Category Archives: Forecasts

SYSPRO: Forecasting and Inventory Optimization for Small & Mid-Sized Businesses, Part I

SYSPRO is an established ERP suite provider (that has been around for over 30 years) that also provides Forecasting, Inventory Optimization, and Warehouse Management solutions in addition to the 40+ other modules that it provides. The Inventory Optimization solution (in SYSPRO 6.1) is one of their newest offerings. It’s built on top of their solid forecasting module and provides SMBs with a good inventory management and optimization solution, especially if they are already running SYSPRO.

While SYSPRO (a Microsoft Gold Certified Partner) isn’t an Oracle or SAP, they are a fairly significant company in the mid-market with over 14K customers in 60 countries that contribute to their 300M footprint. Furthermore, their solution is built on top of the Microsoft .Net platform, integrated with the Microsoft Office suite, and easy to pick up by anyone who is familiar with Microsoft Small Business Products. In particular, if you can use Microsoft Project, you can use their products. This is an appropriate technology stack (and strategy) for most of their target market who are already using Windows and Microsoft (Back Office) Products as it minimizes the new-user learning curve.

Their forecasting solution is quite robust. (For a discussion of forecasting, see the glossary page, which also contains links to some relevant posts.) It allows you to forecast at the individual SKU level and at the product family level, which generally creates more robust long-term forecasts. The forecasting solution can take into account historical data, projected sales, promotions, current stock levels, target stock levels (by location), lead times, and policies and create a (monthly, weekly, or daily) forecast using a variety of algorithms. You can select your preferred algorithm, or let the program choose the algorithm that is the best fit given historical data patterns. The program tracks the current forecast, the draft forecast (revision) under consideration, and the suggested forecast created from the last modelling session, which you can manually alter or revise to create a new draft forecast, which will become the new forecast once approved by an administrator.

The algorithms at your disposal include competition, Holt-Winters additive, Holt-Winters multiplicative, annual seasonal profiles (smoothed and unsmoothed), mean, median, moving average, exponential smoothing (with or without trending), multi-period weighted average (six, twelve, etc.), and a few others. (A good overview of these forecasting models can be found on resample.com.) In each case, the system will generate a forecast and graphically plot it against sales for the last three relevant periods (e.g. if you were forecasting Jan to Dec 2010, it would plot Jan to Dec 2007, Jan to Dec 2008, and Jan to Dec 2009, if available), the current forecast, and the current draft forecast so that you can visually see whether the forecast is in line with historical behavior and what is currently expected. This allows you to quickly spot whether a trend might be out of whack or whether (or not) the revised forecast produces spikes in line with upcoming promotions.

The system will also generate all of the relevant statistical data, including the cumulative forecast error, mean absolute deviation, mean square deviation, mean absolute % error, and tracking signal so that you can check the calculations and understand how much confidence you should have in the result. For each algorithm, it will also allow you to alter any of the controlling parameters (and re-run the forecast at any time). (But you should only do this if you are well versed in the art of forecasting and know what you are doing. However, if you are an expert, it’s great to have all this power to run multiple what-ifs and understand the ripple effects minor deviations in sales trends have on your forecasts, which in turn can effect your optimal inventory strategy.) And, as I noted above, you can do this forecasting at the group / product family level or the individual SKU level. This allows you to quickly generate a robust group forecast and then dive in and alter only those individual SKU forecasts that need to be tweaked to take into account upcoming promotions or new seasonal trends. In addition, you can also restrict the group forecast to any meaningful combination of warehouses, stock codes, suppliers, and product classes — which gives you a lot of power and flexibility in forecast creation. And the more advanced users can set up batch forecasting runs, forecast-over-forecast comparisons, and even Pareto analyses, but this takes us into the realm of inventory optimization, which is the subject of Part II.

In other words, the SYSPRO forecasting module packs a lot of power into a relatively easy to use SMB software solution. And with SYSPRO 6.1, you now get a true Inventory Optimization Solution!

Share This on Linked In

The Basics of Inventory Management, Courtesy of SYSPRO

As a precursor to my future post on SYSPRO’s new Inventory Optimization solution, I thought I’d provide a brief review of their free e-book on Supply Chain, Inventory Management, & Optimization: Skills for Small Businesses, available on request to clients and prospects. While it doesn’t delve deep into inventory (and related supply chain) optimization, it does a great job describing the basics of inventory management and serves as a great introduction to the subject to small and mid-size businesses just beginning to tackle the issue.

When beginning to delve into the issue of inventory management, there are five factors that need to be considered: production, stock, location, transport, and information requirements. Associated with each factor are a number of decisions that need to be made, which are summarized in the following table:

Production capacity

flexibility

facilities

SKU vs Job Lot vs. Cross-Docking

Stock basic vs seasonal vs safety

level

variety

Location supplier proximity

customer proximity

Transportation mode

frequency

flexibility

Information collection

distribution

More specifically,

  • should you centralize production, and increase shipment times to remote locations, or decentralize production and minimize shipment times to any particular customer location?
  • should you maintain high levels of stock to prevent a stock-out, or implement flexible manufacturing and JIT delivery?
  • should you organize inventory by SKU, by Job Lot, or implement Cross-Docking?
  • how does seasonality affect your safety stock levels?
  • ship, rail, truck, air, cableway, pipeline, conveyor, or wire?
  • should you centralize your warehouses, or distribute them?
  • should you implement POS or rely on traditional back-room systems?

The goal is to balance trade-offs to maximize agility, adaptability, and alignment in your supply chain which balances customer service levels and internal operating efficiencies to make sure that you can provide your customers with the right goods, at the right price, at the right time.

As such, you need to be concerned with stock assortment, level, turnover, and associated costs. More specifically, what is the right mix of product at any particular time to maximize turnover and minimize associated costs? Then, you have to acquire the inventory, within working capital constraints, and track real-time utilization to improve future forecasts. This should all be done in accordance with comprehensive inventory management policies, which should be designed to quickly identify and eliminate overstock (before the product spoils or becomes obsolete) and replenish popular items in a timely fashion. These policies can use one or more inventory control methods, which can be manual (like ticker, click, and stub) or automatic (like pos terminals).

And once you’ve got the basics of inventory management down, you can move on to inventory optimization.

Share This on Linked In

Want Better Forecasts? Tie Your Key Stakeholder’s Performance Bonuses To Them!

Share This on Linked In

A recent article on CFO.com on “imperfect futures”, which discussed the great deal of uncertainty in today’s economy and how hard forecasting has become, contained a novel and often overlooked idea for creating more accurate forecasts (on average). See how much your people will bet on them!

Betting, often administered through on-line prediction markets, has famously foretold the results of recent elections and Super Bowl match-ups. and is now being used by some companies to try and create a window into their corporate futures. Some companies, like Ford, have achieved good results with the methodology in preliminary applications in NPD. (Perhaps Ford should have used the methodology more broadly, given the current state of the American automotive sector? But I digress.) Electronic Arts has used it to gauge future release popularity within 2%, four times as good as the usual results achieved from in-person polling.

But I’d take it one step further. I’d make it a key factor in the determination of the bonus of each key stakeholder (who should be coming together from the various business units in the creation of a consensus forecast which combines all of your organizational intelligence) — and penalize them for each percentage point the forecast is off, either-way. You’d get more cooperation that way, since no one would want someone else deciding their fate when they could do something about it. You’d also see more scenarios analyzed in the construction of the forecast, since it would take repeated iterations before you came to a majority agreement. Thoughts?

Recession-Resistant Demand Management Strategies and Tactics

Share This on Linked In

A recent article in i2’s Supply Chain Leader on how resilient companies can manage demand when the past is not a reliable indicator of the future had a great table that compared and contrasted demand tactics and strategies in stable vs. volatile economies that did a great job of summarizing the demand management strategies you need to weather the storm.

The following points from “Recession-Resistant Demand Management Strategies and Tactics” in particular are key:

  • level and seasonality in volatile economy are best determined from the last 2-3 months, not 2-3 years
  • constantly look for patterns that are different from expectations and make adjustments quickly
  • it’s not just about consensus forecasting — triangulate multiple scenarios across functions against leading and trailing indicators to come up with near real-time forecasts
  • combine and push-and-pull strategies to get the best results
  • monitor continuously … today’s patterns will not be tomorrow’s

Six Steps to Better Sales Forecasting and Demand Planning in an Intelligent Enterprise

Share This on Linked In

A good article from last year’s Intelligent Enterprise covered “six steps to better sales forecasting and demand planning”. The process, which should revolve around the four critical components of people, process, information, and technology at its core, is straight-forward and a good guide for any organization that needs to improve its demand planning (since good demand projections are critical to getting good sourcing results).

  1. Conduct an SF/DP process and system assessment
    Start with an independent and unbiased evaluation of your financial and operational performance planning processes and systems that establishes benchmarks for current effectiveness and identifies areas for improvement.
  2. Identify user requirements and project scope
    Define the business requirements, develop clear definitions of information needs (for proper planning), and what product / service lines you will be addressing.
  3. Build a business case that improves value and results
    Decide whether you need better systems, better processes, or both … then, once you’ve quantified the costs, outline the expected improvements in the results and build a business case that will define the expected value and ROI.
  4. Assemble the program and plan
    Once you have approval, it’s time to define the implementation and change management program that will realize the expected benefits. Then determine the communication plan that will insure that each affected individual knows what she has to do, when, to insure project success as well as the value the program will deliver to them (to give them incentive to contribute to the overall success).
  5. Evaluate new technology against the program plan
    As you are implementing the technology, you’ll need to evaluate its effectiveness. This will require specific product evaluation criteria that should be defined in advanced.
  6. Deploy the integrated SF/DP program and set of processes
    Be sure to implement the new processes in a way that minimizes disruptions to the business, culture, and technology infrastructure.

The article also discusses a Maturity Model, the potential impact of failure, and the role that people, process, information, and technology impacts.