An Inflection Point in Supply Chain Planning?

Interesting Thoughts Generated from Two Days at the Logility User Conference in Atlanta

There are I think two related trends that are sending the traditional world of supply chain planning software in some important new directions.

The first is a partial transition away from the strictly hierarchical, cycle driven approach to planning to one that is more continuous and integrates planning and execution at much deeper levels than we have seen in the past.

The second is the rapid ascendance of advanced analytics as the planning engines from software vendors and indeed companies themselves, versus the traditional planning applications we've had for something like 40 years now, but which really came to the forefront in the 1990s from such companies as i2 Technologies, Manugistics and Logility, among others.

Gilmore Says.... We're moving into an era of algorithmic planning that can consider a much wider and complex range of variables in recommending plans. What do you say? Click here to send us your comments

I make these comments after having just returned from the Logility user conference this week in Atlanta. It was a good couple of days, and triggered some thoughts, which I will relay here. As always, I try to make these "trip reports" of value to all readers, not just Logility users, by generalizing as much as I can.

That said, some quick notes on Logility. The company is a wholly-owned subsidiary of American Software, once a leading player in manufacturing and ERP software - it had the first commercial DRP application, just FYI - but that business continues to dwindle down.

Logility represents roughly 70% of American Software revenues, and is now about an $80 million company, with a small portion of that revenue coming from another division, Demand Solutions, which caters more to the small and medium company planning market (though it has a few large customers as well).

In 2015, Allan Dow became President of Logility after a number of years heading sales and marketing. Earlier this month, he was named president of all of American Software. Logility revenues have grown at an annual rate of 8% over the past three years, not bad in what is clearly a currently slow-growth planning market. It has continued to expand its international sales, and is the planning platform for a growing roster of very large companies (e.g., Verizon, Under Armour, Ferguson), something it to an extent lacked in years past.

So back to my themes. Obviously, there has been much talk in recent years about the rise of analytics, though as usual often without a lot of detail and/or obfuscation of what is really different from what we have had until now.

So, for example, a number of software companies have talked about embedding analytics in their planning applications. But what this really means in most cases is making traditional "descriptive" analytics easier to access as part of the planning process, and in a way that is contextually relevant to the work a planner is doing at a given point in time.

That is all well and good, and may save the planner time and reduce the need to go outside the planning application and do some analysis in a spreadsheet. But this is not the game changing type of analytics I am talking about.

Consider, for example, the type of analytics Intel has built to improve its supply chain decision-making. As presented at last fall's CSCMP conference, Intel's new age analytics includes a "supply-demand solver" that looks out over some horizon, and not only identifies where there is a gap in S&OP plans, but also suggests how the issues might be best resolved. Another called "inventory surveillance" provides a neat visual simulation of how inventory will flow across the supply chain to customers over some period of time, and again identifies likely problem areas.

These are not your father's analytics - this is a whole new generation.

At the Logility conference, Dow and VP of R&D Mark Balte gave a nice overview of where this is all headed. Most of us have probably seen the framework for analytics as popularized by Thomas Davenport that starts with "descriptive" - reporting on what has already happened.

A layer above are diagostics - providing insight into why something happened. If you can understand why something happened, then you may be able to develop predictive analytics that anticipate problems or opportunities. And if you can predict problems or opportunities, you may also be able to move into "prescriptive" analytics - what should a company or a planner do when faced with a given situation?

But the game changer is really taking that last category, prescriptive analytics, and moving it into the brave new world of cognitive systems, based on artificial intelligence.

Balte gave a nice summary of the differences between the traditional planning world and this coming era of cognitive solutions, which I summarized in the table below.

As I hope we can agree, if this does indeed play out as summarized above, this will certainly be a major inflection point in the history of supply chain planning. Logility, for example, noted in this world the software will automatically run through perhaps hundreds or thousands of planning scenarios - something that really has to be initiated manually today, meaning a limited number of scenarios are evaluated in practice.

In this world, after running through this huge range of scenarios, the planning software might select the three or four most likely, relevant or impactful for company executives to consider - and of course likely then also recommend the best decision based on parameters the company has previously determined.

And with machine learning, these new age analytics will simply get smarter over time. With most planning applications today founded either in basic heuristics (e.g., DRP) or linear programming for optimization (e.g., network planning), we're moving into an era of algorithmic planning that can consider a much wider and more complex range of variables in recommending plans.

It is going to be a wild ride, that's for sure, with winners and losers in the software market.

Just a few other notes of general interest from the rest of the conference: Logility chief scientist Sean Willems was entertaining and interesting as usual in his day 2 keynote presentation. Willems is also head of the supply chain analytics program at the University of Tennessee, and literally along with an MIT colleague, invented the algorithm for multi-echelon inventory optimization (MEIO) in the late 1990s or so in his master's thesis. A smart man indeed, who came to Logility when it acquired his company, Optiant, which provided an MEIO solution.

We all need decision "playbooks," Willems said, by which he means a structured way for looking at problems and opportunities. He cited the four-chapter playbook used by Jeff Wilkie, CEO Worldwide Consumer at Amazon and graduate of a master's program at MIT that Willems also teaches at currently.

Wilkie's playbook starts with "constraint management" - a process view of the issue. Next is "statistical process control" - how can Lean and Six Sigma principles be applied? Following that is supply chain - what is the cross-functional view of this issue? Finally, "plant managers matter" - an Amazon fulfillment center is really more like a factory than not, with inputs and outputs. How will this issue/opportunity impact factory performance, with the fulfilment center managers greatly impacting system results?

Very interesting.

I was very struck that during several of the case study breakout sessions, companies were simply limited in what they can do in demand and supply planning by a lack of staff. One well-known, global cosmetics firm cannot get to retailer-specific forecasting because it just doesn't have the human bandwidth, for example, a theme that ran through several of the presentations.

As I mentioned to Dow, Logility's strength and partial weakness is its conservative approach to its financial management, market messaging and more. This is a company that simply doesn't get out ahead of its skis very often, though Dow said the company may get more aggressive in terms of its future vision soon, and part of that was on display in Atlanta.

The company has made a few, relatively modest acquisitions in recent years (e.g., MID Retail), but American Software has no debt and is sitting on some $80 million or so in cash. I think we can expect other deals for complementary solutions in coming years.

Do you think advanced analytics will drive a new era of supply chain planning? Why or why not? Is lack of staff a big issue in demand and supply planning? Let us know your thought at the Feedback section below.

Your Comments/Feedback