Getting A Better Grasp On Traceability
Jim Prevor’s Perishable Pundit, July 19, 2007
Few issues are more important to the future of the produce industry than the development of an effective traceability mechanism. And no one should be under the delusion that we have one.
The FDA grew so frustrated at the inability of one of the national companies involved in one of the food safety outbreaks last fall to deliver traceback information that they were on the verge of invoking the Bioterrorism Act, which requires a traceback within 24 hours. Failure to comply would have turned the vendor’s incompetence into a criminal act.
The FDA held back only because it was convinced that the vendor actually had the information but sincerely had no idea where it was — so they cut them a break. The vendor finally came up with the information — two weeks after the event.
Many vendors think they have effective traceability but it turns out that if you call them on Sunday morning, they have absolutely no ability to give you any information. The real goal is time and personnel-neutral traceability, in which answers are as easily available on Christmas day when your vice president of food safety is on vacation climbing Mt. Everest as when he is sitting at his desk.
Early on we turned to Gary Fleming, Vice President of Industry Technology and Standards of the Produce Marketing Association, for insight into this important area. First, in Guest Pundit — Traceability And the Need For A Common Language, Gary explained the crucial role that data standards play in any industry-wide effort to establish trace-back capability. Then, in Guest Pundit — Pairing The Global Language With Technology, he helped us understand how we could actually make this happen. In a third piece, Guest Pundit: Traceability — A Forgotten Piece of Food Safety, Gary traveled to Argentina in search of a specific and effective traceability solution.
When Bruce Peterson resigned from Wal-Mart, people were waiting with baited breath to find out what his next step would be. So when the press release arrived advising that Bruce had entered into a collaboration with Michael McCartney, Principal of QLM Consulting, to promote a traceback effort for the produce industry, we wanted to find out more.
Bruce Peterson Focuses On Traceability detailed his basic point: that the produce industry is more likely to reduce the negative impact of food safety problems by enhancing traceability than through any other single measure.
Now, to find out how we might make this happen, we asked Pundit Investigator and Special Projects Editor, Mira Slott, to speak with Michael McCartney so we could better understand the collaboration between Bruce and Michael and so we could fill in the details about the challenges ahead:
Q: Could you clarify your collaboration with Bruce Peterson? In your pursuit to develop global traceability guidelines, are you looking to provide consulting work with individual companies or is this a public service to try to make traceability more high profile?
A: We would be doing this as a public service, and certainly we wouldn’t be doing this for anything beyond what our costs would be. Both Bruce and I are doing fine with income and not looking to enrich ourselves financially through this venture. We have not been paid by anyone to do this quite frankly. We can offer our services on a cost basis and we’re committed to make this work as opposed to another approach of needing this for bottom line revenue.
Costs are a reflection of where we are getting our support. We have strong support from between 10 to 20 companies that have said they are willing to contribute in some way to have us put together meetings to iron out types of methodologies to create a traceback recall system.
We are looking to facilitate this process and do it at whatever it costs, but at the lowest cost possible. Costs depend on a number of factors: the level of participation, how easy is it going to be to reach common understanding or approach on what is important. Participants may not necessarily be contributing funds directly. For example, a retailer may provide in-kind services, give us part time access to three different folks that could share their expertise and research, etc. The grower might say, here are resources and access to all the information you need.
Q: Are you speaking hypothetically or do you have an actual meeting scheduled or larger summit in the works?
A: We want to set a meeting no later than the end of August, either in Chicago or Dallas and it will be an all day meeting. We need grower/shipper participants and major players in the foodservice and retail industries. Many of these people are already on board, but we need to galvanize everyone in one place, lay out the parameters, exert the leadership.
Readers should know if you are interested and want to participate, come join us. With Bruce and me running this, people can be assured we are going to take the straightest line between two points. Up until now, there have been a lot of divergent activities. We need to converge these activities into single actions. We need industry organization support at a high level. We’re not looking at dollars but for leaders to reach out to their memberships.
Our problem has been we that we have a strong tendency to collaborate and not to dictate, and we’ve been acting in that manner for the last several months, but now we have to take all this collaboration into one room and craft a plan and take action. We’re not saying, ‘do it this way’. We want to examine the methodologies, and who is doing this well. We don’t need vendors and software data guys there. Collectively we must think through problems and come up with common solutions. At that point we pick the best systems, test and compare. I could list five solutions now, but what I want is for people to bring in their solutions and then we can pick the best approaches together. We think the Leafy Greens Marketing Agreement took about a year to get going, and we’re looking at that timeframe.
Q: After this initial meeting, will you be orchestrating a larger summit?
A: Assume for a moment we have 20 companies on board. We’ll call a summit, meet for dinner before hand and spend the next day participating in small group workshops culminating in a set of principles to outline methodology; it’s going to cost x amount of dollars, etc. We’d look to do it at the lowest cost, say someone offers their big executive office in Dallas, Texas, to gather these folks and everyone flies in at his or her own expense. Now we create different working groups that have specific tasks, and create a common data dictionary.
Nothing like this exists today. For example, a retailer might define ‘receiving’ as when the product comes into its distribution center and is offloaded to the floor of the warehouse, another retailer says the product is only received once a quality inspection is completed, and a third retailer doesn’t say product is received until it is posted to inventory.
Further, PMA and the industry have focused on creating a global product identification code, but what we’re saying is before you get there, you have to identify the lot of the parcel of where it’s created, everything from the seed through the supply chain before the product is ready to ship. Up until now, all this information has been largely invisible. We want to make that information more accessible to stakeholders with common definitions and in a structured universal data format.
If the perishables industry was so structured, that would be great, but we have a myriad of systems and many ways of annotating this information now. We don’t have a common approach up the supply chain that a retailer or foodservice company could turn to if they received a food safety inquiry from the public.
Q: Aren’t there systems on the market that already do a great job of traceability as long as everyone in the supply chain is in the loop? How do you solve problems that have stopped traceability from being fully implemented?
A: There are actually a number of different approaches, a lot of methodologies, and dozens and dozens of solutions, so I don’t want to leave the reader with the idea that there is just one. That’s part of the problem; being able to ferret or sort out all those solutions in an open, constructive way to move the process along.
Q: Isn’t this quite complex since you’re dealing with the entire, fragmented supply chain?
A: Essentially there are three major stakeholders: retailers, foodservice operators and suppliers. If you draw a line starting on the left to the beginning of the food chain, you want to know what is in the soil, what chemicals and fertilizers are used, more importantly the seed, what its origin, is it GMO, produced locally, from a major manufacturer, or different seeds over time. Farmers already have all this information, maybe on index cards, written down on a field notebook or journal going back 10 to 20 years.
A key question would be what is the source of the raw material and, once the seed is planted, all things that go into the soil, the weather conditions, water systems, whether the land is welled, is there a natural abundance of rain in the area? Then there’s the testing of that water, sourcing of that water as well as other materials, usually information that is kept pretty privately.
Essentially it involves painting a picture, creating a timeline from the birth of the product. You want to establish what happens from the cradle and follow it all the way to the fridge, or at least to the shelf and point of sale. Along the way, you need to document data points of events that occur. This is about the story — really multiple stories — this product needs to tell.
The first story is the seed, the second is the soil, third pre-harvest and the inspection done within that time all documented on paper. Then the harvest itself involves a series of steps that you need to identify; what parcel, block of land, the lot where it’s harvested accomplished with latitude and longitude global positioning systems, the start time, who actually performed the work, whether it was a hand or tow process, the crew name, crew dates, and employee ID number. You also need the amount of products harvested that day, and if they go into separate containers, those may hold 100,000 products each. Is the product field harvested, graded and inspected to put right into the container to be shipped? What happens to that product during the day? Does the crew move from one field to the next, from one parcel to another?
Northeast or southwest of that parcel a neighbor might have horses or there may be a dairy around. You’re in farmland. You need to be able to identify who your neighbors are as they need to identify who you are. When harvested to another location, when and by whom? What ever container or pale being used needs to be identified. If a tractor or trailer equipment is involved, you need to identify it.
Q: How extensively and thoroughly is this kind of information gathered and documented now?
A: It is being done now; the questionable part is the level of granularity. Certainly the crew and the barn are being identified. But the details vary from farm to farm, practice to practice, and commodity to commodity. For example, the crew may be documented, but with no separate ID’s for each employee and not having that ID directly tied to that particular box. That crew might be at the same farm, which has eight parcels.
In this system, you have variability. The whole harvest area has a series of changes that are critical to traceabilty. You’re a farmer, I’m a farmer, and there are 20 others picking lettuce in bins coming in from these 20 farms that all end up at the same food processor.
You can’t do traceability if you don’t have a unique product identity already established.
Once product arrives, how do you separate each lot or bin or farm? The only way to do this is to create a time gap. If three farmers supplying the same product each have a unique identity, all bets are off once the food processor dumps the product in bins and it’s all blended together. It can happen like this now. If there isn’t a gap between your produce and mine, we lose our identity. We lose our ability to trace back. If an inquiry comes in, we have to look at a three-farm recall instead of a single farm, and what comes before and what comes after.
Assuming we’ve maintained separate identities, and we know it takes a certain number of minutes to process what’s in each bin, we have narrowed in on our ability to track the problem. This is taking a simple example without multiple grades. With oranges for example, you may have five different grades, your farm grades two, three and four, and these have to be sorted out.
Now it gets stored temporarily, either a barcode or RFID label is affixed to the carton, then if it’s in a pallet, it receives another barcode or RFID association to read and match up to a purchase order. Those all go together and now you have a tractor or trailer with all that information, refrigeration number identification, etc. There’s a whole series of transportation things that have to be taken care of. On a macro level, the truck has a number; it may go direct if one load to a distribution center, if not it may pick up other product. It gets logged in, and then it’s inspected. Transportation is another significant event where about 30 events occur that I’ve glossed over.
Now Wal-Mart or Target or Kroger accepts it, assuming everything is ok, and it goes to temporary storage for the next call to action from a store that needs x amount of these cases, say five per store… these cases are broken down into 10 separate pallets with other things to go to the store.
Now it leaves the distribution center, and there could be intervening steps along the way, where the same process is replicated again and again. It could go to a wholesaler and another wholesaler. It could be one ingredient of a bag of salad, and another supplier could provide carrots.
The product goes to the store, and hopefully the produce merchandiser understands what the best use-by date is and the employee rotates the product. As individual items are sold, the retailer has to keep track of the pieces that go into rotation. You may have three farms supplying tomatoes, and normally the retailer will keep track of that because they want to know which tomatoes are selling better.
Then it goes to point of sale, and there’s the problem of non-branded commodities becoming aggregated. A retailer puts Roma tomatoes from three farms together, and doesn’t differentiate which farm. It could be related to supply/demand — a product is selling well in one area of the state, so the retailer may want to divert product from one area or another. Maybe the retailer is doing a promotion with wine, grapes and cheese, always being challenged in managing inventory.
Q: What are some strategies to better document the product’s movement through this complicated supply chain system to improve traceability?
A: One methodology is to set up data bases as a way to input information about the product at each significant point along the supply chain; seed, soil, harvest production, transport, distribution center, store, and all stakeholders are involved in each of these events, and share the private data bases.
Your soil information goes to a separate data base. You may be a farmer, also doing the harvest, so you’d be involved in two data bases. But you might be the food processor with another private data base and only you define who looks at that data. You line up all these notes. You may have eight or 10 different data bases, and depending on the operation, share one if not more.
If someone gets sick and notifies the food service operator or retailer, you may give your food service operator or retailer permission to go through your data base. Now you have a lot of data on a particular product by looking at a unique bar code or RFID number. If you have one individual that has a health concern, that’s one thing. If you have 10 people saying the same thing from 10 places all connected to the same date, time harvest and crew, you know you have a problem. They may or may not have called their local health department, but you have a means to detect and act on the problem.
Q: This sounds like PulseNet. Are you suggesting that individual companies take on this function?
A: You could be the grower, a retailer, or a food service person. You could have the ability to see complaints from 10 restaurants, and all this correlation from one grouping and take immediate action, rather than wait weeks and weeks. You could voluntarily withdraw product from the market and start testing.
Q: The example you give seems quite simplistic and straight forward. Wouldn’t most investigations involve a significantly more complicated and time-consuming process? Will most companies have the expertise or inclination to do the necessary legwork?
A: The food service industry is on the front line. In the grocery store there are many other intervening events. It could be related to the cleanliness with food preparation, meat on cutting board contaminating product. In each event category, there are places where food safety could be compromised.
These data bases could provide the ability for the industry to investigate on its own, define the problem earlier, take more immediate action and do preemptive work. Right now there is not all the information at hand to do that.
Q: What’s your plan to develop these data bases? And what parameters will be put in place for use. Will the FDA and CDC have access to this information?
A: We want to first get the three parties together — retailers, food service operators and suppliers, to find out what data they have already, what format this can be put into, and discuss the rights with sharing partners. We need to do the foundational steps to create this private data base network. It’s only for the industry. We could see PMA and varying others overseeing this data base so the industry becomes the custodians rather than the government.
We need to take a horizontal view of things and get away from a silo view. We need a universal system.
Q: How does your vision relate to other efforts underway such as PMA’s work to develop an international product coding system?
A: We want to incorporate produce industry organization efforts. For example PMA is actively involved in GS1 technology, to get farms and companies to create a product catalogue for all the items they produce. Right now we have a generic catalogue through PLUs. RSS gives the ability to put much more data in a much reduced space. That data can better pinpoint the product’s origins, and is readable throughout the supply chain and certainly very viable, as is RFID. All these things are pieces to the puzzle, a way to document more information. The companies control what information is shared and what’s not in the greater interest. And all this inspires customer confidence.
Bruce [Peterson] and I are trying to drive discussion. There are so many stakeholders and complexities, and up until now all these have been silos. We’re trying to facilitate common dialogue and help create common methodology and approach.
A good analogy is the system Wall Street went through to computerization as a seamless way to trade commodities. The same system that runs through Wall Street is ubiquitous.
Another example is what happened in the airline industry. American Airlines had a proprietary reservation system. United and everyone else had their own systems but knew they were inferior. They approached American, and American enabled the other carriers to utilize its system. American was the big enabler.
A third example is the approach high tech companies used to avoid federal regulation It must be 15 years ago they got together, Microsoft, Intel and IBM, and all agreed to do business the same way. They set up guidelines to become self regulated. These industries have come together and agreed on the way they will do business.
Q: The produce industry is so fragmented though, with challenges unique to perishables.
A: Traceability solutions only work if you have a completely aligned supply chain.
The problem is that’s not the way the whole produce industry works. Not everything is predetermined to end up at Wal-Mart.
We have people in competition, and people that buy from each other. There are all kinds of challenges in the way the industry procures product. We can improve traceability and food safety as an industry if we all agree these are the things we need to keep track of and this is the way to do it. We need an industry-wide solution.
We appreciate Michael’s time and were impressed with Bruce’s resolve to tackle this daunting issue of traceability. Yet we find ourselves baffled by what, exactly, they want to do.
Bruce has been saying that food safety is getting a lot of attention, but effective traceability is not. This is exactly what we have been saying at the Pundit for months, and it is exactly what Gary Fleming of PMA said in three articles for us. All the trade associations and most major companies are in strong agreement.
Beyond this point, however, things get confusing. Here is the best we can understand the situation:
1. Michael wants to create standards for information needed at the farm level.
Although there are no standards stating what detailed pieces of information must be captured at the farm-level (beyond the GTIN capturing the item and the lot number capturing the field and lot from which the item came from), is this really necessary? The information would be hard to standardize as it could vary greatly by commodity, harvest method, etc.
This issue seems to raise a question as to what is the purpose of traceability? The information Michael wants to standardize is more for finding the cause of the problem, rather than isolating the location of the problem.
In other words, the trade’s goal has been more limited. Traceback brings you to the location of the problem — from which other records allow thorough investigation. Michael seem to be saying he wants an FDA officer to be able to not merely see that the product came from the Jones Farm, field 7, lot 6 — but to be able to note the seed source as well.
This seems very hard to standardize as it is such a broad group of issues — seeds, rootstock, etc.
2. Michael wants to create a standard communication protocol with which to communicate this traceability information between trading partners, even potentially with government agencies.
This is where the ideas become very vague. Why would we need a standard communication protocol (e.g. XML) if we were not feeding information into a central repository?
Michael doesn’t seem to want to commit here. A “central repository” of information? This is definitely the “Cadillac” version of retrieving, sending and storing information and one that a lot of small to medium companies would have trouble implementing. Some have mentioned the use of 1SYNC (formerly UCCnet) as the example of what we would be attempting to build. Well, 1SYNC is a very expensive model to build and a very expensive and complex model to participate in. Even the larger produce companies avoid 1SYNC like the plague. Is there a practical way to do this?
3. Determine “when” and with “whom” to share the data.
Giving access rights as to who can see your data is appropriate and an important issue IF you have a central repository of information OR if you have an Intranet-type database accessible by your trading partners. The question is: “Why would a grower/shipper want to make that type of information available to their buyers?” All a buyer would want to know is what do I need to pull off my shelves and out of my inventory AND what other stores did you ship this product to? It is up to the grower to quickly determine where the source is and where did I ship that product. This does not require a central repository.
Overall, everyone agrees with Bruce’s premise that traceability needs more attention. Everyone also agrees that you have to start somewhere or nothing will ever get done. As Bruce is wont to say, you have to create the “will” to do this. This was done in the Leafy Greens Marketing Order by a band of companies getting together with the help of key trade associations to make it happen. It seems that Bruce and Michael envision the same type of activity will be needed for traceability. As such, they hope to rally several companies to support this and then get the trade associations to help facilitate a “solution” — though the precise solution they want support on is still unclear.
It is not clear if this is going to happen. We had something tangible to address with food safety and had some concrete solutions. With traceability, and with everyone claiming they already have a traceability solution, we do not have something tangible to address. PMA has already given solutions via the study it did with CPMA in their report, “Fresh Produce Traceability: A Guide To Implementation”.
The truth is that the reaction to Gary Fleming’s articles was less than we had hoped. It seems that when PMA says “create an effective traceability solution,” the grower community responds “we already have one”.
But we do not.
To some extent all these traceability solutions break down because the nature of the industry is such that the supply chain is destined to be fragmented. When Michael says, “Traceability solutions only work if you have a completely aligned supply chain,” he is not being encouraging about the produce industry finding a traceability solution.
Our take is that the fragmented nature of the industry means that only a central depository of information can work.
Why couldn’t the Center for Produce Safety maintain a confidential database, and the requirement is simply that anyone who buys produce in commercial quantities has to scan or enter a code whenever they buy or sell?
Most of this data transfer could be done electronically so it wouldn’t be too burdensome.
This won’t solve all food safety issues — if the traceability goes back to a processing plant that hasn’t stopped to clean its lines in years, you won’t get a certain demarcation between the product of different farmers. So there are still issues to resolve.
But an industry database would let an FDA official with nothing more than a contaminated bag in his hand trace back to a plant and trace forward of all product produced that day. He could press a button and have a printout of every company in America that has purchased product produced on a particular line on a particular day. You can see who sold product and who might have some left.
And the FDA official could do this on Christmas day, no matter who is on vacation. That would be an excellent start to a traceability system for the industry.
It is not everything Michael would like to see in there, but we don’t want to make the best the enemy of the good.
Many thanks to Michael and to Bruce for their strong efforts in this area.