When We Benchmark, Do We Really Understand the Numbers?


By David Nolan, DVM,
Senior Operations Manager for Cargill Pork

The term benchmark was originally used by surveyors to refer to points with known elevation and geographic position. All land survey measurements were required to be referenced to the known benchmarks to be legally binding; after all, what good is a set of measures and vectors if you don’t have a known starting point. The benchmark seems rather ancient when we consider the technology we have at our fingertips today with global positioning systems but alas, even the GPS systems use benchmarks on earth to function properly. In many cases our farm production measures are much like a survey with no benchmark or known starting point.

As we benchmark our production we must strive to fully implement the premise of the benchmark by understanding the starting point for the numbers. A benchmark without the finer details is an exercise in futility. I will review a few of the benchmarking fallacies I have encountered over the years and encourage you and the industry to continue recent progress in establishing true benchmarks through a deeper understanding of how the numbers were derived.

It would seem that there is no number much more important to the weaned pig producer than born-live. A pig cannot be weaned if it is not both born and living. Never does a visit to a sow farm go by without some mention of born-live or genetics sales presentation, without the latest and greatest breed combination touting a stellar born-live number. It always amazes me how every breed beats every other competitor. How is it possible to have every genetic line ranked number one?

Before I review an actual case study I need to ask how much training is done with the people actually recording the numbers in the barn? Even if you answered any number above zero, have you verified that it is occurring in the manner you trained? We assume the numbers that show up on the computer screen or printed report are accurate because they are calculated by the computer using mathematical equations and highly sophisticated programming, but we all have heard the adage: garbage in - garbage out, and it is true! At a bare minimum staff training is required, and indeed is very important -- managers must have an understanding of the people doing the daily recording of information. We place a tremendous amount of trust in our valued employees, as we should, but at times the employee’s daily tasks get clouded by other priorities and signals we are sending.

The case of the missing born-live: A pod of farms was struggling with born-live numbers. Further analysis showed that not only was the born-live low, but total born was also suffering. After in-depth analysis of the PigCHAMP reports, and comparing the pod of farms against another series of farms with similar genetics in the same production system, production managers jumped to the conclusion that the semen was to blame. The proverbial rabbit was let out of the cage and the chase was on. Since the boar studs for the two farms were different this was the obvious smoking gun. The boar stud was replaced with the same stud that produced semen for the better producing farms. The only problem is that after 6 months on the same stud as the better farms, the born-live still didn’t improve. After another round of
“in-depth” analysis, it was determined that semen was still to blame but this time it was because the method of delivery was different. The switch was made from over-night delivery to air transport with a dedicated courier. As you may have already guessed, 6 months passed and born-live didn’t improve. Was it time for another trip to the farms, only this time with a new focus.

The farms with the low born-live were paid for total pigs shipped; however, they had an employee incentive program for the farrowing house workers that was tied to pre-weaning mortality. Apparently, it wouldn’t be “fair” to have an incentive program based on total pigs shipped when farrowing employees had no direct impact on quality of matings or farrowing rate.

So did the workers in farrowing departments lie about born-live? Although there was incentive to cloud the numbers born, and this may have been done in some cases, for many of the employees it would be more accurate to say they were “cautious” about how they reported the results, so that they did no harm to their team’s performance…or their paycheck. The team had been trained that if a pig is dead and is immediately behind the sow then it must have been a stillborn so it was recorded as such.

But what if a pig is found under the sow on day one? If that pig gets recorded as born-live and laid on by the sow, it will impact the pre-weaning mortality….and it really is not a stillborn because it was not directly behind the sow as the employees were trained. The farrowing house employee sees this as an unfortunate event that they had no control over, since the sow farrowed at night and they weren’t at work so they had no chance to intervene and potentially save this pig. It might just be better not to record that information this time. Or, what about the pig that is alive but it might only weigh 1 pound? We know the chance of survival is slim; and the fact that the pig was born this small is not their fault either, it must be genetics or nutrition. So again, just this time, the pig is not recorded as born-live.

In the case of the missing born-live, you understand the whole picture when you see that the weaned numbers between the two pods of farms were identical. This information had been reviewed in the beginning, but the production managers for the farms with the lower born-live, choose to ignore that piece of information because it didn’t fit their preconceived notion that they were better than the other portion of the system. In the end, a lot of energy and money was spent chasing rabbits that were fabricated by benchmarking against the wrong mark.

There are multiple data points on the sow farm that are subject to misleading information. Although record keeping systems such as PigCHAMP continue to improve data validation, and analysts get more experience finding those hidden inconsistencies, new innovative methods are developed to by-pass those safe guards. When a better mouse trap is built, the mice get smarter.

If the farm focus is on litters per sow per year (LSY) we know a gilt is not classified as a sow until she is mated. What if the gilt is not recorded as bred until the farm crew has confirmed her pregnant, or even worse, the matings are recorded at the time of farrowing ‘just in case something doesn’t work out as planned in the pregnancy’. Surely a farm wouldn’t be this dishonest, but we often see individuals that will massage the information because the drivers are out of their control and they don’t want to be penalized for what someone else is doing. Before we move from LSY, let’s look at the other end of the measure: culling. Should I record the sow as a cull the day I make the decision to cull her or the day the cull truck shows up to remove her from the farm. “It isn’t my fault that the cull truck has already run this week, if management had their act together they would send a cull truck more often”.

Nothing is more clear than pigs weaned right? Either the pig is there or not. Do you reconcile pigs weaned on the sow farm with good pigs that arrive in the nursery? If a weaned pig producer is loading the truck and there is a discrepancy on pig count I am sure the error will tend towards the higher number rather than the lesser, and when the nursery producer receives those same pigs the count will tend towards a more “conservative” number of head received. And then there is the issue of good pigs vs. total pigs and does this number include DOAs which aren’t anyone’s “fault” except transportation. As a general rule the sow farm will ship more pigs than the nursery producer wishes to receive. Are these marginal pigs captured in the benchmark number and used with pride as the sow team touts an amazing PSY number? So can we really trust that 30 PSY number and does it send us on another rabbit hunt that more closely resembles Snipe hunting you might have done on a high school camping trip?

I trust you are beginning to see the problems created by sub-optimization -- or the approach that each department drives to optimize their own area of influence. We often see unintended consequences when one area of the business strives for perfection while ignoring or not understanding the bigger picture.

Although I used the farrowing house to demonstrate this phenomenon there are many more examples. The formulator aims for the lowest cost per ton, while the merchant tries to minimize the incoming cost of those ingredients, often by-passing concern for variation or nutrient content of those ingredients because that information either is not measured or much more difficult to obtain at the speed of commerce. The finishing team measures mortality and feed conversion, but if there is a feed issue it is the mill or nutritionists that must have done something wrong, not the farm, and the same team will lobby to exclude certain closeouts from reports because there was a ventilation failure resulting in extreme mortality that was out of their control.

Let’s look at a simple calculation of gain. Is gain calculated by one of these equations: 1) total pounds live weight shipped – pounds of live weight received, or 2) (average weight shipped – average weight received) X head shipped. These two calculations give completely different answers and both could be correct depending on your system and ability to collect numbers. The second calculation hides the impact of mortality, while the first is said to adjust for its impact. Without the whole story and an understanding of the results, they could lead you on one of those rabbit chases that adds no value.

Are you benchmarking against one of these “adjusted” benchmarks or one with known coordinates? Do you have access to the whole story? The only way to know is to dig deep into the numbers and make sure the service you use, allows you to do the digging and has the standards established up front to minimize the unintended consequences. And don’t hesitate to ask how the number you are focused on could be massaged to reveal a sub-optimized result so you can use the numbers to optimize your entire operation not just a department.

Dr. David Nolan, DVM, is Senior Operations Manager for Cargill Pork. Since 1987, Cargill Pork has led the industry in the production of high-quality pork products. As one of North America’s largest pork processors, Cargill Pork harvests more than 10 million hogs and produces nearly 2.3 billion pounds of pork each year.