My recent series on how to plan and manage in uncertainty and volatility (which conditions have become the “new normal” in many sectors and industries) has generated much interest and many comments. As mentioned in the series, the inspiration came from Kinaxis customers’ case studies presented during the Kinexions 2011 user conference.
Ottawa (Canada)-based Kinaxis has been experiencing a renaissance of sorts lately in these days of dispersed complex supply networks and outsourced and offshore manufacturing (with so-called brand owners and their vast network of suppliers). After over 25 years in existence, and some name changes for both the company and its products since the inception, it is not exactly easy to explain what Kinaxis offers (or even better, where its capabilities start and end in the realm of supply chain management [SCM]).
In a nutshell, the vendor delivers a multi-enterprise supply network planning and fast-acting “what-if” simulation solution for achieving best possible fulfillment decisions within the order-to-delivery lead time. The solution’s current name is RapidResponse Control Tower, which can be offered either on-premise or as an on-demand subscription service. Many customers use Kinaxis to solve issues outside of the lead time, but they gain a lot of agility and flexibility by using RapidResponse within the lead time.
Kinaxis customers use this scalable in-memory (or memory-resident) solution for both longer-term and tactical (near real-time) demand and supply balancing. Based on a number of fast-acting deterministic algorithms, the RapidResponse product provides personal alerts, multi-site visibility, and collaborative analysis functions to help businesses with forecasting, planning, and supply chain performance improvement. The company primarily markets to original equipment manufacturer (OEM) clients in industries with sophisticated supply chains such as the aerospace & defense (A&D), automotive, consumer products, industrial equipment, life sciences, and high-tech sectors.
Many large discrete manufacturing companies with complex supply chain networks and volatile business environments rely on RapidResponse for collaborative planning, continuous performance management, and coordinated response to plan variances across multiple areas of the business. These functional areas include supply chain planning (SCP), demand management, sales and operations planning (S&OP), collaborative planning, forecasting and replenishment (CPFR), project management, workforce optimization, and profitability management. By transcending and replacing disparate planning and performance management tools, Kinaxis customers can realize significant operations performance breakthroughs, because from a single system (control tower, if you will), they can make decisions quickly, collaboratively, and in line with the shared business objectives of multiple stakeholders (trading partners).
These enterprises value the ability to analyze alternative response actions before making a decision about a major trade-off. By proactively modeling and scoring different response alternatives, brand owners can communicate a well-understood and optimal action to their suppliers or contract manufacturers. Some high-profile customers of Kinaxis include Cisco Systems, Honeywell, Toshiba, Avaya, Nikon, Qualcomm, Jabil Circuit, and Raytheon.
Market Validation, at Long Last
Kinaxis is currently growing rapidly (no pun intended) in terms of international expansion, adding new customers (or its existing customers expanding their deployment of RapidResponse to new application areas and adding more sites and users in the process), aggressive hiring, rumors of going public, etc. The company’s blog and related supply chain community have been replete with great discussions in the realm of general SCM themes, and not only about the Response Management niche. There is simply an air of confidence, if not cockiness, around Kinaxis, as noted in Lora Cecere’s Supply Chain Shaman blog post on Kinexions 2011.
But, it hasn’t always been a smooth ride for Kinaxis, or whatever its previous name was during those difficult times in the early 2000s. The reasons for tougher times were multiple, starting with the company’s difficulty in carving out an acknowledged niche in the SCM space.
For a long time, Kinaxis (or Webplan as it was known at that time) was eclipsed by the once “all the rage” advanced planning and scheduling (APS) systems and solutions by former Manugistics (whose former staffers now privately admit to me that Kinaxis was giving a run for its money to their former NetWORKS Planning product line, now part of JDA Demand Planner) and i2 Technologies. Ironically, many ex-i2 and ex-Manugistics employees are now working at Kinaxis.
The difficult times in the high-tech sector in the early 2000s (after the Internet bubble) were an additional hurdle. The roots of the initial adopters of RapidResponse lie in the high-tech, consumer electronics, and contract manufacturing sectors, where forecasts are hardly ever accurate and change is a daily occurrence. Kinaxis was eventually able to successfully build on these capabilities and penetrate other industrial verticals or organizations with similar characteristics (e.g., A&D).
Enterprise resource planning (ERP) vendors were not much of help to Kinaxis, and were not allies either, even though RapidResponse complements their solutions nicely in the realm of operational (re)planning and execution. Most likely, these vendors did not want to admit their limited operational planning and execution capabilities, especially in a multi-enterprise setup (and Kinaxis was also partly to blame for putting down ERP systems instead of trying to build a win-win situation). In any case, RapidResponse is now able to layer on top of virtually any most commonly used ERP system and emulate its materials requirements planning (MRP) logic and re-ordering policies (to a less than two percent discrepancy), and then act as a multi-enterprise brain and produce multi-enterprise master production schedule (MPS), available to promise (ATP), capable to promise (CTP), line balancing, etc., all with incredibly fast re-planning capabilities.
SAP Relationship (NOT)
Not surprisingly, many of Kinaxis’ customers are SAP ERP customers too, but that hasn’t apparently been a good enough reason for the two vendors to have a strategic relationship. For a long time, SAP was touting its well-known SAP Advanced Planner and Optimizer (SAP APO) product suite as the panacea for these situations. As some background, SAP introduced APO in the late 1990s as the counterpart APS offering to then thriving i2 and Manugistics. For more information on the scope of SAP APO, see TEC’s article from 1999 entitled “SAP APO: Will it Fill the Gap?”
However, there have been numerous dissatisfied SAP APO customers who have had to do a lot of manual tweaking to quickly generate a feasible plan and determine order priorities, inventory allocations, and commitments. A number of these customers, who are operating in highly volatile demand arenas (and with inaccurate forecasts from the word go) and deal with many trading partners, are seriously looking at adopting response management capabilities, rather than APS.
Generally speaking, the key capabilities that are required for so-called “Response Management” offerings are the following: multi-user input and collaboration, multi-scenario creation and comparison, and high-speed analytics. These capabilities are essential for companies to quickly react to unexpected events such as a rush order from a very important customer (or, conversely, a major last minute order cancellation), a product quality issue, a supply shortage, or a production line breakdown. SAP APO does not support many of these capabilities, so SAP has had to look for a partner to satisfy these key business needs.
Since November 2010, SAP has been distributing a supply chain solution by a lesser know German vendor ICON as SAP Supply Chain Response Management (SAP SCRM) by ICON, a solution extension to its own SCM suite, SAP SCM (look for TEC’s upcoming 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 called Oracle Rapid Planner in 2009, which can be layered on top of Oracle’s ERP products and other ERP products.
SAP chose ICON over Kinaxis on the grounds that is has better finite resources optimization capabilities, but I find this line of thinking hard to swallow in its entirety. Kinaxis has been a Response Management pioneer of sorts, and–not to knock ICON’s benefits and capabilities–it’s tough to believe anyone could best the “plan, monitor, and respond” approach and scalability that Kinaxis has been promoting so successfully. In addition, isn’t “optimization” part of the SAP APO name, and if so then why have a separate SCRM solution?
Also, given that SAP HANA in-memory capabilities are embedded in almost everything within SAP (see TEC’s article entitled “SAP HANA - One Technology to Watch in 2012″), why not use HANA for responses management as well? Kinaxis’ take here is that columnar in-memory databases (IMDBs) like HANA are not necessarily the best design for handling concurrent what-if scenarios by thousands of users. Instead, nearly three decades ago Kinaxis opted for its proprietary hierarchical IMDB, which is an imbedded component of RapidResponse Control Tower, whose presentation layer emulates a traditional relational database behavior. Reporting outputs are typically insightful end-to-end pegging and planning reports.
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. Part 2 will try to draw some demarcation lines between APS and Response Management by using SAP’s solutions applicability (i.e., APO vs. SCRM).
In the meantime, your views, comments, opinions, etc. about the APS and Response Management software are welcome as usual. 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.
Supply chain optimization is not the problem, it is the execution that is the real bottleneck. The issue in the supply chain is customers try to do all kinds of optimizations and expect suppliers to respond. It is better to collaborate and provide realtime visibility instead of just one sided optimization.
The optimization and respond management works well in long lead time parts operating in a kind of batch or totally outsourced environment, which probably where Kinaxis play
i need such the software to work all module in erp
Optimization is one method that can be applied to planning. However, it does not define planning. Most companies that use planning software, do not use an optimizer.
Some vendors make optimization much more difficult than others. SAP is one of them. However, I have seen many lesser known vendors who do a great job with optimization. Many vendors have evolved past using cost optimization for every problem, and we now have inventory optimization and duration optimization, which are customized for their supply chain domains. This is a very important development. Customization of the optimizer per supply chain domain is one of the main lessons of the history of supply chain optimization. However, most companies don’t have access to these new approaches, in part because they are poorly advised to buy uncompetitive applications by the large consulting companies.
There will always be a need for planning, and in fact there is no real way around it. Being responsive is good, but lead times are a reality. In fact, there may be a false dichotomy at work when contrasting responsiveness with planning. What puts a company in a position to be responsive is in fact good planning. Poor software selection in the planning space should not be used as a logical jumping off point to declare that its time to move on to something new.
A lot more could be done to improve planning implementations. Choosing better software is a good place to start.
I don’t think PJ’s article suggests that planning is not necessary. The question at hand is to what extent the plan is a close match to reality. If it is a close match on a regular basis then there is a role for optimization. If on a consistent basis there is little match between the plan and reality it is better to focus on responsiveness.
Many studies, including one in 2011 by Terra Technologies, show that within CPG few companies have better than a 52% accuracy for weekly forecasts. For NPI products the weekly forecast accuracy drops to 35%. (According to Terra, companies that use their demand sensing technology can improve the weekly forcast accuracy of mature products to 70%.)
Since the primary driver for supply plans is the demand plan, how accurate can the supply plan be if the forecast is only 52% accurate. You can apply as much optimization to the problem you want, but even if you create a supply plan that matches the forecast 100%, you supply plan cannot be more than 52% accurate.
But let us address the question of the supply plan model. As you point out, I don’t think it is only SAP that should be singled out in this context. Most vendors try to use linear optimization (LP) to create a supply plan across multiple distribution and manufacturing nodes, But a supply chain is highly non-linear meaning that the LP is always going to be an approximation. In fact the dichotomy is that to make the LP more accurate means that it will take much longer to run. But the issue doesn’t stop with the model accuracy because the input data is often an estimate at best and inaccurate at worst. By input data I mean lead times, yields, costs, capacity, throughput, etc. Even of the input data is 100% accurate, it is really an average value, often with quite large variance with COV’s of greater than 0.25 not being uncommon. So exactly how optimum is the optimized supply plan? (As a side note, we should be scrapping plan conformance as a metric. It is like Einstein’s definition of insanity: ‘Doing the same think over and over again, and expecting a different result.’)
Dick Ling - the ‘father’ of S&OP - and Andy Coldrick capture my opinion best when it comes to the use of optimization in the creation of a supply plan,namely that it should be ‘Roughly right, not precisely wrong’.
But Sean is correct: Everyone HAS to plan.
But it is a false dichotomy to position it as planning OR responsiveness. Companies need planning AND responsiveness.
But what is responsiveness and how can we measure it? In Lean there are 2 terms that I think are key, name ‘Time to Detect’ and ‘Time to Correct’.
As Sean writes, ‘lead times are a reality’, but every minute that is wasted before knowing that the reality is different from the plan, every minute that is wasted before determining the consequence of this mismatch, and every minute that is wasted in evaluating how the supply chain can respond to this mismatch is a minute taken away from the physical supply chain’s ability to respond in a profitable manner. Gartner uses the term ‘demand translation’ in their Demand-Driven Value Network concept to describe this ability.
Undoubtedly there is much that can be done in the physical supply chain to improve agility and responsiveness, but changing poured concrete is expensive and takes a long time. Compressing the ‘Time to Detect’ and ‘Time to Correct’ is where companies can make enormous strides in improving responsiveness quickly.
Even then, what does responsive mean? I’d argue that telling a customer 1 week earlier that you are going to deliver late is being responsive, because this give your customer time to adjust their plans. Being responsive is not about recklessly adjusting the supply plan at any cost to meet customer demand. But undoubtedly one of the primary measures of responsiveness should be on-time delivery to original customer request.
Lastly, the basis of inventory optimization is the idea that we can absorb demand variability with inventory buffers. The problem is that with a 52%, OK, let’s say 70% on the assumption that everyone has deployed sophisticated demand sensing capabilities, forecast accuracy, that is an enourmous amount of inventory that will be required. And that’s assuming that this year’s deamnd aptterns and product portfolio are the same as last years’. When last did that happen?
People reading my comments may come to the conclusion that I am anti-optimization and then be surprised to learn that I studied optimization at the PhD level. This is another false dichotomy because the issue isn’t optimization OR responsiveness.
But I would argue that your next big supply chain breakthrough isn’t going to come from learning to plan better - that is what we have been doing for the last 30 years - but rather from learning to respond profitably to real demand.
So, is the right question “how should a company optimize its responsiveness given its market and its ecosystem?”
That is an excellent way to characterize the approach I would advocate. I would sip the word ‘profitable’ into the description as a way of capturing the trade-offs that need to be made.
Undoubtedly an aspect of that approach should be how to take asset utilization into consideration.
How Much Supply Chain Optimization Do We Really Need? â?? Part 1 » The TEC Blog…
How Much Supply Chain Optimization Do We Really Need? â?? Part 1 » The TEC Blog…