White Paper

Master Yield Management - 10 Steps To A Competitive Advantage

JustFood

Yield analysis and management is a complicated topic that most wouldn’t consider a potential source of competitive advantage. But there is some compelling research that suggests that a robust, evidence-based approach to analyzing and optimizing yield can have an enormous impact on a company’s bottom line.

This white paper covers:

  • Why the most common approach to improving yield doesn’t lead to excellence
  • How to determine the best possible yield and calculate a value known as the ‘yield index’
  • Why the yield index should be at the center of your yield optimization efforts
  • A 10-step process for applying these principles, and profiting from them, in your own company

There is some discussion of regression analysis and modeling. Some familiarity with those concepts is helpful but not required.

Introduction

Every company performs yield analysis to some degree, examining where unnecessary waste happens between raw materials and finished goods and making efforts to reduce that waste. Some companies have a more formal approach than others in seeking out these incremental improvements, but most act on the assumption that if improvements occur on a regular basis, management is doing a good job.

The logical flaw in that assumption, and the danger of using your own past performance as your only benchmark, is that it’s possible to improve frequently and still not be very good.

While it’s true that any yield improvement is desirable, management needs more information to know if they’re doing a good job. Specifically, they need to know what the best possible yield is and how close they are to it. Then they’ll have an objective standard to compare themselves to, rather than comparing themselves to their own past performance which may or may not be a good standard.

The problem of defining the best possible yield

With so many steps and variables involved in transforming raw ingredients into finished products, it’s very hard to know the maximum possible yield. When you consider that some waste is necessary for the transformation process and is therefore wanted, and that some processing steps (e.g. rinsing) can add rather than subtract weight, it’s clear that simply comparing the weight of the raw material to the weight of the finished goods will not give you a good idea of your yield efficiency and where your best opportunities for improvement are. You need a more sophisticated approach.

Introducing the yield index

The solution to the problem of quantifying maximum yield and getting the associated benefits is solved in the research of Derk Somsen and Anthony Capelle. Their research proves that, when food companies know the theoretical maximum possible yield, they can improve profits significantly by taking more systematic and effective steps to reduce unwanted waste.

Their approach uses a new concept they call the yield index. The yield index is a ratio.

YIELD INDEX = ACTUAL YIELD / HIGHEST POSSIBLE YIELD

A yield index of 1.00 means the actual yield and the highest possible yield are the same; there is no unwanted waste. A yield index of 0.92 would indicate 92% raw material efficiency — a little room for improvement. A yield index of 0.50 would tell you half of your lost mass is unnecessary and preventable.

A 10-step process

Somsen and Capelle developed a 10-step process for determining a company’s yield index and building a program of continual improvement around it. It works because it’s rooted in the scientific method and objective standards of performance rather than just historical ones.

  1. Set a baseline by which you can track future improvement by measuring your production yield. Use whatever method is convenient and familiar for you.
  2. Characterize the raw material. Describe every parameter of the raw material that can possibly affect production yield. In Somsen and Capelle’s work, they examine potatoes that are to become French fries, and identify parameters such as the potatoes’ bumpiness, thickness of skin, number of eyes, average size and standard deviation of sizes, number of potatoes per kilogram, the potatoes’ age and their orientation during cutting. You’ll similarly have to consider every possible variable in the transformation of your own raw materials that could affect yield.
  3. Characterize the optimum process by using the scientific method.
    1. Develop hypotheses on how you could achieve the most efficient transformation possible. Don’t be limited by existing constraints, procedures or policies, but be practical and examine only possibilities that could actually be implemented.
    2. Conduct experiments to test the hypotheses.
    3. Observe the results.
    4. Draw conclusions and refine your understanding of where unwanted losses occur with further experiments.
  4. Implement the optimum transformation process you discovered in the previous step and take careful measurements. The measurements will be the data points used in further analysis and modeling. Don’t expect to witness significant improvements at this stage; that will come later.
  5. Develop a model based on the data you gathered in the prior step. This is where some special knowledge and skills come in, namely regression analysis and modeling.

    Regression analysis helps you find the formula that best describes your data points, which means you can use historical information to make predictions about the future. It will, for example, tell you precisely what the relationship is between the number of eyes on your potatoes and production yield. With that information, you can build a spreadsheet model that uses your raw material parameters as inputs and gives you the best possible yield as its output, making it possible to calculate your yield index. Ideally, those parameters will be recorded when the raw materials are received. For help with this, Somsen and Capelle suggest Applied Linear Statistical Modelsby Neter, Kutner, Nachtsheim and Wasserman.

  6. Make the model convenient for practical use by turning it into a spreadsheet (or, better yet, embedding it into your business software) so workers on the production line can use it. Make sure it’s practical. For example, since the relative amount of surface area on a potato reduces its potential yield into French fries, the average volume of your latest shipment of potatoes would be a great predictor of the maximum yield potential, but measuring the volume of each potato isn’t practical on a busy production line. The number of potatoes per pound is a close-enough proxy and much easier to measure.
  7. When you’re confident of the software or model you’ve developed, implement it. Train employees on the important concepts of yield analysis so they’ll be able to make the most of it. This initial implementation will be the benchmark when the whole company adopts this approach, so full buy-in and support from senior management is critical. Choose an interval that makes sense for your business – by shift, batch, day or week – to measure all the key raw material parameters you identified earlier and input them into the model to determine the best possible yield. Then measure the actual yield and compute the yield index.

    Take special note: yields greater than 1 are possible in practice and often indicate inefficient dissection when dealing with protein. Look to see if the yield index of less valuable parts greatly exceeds the yield index for more valuable parts. If it does, dissection is probably not cutoff efficient – too much yield is being attached to the less valuable parts. This is bad for profitability, so treat very high yield indexes with suspicion.

  8. Carefully examine the results of your implementation to gain a better understanding of where losses are occurring and which losses are unwanted versus necessary for the transformation process. Apply the 80/20 rule and focus your efforts on the small number of operations that are generating the greatest proportion of unwanted losses. Measure and celebrate success often to maintain enthusiasm for the program, and make continual improvements, always reaching for a yield index of 1 (or close enough to it that raising yield any further would consume more value than it would generate).
  9. Benchmark performance so that employees and departments across the organization can learn from the best. There should be a formal process by which the knowledge and skills of leading workgroups can be shared so that all parts of the organization can strive towards that standard.
  10. Build more advanced models that can help you, for example, predict next year’s yield based on pre-harvest information from farmers, or develop simulations to gain a better understanding of where unwanted losses are coming from in areas where you struggle to improve the yield index.

Conclusion

This is an involved process that can’t be entered into lightly. Somsen and Capelle estimate that following their process in a typical production facility will take 1.5 full-time employees about 1.5 years to complete, so the investment in time and money is significant. But their work proves the value of the effort. The French fry production facility used as the test subject in their research saw their yield index improve from 0.77 to 0.90 over a ten year period, with the entire improvement going directly to the bottom line.

With the right commitment, skills and investment, you can approach yield analysis and optimization in a way your competitors likely aren’t, improving your margins and profitability and turning your new mastery of production yield into a sustainable competitive advantage. There are few parts of your business where so much stands to be gained with so little risk.