I saw the other day on LinkedIn (my least favorite social platform) that @Dr Jakob Beer published an e-book on ABC analysis for warehouse design. It is (aptly) titled “Understanding ABC Analysis for Warehouse Design Decisions.” [link] I had to get it, even braving the German-language checkout page to buy a copy! At 50EUR, it was a little pricey. But considering that the table of contents promised to tell me virtually everything I need to know about ABC analysis, I made the leap.
The quick reaction: it is well worth the investment. Especially if you are in the market for warehouse design or automation implementation, reading this book could easily save you months of work and 5, 6, or 7 figures in cost. I would see the most value if you are an Operations manager (Director, VP) in charge of DCs/FCs, or are purchasing or implementing an automation system.
You probably know about and have used ABC analysis, and much of the book analytical content is probably in industrial engineering curricula. Beer’s unique value is putting a layer of practical insights on the tools that is likely not found in textbooks, and giving a list of what to watch out for in design analysis.
Analysis Creates Value Through Good Requirements
Value in design is created (or lost) in getting requirements right. And ABC analysis – that is, correct ABC analysis—is fundamental to developing good requirements. While ABC analysis is a commonly used tool, it is commonly used incompletely or used with poor data (or both!). Beer also points out that ABC is not the end of the analysis, but the vehicle or tool by which to get to the point where real business decisions can take place.
So being able to spot poor analysis is an invaluable skill for Operations managers. This is because poor analysis leads to poor requirements, and poor requirements get translated into operations where they are not easily fixed.
As examples: I have seen cases where incomplete analysis created unworkable, or simply less operable systems, in projects valued at 10s of Millions of dollars. Beer articulates and addresses the core problems I saw. In one case this was because of the incorrect inclusion of irrelevant corrugate in a solution SKU storage profile. In another, it was incorrect analysis of cube profiles that led to dramatic under-utilization of storage space. And timing is always an issue.
Further, work like this is usually packaged inside of other textbooks or references and is usually dealt with superficially as part of a larger warehousing or industrial engineering survey of knowledge. So to see an in-depth, focused PDF reference to this topic is very helpful because it is accessible.
I also tend to index this sort of work against Dr Ed Frazelle’s “World Class Warehousing.” The profiling section in WCW is worth its weight in gold. Indeed it mentions more types of profiles in its analysis guide than Beer does. But Beer goes over and beyond by talking more about the data and the fundamentals of the toolkit plus real-world cautions and drawbacks, while Frazelle gives more of an analytical checklist for warehouse and inventory storage design and then moves onto different technologies and operating principles.
Project Manager Point Of View
As a consultant on intralogistics project management such as warehouse implementations, I have the opportunity to see and be involved in business case development and warehouse design cycles. The knowledge in this book is both on how to use the analytical tools and, more importantly, how to be properly critical of analysis.
This has two sides. When preparing vendors to offer a solution, Operations teams should know what they’re asking for. This means being able to provide the appropriate data and forecasts to make design recommendations.
Then on the back end, the Operations and engineering teams must be able to critique a vendor’s analysis, ensuring (for example) that the correct items are included, that the time period is relevant, that seasonality is properly captured, that all the required profiles are present and done correctly, and so on. The vendor may either not know about gaps in the analysis, or may be presenting incomplete work.
If an Operations team or supporting consultant (ahem!) is not able to critique an analysis, they risk the success of the project. This book can help develop the acumen and checklist topics to review in this critique.
Foundational Concepts
Beer gives a short history of ABC analysis and how it underpins much analysis in warehouse design. It is a basic, fundamental, powerful tool in deriving the parameters for how to design a system. In fact, ABC analysis – the practice of segmenting flows of the occurrence of something by the effects of that something, more popularly known as a Pareto distribution – is a close cousin of the applying the normal distribution of events and deciding whether to design for a 95th percentile frequency occurrence, 99th percentile, or even a 50th percentile occurrence.
There are a few underlying themes in the book.
Key foundational concepts:
- An analytical framework: Beer describes in ABC analysis a tool for a world-view of operations, a *philosophy* or approach to triaging the huge amounts of data facing the operations manager today, whether on forward logistics, suppliers, errors, or returns. Any dataset should be reviewed through this lens for insights on how to deal with it.
- Orthogonality of analyses: As a framework, the analysis tool can be applied across many different dimensions of activity.
- You can look at data over time, or value, or physical characteristics, or upstream- and downstream- process effects. This analysis can even be applied to activities with Safety consequences. A value view leads to different analysis and consequences than an activity or dimensional or commercial view.
- It is important to understand what you want before the analysis, and ensure that your conclusions follow logically from the analysis you’ve conducted.
- Developing data: Data is critical to this analysis. You can use WMS data, OR Hand-gathered data. Use Audits. Or other tracking. But you have to develop the data! If this means generating audits through systems like @QualVis, then you need to do it!
- But You Need Clean Data: For example, remove materials & duplicate SKUs or interim SKUs developed in manufacturing or kitting processes, or replen movements, or transfers. This type of “catch” is easy for vendor analysts to miss, and hence is why vendor sales analysis can be so dangerous.
Technology and Application Specifics
I enjoyed the discussion of Cube Storage vs ASRS vs Manual Picking in the book. The explanations in the book were clear and most helpful in fitting the creation of charts to how they affected the actual design and application of technology in distribution settings.
Further, Beer gives examples of how the analysis can be used in every-day warehousing ops. He points out that the toolkit can be applied to quality errors, safety conditions, and customer events, to identify main areas of focus for efforts. The specific applications are left as an exercise to the reader based on the reader’s setting.
ABC Analysis Cautions and Drawbacks
Beer helpfully includes a whole sections on shortcomings of the ABC analytical tools. This is essentially the flip side of the flexibility of analysis orthogonality described above: Whenever a decision is made to include an ABC analysis on some parameter (value, cube, activity, time), it necessarily excludes analysis on other variables.
Some specific limitations include:
- Limitations of historical data: The past is not the future! So your data is not exactly what will happen, and you need to take some risk on judgments.
- Timeframe sensitivity will require judgment about what timeframe most closely represents “the” business. The ABC distributions (and SKUs within the different segments) will change over time.
- ABC ignores SKU mix and demand variability (which is addressed later with XYZ and other analyses)
- ABC ignores product interdependencies and affinities.
- ABC does not consider promotions.
- Using different units of analysis will yield different results. E.g. product value vs volume, for example, yield very different profiles.
This means that, especially as it matters for time and SKU-affinity, ABC analysis can give incomplete or missing pictures of activities. The analyst must be careful of how he chooses the inputs and frames the outputs.
Nuts and Bolts of Analysis Included
Beer includes examples of how to complete the ABC analysis for warehouse design. He steps through useful Python and Excel techniques, command-by-command, to get the answers you need.
While the math is quite straightforward, it is nice to have a cookbook to produce consistent results from dataset to dataset.
He also shows some big-dataset techniques for working in Excel which is useful for the ops professional who doesn’t have a data analyst and can’t learn SQL in a short time. Perhaps some other references would be helpful, but this is enough to get someone going if they did not know where to start.
Useful Video Links
The book thoughtfully includes QR codes to videos on how to accomplish the more technical topics. I didn’t review those, but it is nice to have video tutorial links on the exact examples being showed.
Topics For Next Time
It’s easy to Monday-morning quarterback, especially when you’re not the one doing the work. But that said, there are a couple points I think would improve the work.
I would have liked to see more ABC analysis critiques on other Goods-to-person (e.g. Kiva type systems). The analysis on ASRS vs Cube Storage vs Manual warehouse picking was very helpful and illustrative of the principles. A review of Kiva-style GTP combines aspects of all of the above, plus we must consider cycle counting on top of picks in the activities.
While picking gets most of the attention for its high labor content, we can’t sleep on the other areas! What are the ABC applications for receiving and putaway? Are there any technological analogues to the Cube Storage vs ASRS discussion that we should review carefully on the inbound side of the equation?
Another area I’d recommend in any next editions is alternatives. The profiling / ABC analytical technique seems to be the only tool available to engineers for system designs. This is because systems are discrete and, apart from any scaling by adding cube storage or AGVs, are fixed assets. Therefore the design parameters are ultimately deterministic. Will any other techniques (regression or machine learning models, for example) eventually supplant or enhance ABC analysis to drive design of these systems?
Conclusion
Again, I recommend this as a reference: Read it once for the knowledge of what’s in it, and then keep it handy for application whenever the need presents itself.
VP Operations and Directors and Industrial Engineers will find the most bang for the buck here because it addresses high-cost situations. However, the analysis applies to every-day operations too, so it may make good training material for Ops Managers and Supervisors to read. The practical applications on a daily basis (for example, to quality errors) may need more explication and examples in the actual warehouse environment where data may not be as readily available.
And project managers, like here at PL Programs, should read it too. It is a good guide to help ensure that requirements for big-consequence decisions are well-understood.