In a recent LinkedIn post, I detailed four questions that, when answered, will reveal how forecasts are being used in your business.  In this article, we’ve listed questions you can ask that will reveal how forecasts are created.

1. When we ask users how they create forecasts, their answer will often be “we use history.” This obviously isn’t enough information, as there are different types of demand history that require different forecasting methods. If you are using historical data, then make sure to find out if you are using an averaging model, a trending model, a seasonal model, or something else to forecast.

2. Once you know the model used, ask about the parameter values of those models. The forecast output of an “average” will differ, sometimes significantly, depending on the number of periods you are averaging.  So, find out whether you are using an average of the last 3 months, 6 months, 12 months, etc.

3. If you are using trending models, ask how the model weights are set. For example, in a trending model, such as double exponential smoothing, the forecasts will differ significantly depending on how the calculations weight recent data compared to older data (higher weights put more emphasis on the recent data).

4. If you are using seasonal models, the forecast results are going to be impacted by the “level” and “trending weights” used. You should also determine whether seasonal periods are forecasted with multiplicative or additive seasonality.  (Additive seasonality says, e.g., “Add 100 units for July”, whereas multiplicative seasonality says “Multiply by 1.25 for July.”) Finally, you may not be using these types of methods at all.  Some practitioners will use a forecast method that simply averages prior periods (i.e., next June will be forecasted based on the average of the prior three Junes).

5. How do you go about choosing one model over another? Does the choice of technique depend on the type of demand data or when new demand data are available? Is this process automated? Or if a planner chooses a trend model subjectively, will that item continue to be forecasted with that model until the planner changes it again?

6. Are your forecasts “fully automatic,” so that trend and/or seasonality are detected automatically? Or are your forecasts dependent on item classifications that must be maintained by users? The latter requires more time and attention from planners to define what behavior constitutes trend, seasonality, etc.

7. What are the item classification rules used? For example, an item may be considered a trending item if demand increases by more than 5% period-over-period. An item may be considered seasonal if 70% or more of the annual demand occurs in four or fewer periods. Such rules are user-defined and often require overly broad assumptions. Sometimes they are configured when a system was originally implemented but never revised even as conditions change. It’s important to make sure any classification rules are understood and, if necessary, updated.

8. Does the forecast regenerate automatically when new data are available, or do you have to manually regenerate the forecasts?

9. Do you check for any change in forecast from one period to the next before deciding whether to use the new forecast? Or do you default to the new forecast?

10. How are forecast overrides that were made in prior planning cycles treated when a new forecast is created? Are they reused or replaced?

11. How do you incorporate forecasts made by your sales team or by your customers? Do these forecasts replace the baseline forecast, or do you use these inputs to make planner overrides to the baseline forecast?

12. Under what circumstances would you ignore the baseline forecast and use exactly what sales or customers are telling you?

13. If you rely on customer forecasts, what do you do about customers who don’t provide forecasts?

14. How do you document the effectiveness of your forecasting approach?  Most companies only measure the accuracy of the final forecast that is submitted to the ERP system, if they measure anything. But they don’t assess alternative predictions that might have been used. It is important to compare what you are doing to benchmarks. For example, do the methods you are using outperform a naïve forecast (i.e., “tomorrow equals today,” which requires no thought), or what you saw last year, or the average of the last 12 months.  Benchmarking your baseline forecast insures you are squeezing as much accuracy as possible out of the data.

15. Do you measure whether overrides from sales, customers, and planners are making the forecast better or worse? This is just as important as measuring whether your statistical approaches are outperforming the naïve method.  If you don’t know whether overrides are helping or hurting, the business can’t get better at forecasting – you need to know which steps are adding value so that you can do more of those and get even better. If you aren’t documenting forecast accuracy and conducting “forecast value add” analysis, then you aren’t able to properly assess whether the forecasts being produced are the best you could make.  You’ll miss opportunities to improve the process, increase accuracy, and educate the business on what type of forecast error is to be expected.

We often encounter Excel-based reorder point planning methods.  In this post, we’ve detailed an approach that a customer used prior to proceeding with Smart.  We describe how their spreadsheet worked, the statistical approaches it relied on, the steps planners went through each planning cycle, and their stated motivations for using (and really liking) this internally developed spreadsheet.

Their monthly process consisted of updating a new month of actuals into the “reorder point sheet.”  An embedded formula recomputed the Reorder Point (ROP) and order-up-to (Max) level.  It worked like this:

• ROP = LT Demand + Safety Stock
• LT Demand = average daily demand x lead time days (assumed constant to keep things simple)
• Safety Stock for long lead time parts = Standard deviation x 2.0
• Safety Stock for short lead time parts = Standard deviation x 1.2
• Max = ROP + supplier-dictated Minimum Order Quantity

Historical averages and standard deviations used 52-weeks of rolling history (i.e., the newest week replaced the oldest week each period).  The standard deviation of demand was computed using the “stdevp” function in Excel.

Every month, a new ROP was recomputed. Both the average demand and standard deviation were modified by the new week’s demand, which in turn updated the ROP.

The default ROP is always based on the above logic. However, planners would make changes under certain conditions:

1. Planners would increase the Min for inexpensive parts to reduce risk of taking an on-time delivery hit (OTD) on an inexpensive part.

2. The Excel sheet identified any part with a newly calculated ROP that was ± 20% different from the current ROP.

3. Planners reviewed parts that exceed the exception threshold, proposed changes, and got a manager to approve.

4. Planners reviewed items with OTD hits and increased the ROP based on their intuition. Planners continued to monitor those parts for several periods and lowered the ROP when they felt it is safe.

5. Once the ROP and Max quantity were determined, the file of revised results was sent to IT, who uploaded into their ERP.

6. The ERP system then managed daily replenishment and order management.

Objectively, this was perhaps an above-average approach to inventory management. For instance, some companies are unaware of the link between demand variability and safety stock requirements and rely on rule of methods or intuition exclusively.  However,  there are problems with their approach:

1. Manual data updates
The spreadsheets required manual updating. To recompute, multiple steps were required, each with their own dependency. First, a data dump needed to be run from the ERP system.  Second, a planner would need to open the spreadsheet and review it to make sure the data imported properly.  Third, they needed to review output to make sure it calculated as expected.  Fourth, manual steps were required to push the results back to the ERP system.

2. One Size Fits All Safety Stock
Or in this case, “one of two sizes fit all”. The choice of using 2x and 1.2x standard deviation for long and short lead time items respectively equates to service levels of 97.7% and 88.4%.    This is a big problem since it stands to reason that not every part in each group requires the same service level.  Some parts will have higher stock out pain than others and vice versa. Service levels should therefore be specified accordingly and be commensurate with the importance of the item.  We discovered that they were experiencing OTD hits on roughly 20% of their critical spare parts which necessitated manual overrides of the ROP.  The root cause was that on all short lead time items they they were planning for an 88.4% service level target. So, the best they could have gotten was to stock out 12% of the time even if “on plan.”   It would have been better to plan service level targets according to the importance of the part.

3. Safety stock is inaccurate.  The items being planned for this company are spare parts to support diagnostic equipment.  The demand on most of these parts is very intermittent and sporadic.  So, the choice of using an average to compute lead time demand wasn’t unreasonable if you accept the need for ignoring variability in lead times.  However, the reliance on a Normal distribution to determine the safety stock was a big mistake that resulted in inaccurate safety stocks.  The company stated that its service levels for long lead time items ran in the 90% range compared to their target of 97.7%, and that they made up the difference with expedites.  Achieved service levels for shorter lead time items were about 80%, despite being targeted for 88.4%.    They computed safety stock incorrectly because their demand isn’t “bell shaped” yet they picked safety stocks assuming they were.  This simplification results in missing service level targets, forcing the manual review of many items that then need to be manually “monitored for several periods” by a planner.  Wouldn’t it be better to make sure the reorder point met the exact service level you wanted from the start?  This would ensure you hit your service levels while minimizing unneeded manual intervention.

There is a fourth issue that didn’t make the list but is worth mentioning.  The spreadsheet was unable to track trend or seasonal patterns.  Historical averages ignore trend and seasonality, so the cumulative demand over lead time used in the ROP will be substantially less accurate for trending or seasonal parts. The planning team acknowledged this but didn’t feel it was a legitimate issue, reasoning that most of the demand was intermittent and didn’t have seasonality.  It is important for the model to pick up on trend and seasonality on intermittent data if it exists, but we didn’t find their data exhibited these patterns.  So, we agreed that this wasn’t an issue for them.  But as planning tempo increases to the point that demand is bucketed daily, even intermittent demand very often turns out to have day-of-week and sometimes week-of-month seasonality. If you don’t run at a higher frequency now, be aware that you may be forced to do so soon to keep up with more agile competition. At that point, spreadsheet-based processing will just not be able to keep up.

In conclusion, don’t use spreadsheets. They are not conducive to meaningful what-if analyses, they are too labor-intensive, and the underlying logic must be dumbed down to process quickly enough to be useful.  In short, go with purpose-built solutions. And make sure they run in the cloud.

Smart IP&O is powered by the SmartForecasts® forecasting engine that automatically selects the most appropriate method for each item.  Smart Forecast methods are listed below:

• Simple Moving Average and Single Exponential Smoothing for flat, noisy data
• Linear Moving Average and Double Exponential Smoothing for trending data
• Winters Additive and Winters Multiplicative for seasonal and seasonal & trending data.

This blog explains how each model works using time plots of historical and forecast data.  It outlines how to go about choosing which model to use.   The examples below show the same history, in red, forecasted with each method, in dark green, compared to the Smart-chosen winning method, in light green.

Seasonality
If you want to force (or prevent) seasonality to show in the forecast, then choose Winters models.  Both methods require 2 full years of history.

`Winter’s multiplicative will determine the size of the peaks or valleys of seasonal effects based on a percentage difference from a trending average volume.  It is not a good fit for very low volume items due to division by zero when determining that percentage. Note in the image below that the large percentage drop in seasonal demand in the history is being projected to continue over the forecast horizon making it look like there isn’t any seasonal demand despite using a seasonal method.

Statistical forecast produced with Winter’s multiplicative method. 

Winter’s additive will determine the size of the peaks or valleys of seasonal effects based on a unit difference from the average volume.  It is not a good fit if there’s significant trend to the data.  Note in the image below that seasonality is now being forecasted based on the average unit change in seasonality. So, the forecast still clearly reflects the seasonal pattern despite the down trend in both the level and seasonal peaks/valleys.

Statistical forecast produced with Winter’s additive method.

Trend

If you want to force (or prevent) trend up or down to show in the forecast, then restrict the chosen methods to (or remove the methods of) Linear Moving Average and Double Exponential Smoothing.

 Double exponential smoothing will pick up on a long-term trend.  It is not a good fit if there are few historical data points.

Statistical forecast produced with Double Exponential Smoothing

Linear moving average will pick up on nearer term trends.  It is not a good fit for highly volatile data

Non-Trending and Non-Seasonal Data
If you want to force (or prevent) an average from showing in the forecast, then restrict the chosen methods to (or remove the methods of) Simple Moving Average and Single Exponential Smoothing.

Single exponential smoothing will weigh the most recent data more heavily and produce a flat-line forecast.  It is not a good fit for trending or seasonal data.

Statistical forecast using Single Exponential Smoothing

Simple moving average will find an average for each period, sometimes appearing to wiggle, and better for longer-term averaging.  It is not a good fit for trending or seasonal data.

Statistical forecast using Simple Moving Average

I’ll bet your maintenance and repair teams would be ok with incurring higher stock out risks one some spare parts if they knew that the inventory reduction savings would be used to spread out the inventory investment more effectively to other parts and boost overall service levels.

I’ll double down that your Finance team, despite always being challenged with lowering costs, would support a healthy inventory increase if they could clearly see that the revenue benefits from increased uptime, fewer expedites, and service level improvements clearly outweighed the additional inventory costs and risk.

A spare parts tradeoff curve will enable service parts planning teams to properly communicate the risks and costs of each inventory decision.  It is mission critical for parts planning and the only way to adjust stocking parameters proactively and accurately for each part.  Without it, planners, for all intents and purposes, are “planning” with blinders on because they won’t be able to communicate the true tradeoffs associated with stocking decisions.

For example, if a proposed increase to the min/max levels of an important commodity group of service parts is recommended, how do you know whether the increase is too high or too low or just right?  How can you fine-tune the change for thousands of spares?  You won’t and you can’t.  Your inventory decision making will rely on reactive, gut feel, and broad-brush decisions causing service levels to suffer and inventory costs to balloon.

So, what exactly is a spare parts tradeoff curve anyway?

It’s a fact-based, numerically driven prediction that details how changes in stocking levels will influence inventory value, holding costs, and service levels.  For each unit change in inventory level there is a cost and a benefit.  The spare parts tradeoff curve identifies these costs and benefits across different stocking levels. It lets planners discover the stock level that best balances the costs and benefits for each individual item.

Here are two simplified examples. In Figure 1, the spare parts tradeoff curve shows how the service level (probability of not stocking out) changes depending on the reorder level.  The higher the reorder level, the lower the stockout risk.  It is critical to know how much service you are gaining given the inventory investment.  Here you may be able to justify that an inventory increase from a reorder point of 35 to 45 is well worth the investment of 10 additional units of stock because service levels jumps from just under 70% to 90%, cutting your stockout risk for the spare part from 30% to 10%!

​

Figure 1: Cost versus Service Level

Figure 2: Service Level versus Size of Inventory

In this example (Figure 2), the tradeoff curve exposes a common problem with spare parts inventory.  Often stock levels are so high that they generate negative returns.  After a certain stocking quantity, each additional unit of stock does not buy more benefit in the form of a higher service level.  Inventory decreases can be justified when it is clear the stock level is well past the point of diminishing returns. An accurate tradeoff curve will expose the point where it is no longer advantageous to add stock.

By leveraging #probabilisticforecasting to drive parts planning, you can communicate these tradeoffs accurately, do so at scale across hundreds of thousands of parts, avoid bad inventory decisions, and balance service levels and costs.  At Smart Software, we specialize in helping spare parts planners, Directors of Materials Management, and financial executives managing MRO, spare parts, and aftermarket parts to understand and exploit these relationships.

Sometimes a statistical forecast just doesn’t make sense.  Every forecaster has been there.  They may double-check that the data was input correctly or review the model settings but are still left scratching their head over why the forecast looks very unlike the demand history.   When the occasional forecast doesn’t make sense, it can erode confidence in the entire statistical forecasting process.

This blog will help a layman understand what the Smart statistical models are and how they are chosen automatically.  It will address how that choice sometimes fails, how you can know if it did, and what you can do to ensure that the forecasts can always be justified.  It’s important to know to expect, and how to catch the exceptions so you can rely on your forecasting system.

How methods are chosen automatically

The criteria to automatically choose one statistical method out of a set is based on which method came closest to correctly predicting held-out history.  Earlier history is passed to each method and the result is compared to actuals to find the one that came closest overall.  That automatically chosen method is then fed all the history to produce the forecast. Check out this blog to learn more about the model selection https://smartcorp.com/uncategorized/statistical-forecasting-how-automatic-method-selection-works/

For most time series, this process can capture trends, seasonality, and average volume accurately. But sometimes a chosen method comes mathematically closest to predicting the held-out history but doesn’t project it forward in a way that makes sense.  That means the system selected method isn’t best and for some “hard to forecast”

Hard to forecast items

Hard to forecast items may have large, unpredictable spikes in demand, or typically no demand but random irregular blips, or unusual recent activity.  Noise in the data sometimes randomly wanders up or down, and the automated best-pick method might forecast a runaway trend or a grind into zero.  It will do worse than common sense and in a small percentage of any reasonably varied group of items.  So, you will need to identify these cases and respond by overriding the forecast or changing the forecast inputs.

How to find the exceptions

Best practice is to filter or sort the forecasted items to identify those where the sum of the forecast over the next year is significantly different than the corresponding history last year.  The forecast sum may be much lower than the history or vice versa.  Use supplied metrics to identify these items; then you can choose to apply overrides to the forecast or modify the forecast settings.

How to fix the exceptions

Often when the forecast seems odd, an averaging method, like Single Exponential Smoothing or even a simple average using Freestyle, will produce a more reasonable forecast.  If trend is possibly valid, you can remove only seasonal methods to avoid a falsely seasonal result.  Or do the opposite and use only seasonal methods if seasonality is expected but wasn’t projected in the default forecast.  You can use the what-if features to create any number of forecasts, evaluate & compare, and continue to fine tune the settings until you are comfortable with the forecast.

Cleaning the history, with or without changing the automatic method selection, is also effective at producing reasonable forecasts. You can embed forecast parameters to reduce the amount of history used to forecast those items or the number of periods passed into the algorithm so earlier, outdated history is no longer considered.  You can edit spikes or drops in the demand history that are known anomalies so they don’t influence the outcome.  You can also work with the Smart team to implement automatic outlier detection and removal so that data prior to being forecasted is already cleansed of these anomalies.

If the demand is truly intermittent, it is going to be nearly impossible to forecast “accurately” per period. If a level-loading average is not acceptable, handling the item by setting inventory policy with a lead time forecast can be effective.  Alternatively, you may choose to use “same as last year” models which while not prone to accuracy will be generally accepted by the business given the alternatives forecasts.

Finally, if the item was introduced so recently that the algorithms do not have enough input to accurately forecast, a simple average or manual forecast may be best.  You can identify new items by filtering on the number of historical periods.

Manual selection of methods

Once you have identified rows where the forecast doesn’t make sense to the human eye, you can choose a smaller subset of all methods to allow into the forecast run and compare to history.  Smart will allow you to use a restricted set of methods just for one forecast run or embed the restricted set to use for all forecast runs going forward. Different methods will project the history into the future in different ways.  Having a sense of how each works will help you choose which to allow.

Rely on your forecasting tool

The more you use Smart period over period to embed your decisions about how to forecast and what historical data to consider, the less often you will face exceptions as described in this blog.  Entering forecast parameters is a manageable task when starting with critical or high impact items.  Even if you don’t embed any manual decisions on forecast methods, the forecast re-runs every period with new data. So, an item with an odd result today can become easily forecastable in time.