP APO name, and why then did SAP introduce a separate SCRM solution? It seems we may be talking about different kinds of optimization. Maybe in certain situations it's more appropriate to use one kind of optimization versus the other.

Optimization Explained (in SAP Lingo)

SAP apparently draws a distinction between Capital-O "Optimization" and so-called heuristic "Search" optimization (i.e., searching for the best alternative amidst multiple possible options). Capital-O Optimization in APS is about finding a mathematical optimum, typically minimum costs or maximum profits (in a "boil the ocean" manner). On the other hand, heuristic Response Management processes attempt to provide a solution to a problem that may or may not be optimal. Often, "Search" involves criteria other than cost or profit, such as rules and priorities.

Both Capital-O Optimization and "Search" heuristics in SAP’s related products can respect material or capacity scheduling constraints, unlike generic (fast) material resource planning (MRP) products or the Supply Network Planning (SNP) heuristic module within SAP APO, which largely run unconstrained. These rapid (or not) MRP solutions then have to do some post-processing to try to smooth out the "mountain peaks" (capacity or material peaks) into "valleys" to respect constraints, but these are not quality executable solutions. In other words, SAP believes that the unconstrained scheduling solutions are suitable for high-level planning tasks such as sales an operations planning (S&OP), but they do not generate operating plans of high quality.

Like Kinaxis RapidResponse, ICON-SCM offers a user interface (UI) to allow users to make judgments and adjustments to the plan. According to SAP, the difference with Kinaxis is that ICON-SCM starts from a higher quality plan, since it is the result of a near optimal heuristic, while Kinaxis starts from an adjusted fast-MRP plan, which requires more adjustment before it can be operational.

On the other hand, Capital-O Optimization tends to be slow, and not suitable for never-ending what-if scenarios in highly volatile environments. Fast MRP and "Search" are fast, as their name implies, often run in-memory, and thus are more useful for evaluating what-if situations. While you sacrifice quality (perfection) of the plan, you can still get a "feasible/infeasible" or go/no-go answer quickly.

In concrete terms, Capital-O Optimization (i.e., SAP APO) is well suited for process and semi-process manufacturing, such as consumer products. It is also suited for situations where cost minimization is critical, i.e., where transportation costs take a large piece of the margin, and therefore the sourcing is based predominantly on distance travelled. For its part, "Search"  (i.e., SAP SCRM) is better suited for items with more extensive bills of materials (BOMs), multiple capacity and/or material constraints, long supply lead times, multiple supply alternatives, and extreme customer competition, where prioritizing customers and allocating supply are key processes.

These Response Management systems start with demand priority management by applying flexible priority rules to demand planning. Then they match supply and demand by efficiently planning deliveries and planned orders, and fully pegging orders and material requirements throughout all tiers of the supply network. Finally, the what-if analysis must be transparent, easily understood, and explainable. Reporting outputs are typically insightful end-to-end pegging and planning reports.

Essentially, SAP is recommending the aforementioned SNP Optimization module within SAP APO in industries where true cost optimization is important, and is recommending SCRM where it is not. There is some overlap between SCRM and the SAP Capable-to-Match (SAP CTM) capability within SAP APO, but CTM is a more robust solution for complex characteristic-based "down-binning" (rules-based product substitution) situations in the semiconductor space. SAP continues to recommend CTM in those industries, but is seeing greater demand for SCRM where those exacting requirements are not present.

Kinaxis RapidResponse vs. SAP APO

Apart from their different approaches to optimization, and thus the products' varying level of appropriateness for certain audiences, another likely reason why SAP did not embed Kinaxis as a globally more established response management provider (in addition to possible ego clashes between the two companies) is because Kinaxis covers a larger scope than SAP SCRM by ICON. Kinaxis has one solution that covers S&OP, demand planning, supply planning, multi-enterprise master production schedule (MPS), new product introduction (NPI), capacity planning, inventory management, supplier collaboration, demand collaboration/collaborative planning, forecasting & replenishment  (CPFR), CTP, clear to build, line of balance, expiry data analysis, alternate sourcing, yield analysis, part substitution, inventory allocation, etc.In other words, Kinaxis RapidResponse is closer in functional scope to SAP APO or JDA Planner than to SAP SCRM by ICON (and Oracle Rapid Planner for that matter). Basically, Kinaxis plays in both the APS and Response Management game.

SAP APO offers a variety of methods including fast-MRP, rules-based heuristics, and optimization. These algorithms browse through all of the available resources (machines, labor, suppliers, available time, etc.) and match them with the demand in order to generate replenishment orders. The core APO modules include demand planning, SNP, production planning and detailed scheduling, supply network collaboration, global available to promise (ATP), and common interface function (to provide a common data structure across all the modules).

One could create a similar list of modules and capabilities for the APS offerings by JDA Software, Logility, Infor (the Infor Advanced Planner and Infor Advanced Scheduler products), Adexa, Oracle, AspenTech, Preactor, WAM Systems, etc. The traditional problem for users has been to understand the business logic of each rule (algorithm), and how it affects the other functions in the enterprise, not to mention the correlation between all of these modules and their rules when they are implemented concurrently. Thus, the system is typically unable to alert users in other functions that there is an issue that is important to them or to solicit their input.

How Much Human Intervention?

More importantly, any optimized plan is based upon assumptions, in many cases about the relative importance of different metrics such as inventory levels vs. customer service levels that can’t even be represented in the same units of measure (UOM). Moreover, these assumptions change over time, but the changes are all too often not reflected in the optimization model because of the major effort required to make the changes.

Fundamentally, though, Kinaxis believes that human judgment is just as important as machine intelligence. Humans are far better at understanding nuance and ambiguity and using this to reach a compromise. Nearly all decisions, but especially cross-functional decisions, are about competing objectives and therefore can only be resolved through compromise. Kinaxis' approach is to provide humans with very fast analytics that allow them to quickly achieve a very good understanding of the likely consequences of their actions.

Kinaxis feels that making the right choice from among a set of possible actions is best left to humans. In contrast, optimization-based APS solutions give the human the answer, when in truth it is just one answer of many. In other words, they don't present a set of alternatives. But because machines have a restricted “span of experience” and can only investigate alternatives within their span of experience, choosing between alternatives is exactly where humans always outperform machines.

In addition, Kinaxis RapidResponse can do a full "as planned" cost analysis, including the ability to model time-phased costs, whereas most other vendors can only support unit-level planning, with perhaps some past analysis for the financials. Kinaxis heuristics are designed to minimize lateness, but demand can be prioritized. For example, if demand is prioritized by margin contribution, the supply plan generated will result in the greatest margin with the least lateness.

Contrary to SAP's claims, Kinaxis also has finite materials and capacity capabilities, as it could not survive in the high-tech contract manufacturers’ space without finite material capability, or in the auto suppliers space without finite capacity. Admittedly, Kinaxis does not do detailed capacity scheduling and sequencing, but most of its customers are trying to solve the problem of what happens between rather than within facilities. Other areas that Kinaxis does not cover are transportation management and warehousing.

It appears that the biggest conceptual difference between the Kinaxis approach and that of the other vendors in this space is that the latter believe that a perfect/optimal plan can be created if only one tries hard enough. And that once the (im)perfect plan has been generated, the organization just needs to execute according to the plan, even to the point of promoting the use of Plan Conformance as a key metric for measuring supply chain performance.

In contrast, Kinaxis believes that everyone has to plan, but that planning is only the starting point because the plan is never perfect. Since the plan is never perfect it is those other capabilities of monitoring and responding that provide breakthrough performance.

Time Alone Will Tell...

It will be interesting to watch how currently upbeat Kinaxis will compete with SAP, Oracle, Logility, and JDA down the track. The strength of RapidResponse lies in supply chain environments that undergo constant change. As mentioned in Part 1, the roots of the initial deployments of the application lie in the high tech, consumer electronics, and contract manufacturing sectors where forecasts are hardly ever accurate and change is a daily occurrence.

As also mentioned in Part 1, Kinaxis has been able to successfully build on these capabilities and penetrate other industrial verticals or organizations with similar characteristics. In addition, the solution has expanded into some horizontal areas such as S&OP, profit and loss (P&L) scenario simulations, workforce management (WFM) scenarios, project management (as to avoid costly contractual penalties), etc.

RapidRasponse is currently more flexible than its APS competitors in user interaction with the user interface (UI) and in utilizing the various decision-support functionalities. When one speaks with multiple Kinaxis users, they all indicate that the application is easy to use and grasp, and provides lots of flexibility in methods to accomplish more rapid sequential and scenario-based planning.

Dear readers, what are your views, comments, opinions, etc. about the Rapid Response and APS software market in general? We would also be interested in your experiences with these software categories (if you are an existing user) or with your current (possibly ineffective) practices, and your general interest to evaluate these solutions as prospective customers.
comments powered by Disqus