Seven Questions Before Selecting a Post-Harvest Preservation System

Last January, a packing house manager came to us with a specific concern: "We have spoken with three vendors this season. They all say their coating extends shelf life. How do we know which one to believe?" It is the right place to begin, yet most vendor reviews leave it out.

Use this article as a question list for any post-harvest preservation vendor, including Ecogear. We wrote it because evaluations often emphasize headline promises, such as longer shelf life or fewer rejected loads, while overlooking the documents and operating details that decide whether the system works at your site. We ask ourselves the same questions.

Question 1: Which Food Safety Certifications Cover the Coating?

Request exact regulatory clearances rather than accepting a broad "food-safe" statement. A product can be food-safe under one regulatory framework and not another. Those claims are not interchangeable, and a vendor who merges them is either uninformed or imprecise.

For Korean operations and East Asian export, request Korean Food Drug Administration (KFDA) records for every formulation ingredient. Check the category for coating agents used on fresh produce. Also ask whether the ingredients are registered as food additives or general food contact materials, because the documentation rules differ.

For Japan export, remember that Japan's positive list for food additives permits only substances explicitly approved there. Approval in Korea does not guarantee inclusion on Japan's list. Request documentation showing compliance with the Japan Food Sanitation Act for the coating intended for your crop types.

For EU or US export, request E-number listings for wax components and GRAS status for protein-derived components. Certain formulations may also require an EU Novel Food assessment.

A vendor unable to provide documentation for each target market may leave your product vulnerable to detention at the export border. A general "food-safe" certification is not enough when produce crosses regulatory boundaries.

Question 2: What Data Supports the AI Routing Accuracy?

When a vendor uses AI for routing or prioritization, it should explain the model's prediction, the benchmark for a correct result, and the validation data behind it.

Do not ask only, "how accurate is your AI?" Ask instead: "Can you show me a precision-recall summary or confusion matrix from validation, broken down by crop type?" Those are the key measures for batch triage. Precision shows how many batches marked high-risk truly needed urgent treatment. Recall shows how many batches that could have benefited from treatment were actually flagged. Optimizing precision alone can produce a high accuracy claim while overlooking many at-risk batches.

We use the same test ourselves. Prospective partners can review a summary of our pilot-season validation data. We do not quote one precision figure here because results change with crop type and sensor setup, making one aggregate figure misleading. We can provide the method and the breakdown by crop type.

A single accuracy figure is not useful for judging performance at your facility unless the vendor identifies the crop type, sensor setup, and season behind it.

Question 3: What Is the Integration Schedule and Who Handles It?

A credible answer identifies the person responsible for every integration task. A "four-week integration" that needs your IT coordinator full-time for three weeks is not the same as one where vendor staff configure the system and your team spends a day training and signing off.

Ask directly: who sets up the connector between your sorting line and the vendor platform? Who translates your line's field names into the required schema? Who trains sorting and receiving staff? After go-live, who gets the first call when something fails?

Ask about prerequisites too. Some sorting line manufacturers must enable data export before a third-party connection can begin. That requirement may add one to three weeks and remains outside the vendor's control. If it is buried inside the quoted timeline, the vendor is understating time to live.

Question 4: How Is Treated Cost Calculated Per Kilogram?

For packing house planning, monthly subscription price is not the most useful measure. Cost per kilogram treated ties directly to the operation.

Work out the full cost per kilogram: add coating cost to the monthly subscription, then divide by the usual treated volume each month. Compare that result with the average cost per kilogram of dock-rejected produce, including both product value and the handling cost of a returned shipment.

For most operations, the preservation system should make economic sense. If it does not, have the vendor run the calculation with your actual throughput. A positive unit-economics case based on your own volume should be clear before you sign. If it is not, resolve whether pricing or expectations are the problem.

Ask whether price changes with volume. A flat monthly subscription covering a defined throughput may cost more per kilogram at low volume than a per-kilogram fee, while the reverse may apply at high volume. Request the complete pricing model in writing before finishing the evaluation.

Question 5: Is a Pilot Available Before Commitment?

A vendor that stands behind its system should propose a defined pilot, with a fixed scope and success measure agreed in advance. Use a metric your facility can measure, such as dock rejection rate, average shelf-life days per treated batch, or both.

Pilot points to negotiate include the minimum batch volume needed for meaningful results on your crop type, since one or two batches cannot support a conclusion, the pilot duration, at least one full harvest cycle for your primary crop, and the success threshold. Agree what result means proceed and what result means the system is not working for your operation.

Be wary of pilots offered only within commercial contract terms, where meeting the threshold obligates you to buy. That is a purchase with a cancellation window, not a true proof-of-concept trial. We offer a free sample kit program so prospective partners can run the system on their own produce before making a financial commitment.

Question 6: What Happens When the System Flags a Mid-Shift Error?

This tests operating resilience rather than technology. During a busy peak-harvest morning, the AI dashboard stops generating urgency scores while the coating crew waits at the line. What is the fallback?

The answer needs both a technical and a human fallback. Does the system default to "treat all," "treat none," or "hold and wait," and which state suits your operation? Is immediate phone support available, and what response time applies during packing hours?

"We will get back to you within 24 hours" does not meet the needs of a production-critical system. It may work during off-season setup, but not during a July peach run. Ask for support hours and response targets during your primary harvest window, then test those commitments in the pilot.

Question 7: Who Will Handle Ongoing Support?

This is an accountability question, not a technology question. Request the name and direct contact details of the person responsible for operational problems. Not a ticketing system, a named person.

If your packing season peaks between June and September, "business hours KST" may not cover a 4 AM shift start. Confirm that your named contact is available during actual operating hours. If after-hours issues go to another team or a ticket queue, ask for the guaranteed response time for a mid-shift production problem.

At Ecogear, we target under two hours for dashboard or routing problems during Korean daytime hours in commercial packing seasons, and under four hours outside those hours. These targets are written into our pilot agreements. That is the standard we set for ourselves, and every preservation vendor should state its support commitment just as clearly.