Importer Guide

Can a Third-Party Inspector Guarantee Food Safety?

Last updated: 4 October 2026 · By the FoodGate Audit inspection team (ISO 17020 accredited) · Get a free quote in 24h →

No. A third-party inspection examines a sample of the lot, on one day, at one place, and an inspector who promises a guarantee is promising more than the method can deliver. On a typical container of 1,200 boxes, a standard ISO 2859-1 plan opens 80, and a lot in which 1 box in 100 carries a critical defect still passes that plan about four times out of five. What an inspection does deliver is a documented, measurable reduction of risk, and a report that proves exactly what was loaded, when, and in what condition.

Key Takeaways

These are the points a quality manager needs settled before telling management what an inspection report covers.

Point	Details
Buy a stated sampling plan, not a promise	Ask for the inspection level, the AQL per defect class and the sample size in writing, before the season.
Read a pass as a probability	A passed lot is one whose sample stayed under the acceptance number, not one proven free of defects.
Never read a pass as a residue clearance	499 of the 541 EU notifications on Egyptian produce concern hazards no inspector can see: residues, pathogens, toxins.
Decide who covers the voyage	A loading inspection stops at the seal; temperature and handling after that need their own evidence.

What the EU record says an inspection would have to catch
The arithmetic of 80 boxes out of 1,200
What the report proves, and where its proof stops
Why "guaranteed safe" is the wrong thing to buy
Where an inspection earns its place
Recommended
Frequently Asked Questions
Sources

What the EU record says an inspection would have to catch

The table sorts the 541 EU RASFF notifications on Egyptian fruit, vegetables, herbs and spices validated between 28 November 2019 and 22 September 2026 by the hazard category recorded on each notification, and adds what could have found that hazard before loading.

Hazard category recorded	Notifications	What can find it at origin
Pesticide residues	458	A laboratory analysis of a drawn sample, never the eye
Pathogenic micro-organisms	29	A microbiological analysis of a drawn sample
Natural toxins and mycotoxins	12	A laboratory analysis
Additives, allergens, pollutants, composition and other chemical hazards	20	A laboratory, or a check of labels and declarations
Foreign bodies	1	Visual inspection, on the boxes opened
No category recorded	23	Depends on the subject: these include four foreign body findings and two temperature failures on frozen consignments

Source: FoodGate Audit Egyptian produce dataset, built from the EU RASFF Window, counted by hazard category on 4 October 2026. Two notifications carry hazards in two categories, so the rows add up to 543.

The proportion is the whole argument of this article in one number. Taken together, residues, pathogens and toxins account for 499 of the 541 notifications, 92 percent, and none of them is visible to an inspector standing in a packhouse. On those hazards an inspection without a laboratory order proves nothing at all, and with one it proves something about the sample that reached the laboratory.

The arithmetic of 80 boxes out of 1,200

The AQL glossary entry on this site gives the reference case: a 40 foot container of mangoes, around 1,200 boxes, inspected at General Inspection Level II of ISO 2859-1, which sets a sample of 80 boxes, with critical, major and minor defects each judged against its own AQL. Under normal single sampling, the standard's acceptance number for that sample is 1 at AQL 0.65, 5 at AQL 2.5 and 7 at AQL 4.0. The other 1,120 boxes are not opened.

What that means for a buyer can be calculated exactly. The table gives the probability that a lot of 1,200 boxes passes, for a given share of defective boxes actually present in it.

Defective boxes in the lot	Critical defects, AQL 0.65 (accept 1)	Major defects, AQL 2.5 (accept 5)
1 percent (12 boxes)	81 percent	above 99 percent
2 percent (24 boxes)	52 percent	above 99 percent
5 percent (60 boxes)	8 percent	80 percent
8 percent (96 boxes)	below 1 percent	37 percent
10 percent (120 boxes)	below 1 percent	17 percent

Calculated by FoodGate Audit with the hypergeometric distribution, sample of 80 drawn at random from 1,200 boxes, acceptance numbers from ISO 2859-1 Table 2-A.

Read the first column slowly. A lot with a dozen boxes carrying a critical defect, the class reserved for food safety, allergens and foreign bodies, is accepted four times in five. That is not a failure of the inspector; it is what the plan was designed to do. AQL sampling protects against lots that are bad on average, and it is efficient at that: a lot with 10 percent major defects is refused more than four times in five. It was never designed to find the one box in a hundred.

Residue testing follows the same logic at a different scale. Under Commission Directive 2002/63/EC, the method behind official EU residue checks, maximum residue levels for plant products apply to a composite sample derived from one to ten primary samples, meant to represent the average residue level in the lot. A compliant laboratory result is therefore a statement about an average, which is also how the border will judge the lot, and silent about any single carton.

What the report proves, and where its proof stops

An inspection report is strong evidence on a short list of facts, and a quality manager should be able to name them. It proves the identity of what was inspected: the container and its seal, the lots by reference, the cartons by count and marks. It proves the date and place. It proves the measured condition of the sample on that day, with each defect counted against the specification, the pulp temperature and Brix where they were taken, and photographs that tie each finding to a pallet. A redacted sample inspection report shows what that looks like on an Egyptian lot.

Its proof stops at four edges. The first is the unopened boxes, whose condition the report infers from the sample but does not observe. The second is chemistry: without a laboratory order, the report says nothing about residues or pathogens, whatever its pass or fail line suggests to a reader in a hurry. The third is time: once the doors are sealed, temperature, ventilation and handling during the voyage belong to someone else's evidence. The fourth is the field itself, since an inspection sees the crop after harvest and cannot reconstruct what was sprayed on it, or when.

Pro Tip: before the season, send your inspection provider one question in writing: "On a lot of 1,200 boxes under your standard plan, what is the probability that a lot with 2 percent critical defects passes?" The answer, 52 percent under the plan described above, belongs in the inspection order next to the plan itself, where nobody can later remember it differently.

Why "guaranteed safe" is the wrong thing to buy

Buyers ask for a guarantee because their own customer asks them for one, and the request passes down the chain until it reaches someone willing to say yes. That someone is the problem. A provider that writes "guaranteed" on a report has either used the word loosely or claimed a certainty the sampling cannot produce, and in both cases the document will be read against its wording on the day a lot is refused.

What can be bought is a measured share of the risk, and the method that measured it. A stated plan can be tightened where the record justifies it: General Inspection Level III instead of II, a lower AQL for critical defects, a residue scope written from the substances the EU actually finds on that product. Each choice moves a number in the table above, and none of them makes the number zero. Accreditation works the same way. ISO/IEC 17020 accreditation assesses whether a body is competent, impartial and consistent in how it inspects; it does not certify that the goods are safe, and an inspection body is not a certification body. FoodGate Audit holds accreditation number 032506 from the Egyptian Accreditation Council for the inspection of fruit and vegetables, and what that accreditation covers is set out, register entry included, on its own page.

There is also a commercial reason to prefer the honest version. A buyer who has documented a sampling plan, a laboratory scope and a loading record has evidence to bring to a supplier, an insurer or a customer when something goes wrong. A buyer who relied on a guarantee has a sentence.

Where an inspection earns its place

Inside those limits, a pre-shipment inspection reduces risk in ways that are hard to get any other way. It refuses lots that are bad on average before freight is spent on them, it draws the laboratory sample itself, from cartons it picked at random in the load rather than from a crate the supplier set aside, and it fixes the identity of the load so that a later dispute is about facts.

FoodGate Audit is an inspection body accredited to ISO/IEC 17020 that neither buys nor sells the produce it inspects. Its pre-shipment inspection applies ISO 2859 sampling unless the buyer specifies another plan, reports lot size, sample size, defects found and the accept or reject decision for each AQL class, photographs every pallet, and delivers the report within 24 hours; residue and microbiological samples go to a laboratory when the buyer orders them. A sampling plan and a laboratory scope for the season are agreed through the contact page, and where the risk sits after the seal, end-to-end inspection extends the record to destination. A report whose origin you doubt raises a separate question, answered on the page about fake inspection reports in the produce trade.

AQL (Acceptable Quality Level): inspection levels, sample sizes and the three defect classes behind every pass or fail.
Sample inspection report: the facts a report proves, laid out on a real Egyptian lot.
Is your Egyptian inspector really ISO 17020 accredited?: what accreditation covers, and how to check it in a public register.
EU residue limits on Egyptian produce: the limits a laboratory scope should be written against.

Shipping from Egypt this season? Put independent eyes at the packhouse.

Get a Free Quote →

FAQ

Frequently asked questions

If a lot passes at AQL 2.5, how many defective boxes can still be in it?

More than most buyers assume. With a sample of 80 boxes from 1,200, a lot in which 5 percent of the boxes carry a major defect, 60 boxes in all, passes about 80 percent of the time. AQL 2.5 describes the quality level the plan accepts on average, not a ceiling on the lot.

Can I ask for a larger sample or a stricter plan?

Yes. The buyer sets the sampling plan, and an inspector applies it: a higher inspection level draws a larger sample, and a lower AQL for critical defects lowers the acceptance number. Both improve the odds of catching a bad lot, and neither turns a sample into a guarantee.

What should I tell my management a passed report means?

That the sample drawn from that container, on that date, met the agreed specification under a stated sampling plan, and that the lot was identified by container, seal and lot references. It does not mean every box was checked, and it says nothing about residues or pathogens unless a laboratory analysis was ordered and is attached.

Does an inspection cover microbiological risks such as Salmonella?

Only through a laboratory. Pathogenic micro-organisms appear in 29 of the 541 EU notifications on Egyptian produce since November 2019, and 22 of those 29 were notified once the produce was already inside the EU market. If your product carries that risk, the microbiological criteria have to be named in the inspection order.

Does a loading inspection cover what happens during the voyage?

No. It records the condition and temperature of the goods and the seal number at loading, and stops there. Two of the EU notifications on Egyptian produce describe a temperature failure on a frozen consignment, one of them during transportation, which is a risk only a record running past the seal can document.

Sources

FoodGate Audit, Egyptian produce RASFF dataset, 541 notifications validated from 28 November 2019 to 22 September 2026, counts by hazard category and classification recalculated on 4 October 2026: https://foodgateaudit.com/egyptian-produce-open-data.html
European Commission, RASFF Window, notifications on products originating from Egypt, consulted through the dataset above: https://webgate.ec.europa.eu/rasff-window/screen/search
ISO 2859-1:1999, Sampling procedures for inspection by attributes, Part 1, Tables 1 and 2-A (sample size code letters, single sampling plans for normal inspection): https://www.iso.org/standard/1141.html
Commission Directive 2002/63/EC of 11 July 2002 establishing Community methods of sampling for the official control of pesticide residues, Annex, section 2: http://data.europa.eu/eli/dir/2002/63/oj
FoodGate Audit, AQL glossary entry, reference case of a 40 foot mango container: https://foodgateaudit.com/glossary/aql.html
EGAC register entry for Food Gate Audit, Inspection Bodies 032506, consulted 4 September 2026: https://egac.gov.eg/en/entity-details/?tc=apAtggxdeHRVKtBoK2XMa4gf9Nxk6utnXfHjE9mI