The call you never want to take comes at 6:42 am. The vessel, loaded three days ago, crossed to Qingdao and is now sitting at berth waiting for the discharge sample to clear. The message is short: ash content is out of spec, cargo rejected pending commercial resolution, and the rest of the day disappears into a scramble. You already know how it ends. Credit note, rebate, an uncomfortable call with the commercial team, and a quiet note in the customer file that won’t come up again until renewal negotiations in eighteen months.
If you work in bulk commodity logistics, you have either taken that call or you have watched a colleague take it. And the single most frustrating thing about it is almost never that the product was bad. The product was fine at the pit. The stockpile analysis looked clean. The blend plan matched the contract. Somewhere between the face sample and the vessel holds, something drifted, and nobody saw it in time.
This playbook is for the logistics manager who has the analysis data, the survey results, and the shipping programme, yet still cannot answer the single question that matters: will this next vessel load on-spec? It walks through why quality information fails to travel from mine to vessel in most operations today, what the cost actually looks like on the P&L and in the customer relationship, and then, in concrete, staged terms, how to build a continuous quality thread that catches deviations days before loading rather than weeks after rejection. No rip-and-replace. No technology theatre. Just the operational shifts that stop off-spec cargo from reaching the vessel in the first place.
Before you can fix the problem, it helps to sit with what actually happens on the day a cargo is rejected. From your end, it is a phone call. From the customer’s end, it is something more corrosive: a confirmation that your quality assurance does not work the way you said it did.
The buyer, typically a steel mill procurement team, a utility’s fuel desk, or the trading house acting on their behalf, receives the discharge sample result along with the certificate of quality that your team issued at the load port. The gap between the two is where the conversation lives. If the discharge result is close to spec, it is a polite negotiation about rebate percentages and blending allowance. If the gap is wide, the tone shifts. Japanese and Korean procurement teams in particular treat a significant off-spec event as a system failure, not a product failure, and their supplier audit process the following year will reflect that.
That is the part most operational teams don’t see. The off-spec event itself is expensive: the rebate, the demurrage on the delayed discharge, and the rail cars held in the yard while the dispute resolves. But it is the second-order cost that matters. A customer who has been through two material off-spec events in eighteen months is already having quiet conversations with your competitors. And they will not tell you that until the offtake is up for renewal.
The deepest frustration, as most logistics managers describe it, is that the data was almost certainly there. The pit analysis flagged a soft boundary. The stockpile build-up note mentioned that a different seam was added last week. The preliminary load port sample showed a value the surveyor thought was within tolerance, but no one double-checked it against the contract spec. Each of those signals lived in a separate system, an email, a Word document, a spreadsheet tab, a handwritten note, and the person who could have connected them was not in the room at the moment it mattered.
To fix a broken thread, you have to understand where it breaks. In almost every bulk commodity operation, the same five rupture points show up.
The geologists run the face samples, the lab produces a report, and someone on the mine admin side emails it to a distribution list. It is a PDF. It sits in an inbox. Nobody questions the numbers as the lab is trusted, but the numbers never enter a system that can be queried later. When the logistics team needs to know what the ash content on the current reclaim face was three days ago, they cannot look it up. They ask. Somebody finds the email. Somebody reads the number aloud. Somebody writes it down.
That is not a criticism of anyone in the chain. It is a consequence of how analysis data has always been produced, as a technical document for a technical audience, delivered to people who will then use their judgment on what to do with it. It works beautifully when the operation is small, and the people involved sit in the same building. It breaks the moment volumes scale or the team is distributed across sites.
The blend plan for next week’s Capesize was calculated last Tuesday. At that point, Stockpile C3 was 82 per cent of the way to its target grade, with 14,000 tonnes of late-addition material that the blend assumed would integrate evenly into the reclaim profile. By this Monday, the stockpile will have had a further 9,000 tonnes added, some of it from a different seam, as the original plan’s feed had been temporarily diverted for maintenance. Nobody redid the blend calculation. Nobody raised the change in the shipping programme review because no one in the review was tracking stockpile movements at that level of detail.
One of the most common failure modes in bulk commodity quality is that the physical stockpile evolves faster than the model of the stockpile. In iron ore, it manifests as drift in iron content or moisture excursions. In coal, as ash or calorific value. In agricultural commodities, such as moisture, protein, or foreign material. The mechanism is the same: the blend plan is a calculation that assumes the stockpile is a known, static entity. The stockpile is a living, moving, daily-changing entity.
The independent surveyor arrives at the berth the day before loading, takes representative samples during the load, runs preliminary checks, and issues the certificate of quality. They do their job to contractual standards. They cannot, however, second-guess the upstream stockpile profile, and they are not necessarily aware of the specific contract spec nuances for this cargo unless someone has briefed them. If the surveyor’s preliminary result falls within the surveyor’s general tolerance band but outside the specific contractual tolerance for this offtake, they will typically issue the certificate unless the shipper flags the gap. The shipper cannot flag the gap if they are looking at a different contract document than the one the surveyor has on their tablet.
Quality assurance in a resource business sits under the technical or operational function. Logistics sits under the commercial or supply chain function. They use different platforms, attend different meetings, report to different directors, and, in most operations, speak to each other formally once a week at the shipping programme review. Off-spec risk lives precisely in the handover between these two functions, and the handover is almost always email-based, verbal, or a spreadsheet neither team fully owns.
A good quality team will flag deviations. A good logistics team will act on them. But the mechanism that carries the flag from one team to the other is informal, manual, and dependent on the right individuals being in the right meeting. When people are on leave or new to the operation, the flag does not travel.
The discharge port sample, when it eventually comes back, is treated as a commercial document. It goes to the contracts team. It may or may not be copied to operations. It rarely makes its way back to the mine’s geology or processing team, meaning the upstream operation never receives the closed-loop feedback needed to tune future extraction or processing decisions. The single most valuable source of quality truth, the customer’s own assessment at destination, is siloed from the people whose decisions actually influence quality at source.
Put these five ruptures together, and you get the pattern most operations live with: a quality system that looks rigorous on paper, produces genuinely good data at every stage, and still allows off-spec cargoes to load because the data does not travel between stages fast enough or clearly enough to be acted on.
It is useful to put numbers around this because the business case for fixing the thread is almost always built in the wrong place. Most operations count the rebate and the demurrage as the cost of doing business and move on. The real cost is seven categories deep.
Against that seven-category cost, the investment required to close the quality thread is modest. Which is why a pragmatic playbook exists: and why, if you have not built one yet, the economics are almost always favourable.
A continuous quality thread is not a single product, a single system, or a single process. It is the operational discipline of making sure the quality information generated at each stage of the supply chain, mine, stockpile, rail, port, load, voyage, discharge, is captured in a machine-readable form, connected to the same cargo identifier, and made queryable by the people who need to make loading decisions.
Building it is a staged exercise. I have seen it done badly: as an 18-month ERP project that never quite worked, leaving the operation with an expensive piece of Software nobody trusted. I have also seen it done well: in six months, stage by stage, with each stage delivering a tangible capability before the next begins.
Here is the staged version.
The foundation of the thread is that every analysis result, face sample, grade control, stockpile profile, and reclaim sample is stored as structured data tied to the physical entity it describes. Not an attachment to an email. Not a PDF in a folder. A record with a timestamp, a location, a sample method, a lab, a result set, and a status.
This does not require a new lab information system if you already have one. It does require that the data from the lab system be fed downstream into the logistics view, rather than emailed as a summary. The test for whether stage one is done: can a logistics coordinator, on a Tuesday morning, answer the question, “What is the average ash content of the material that has been added to Stockpile C3 in the last seven days,” without emailing anyone? If the answer is no, stage one is not done.
In Practise, this is a data integration exercise. Typically, the lab system exposes an export. Typically, the stockpile management system has a field for analysis. Typically, nobody has joined them. The work is often a week of integration plus a month of data cleaning.
Once analysis data is structured, the stockpile model needs to stop being a weekly spreadsheet snapshot and start being a continuously updated view. Every addition to the stockpile, with its analysis, updates the blended profile. Every reclaim from the stockpile is measured against the current profile, not the profile from seven days ago.
The mathematics of stockpile blending is not new. Mining engineers have done it for decades. What is new is doing it in a place where the logistics team can see it and act on it, with alerts that fire when the blended profile drifts outside the tolerance band for the next-nominated cargo’s contract spec.
The critical output of stage two is an alertable stockpile model. When the projected ash content of Stockpile C3 exceeds 0.3 per cent of the next cargo’s upper tolerance, the logistics manager and the quality manager receive a notification with the underlying data and suggested corrective actions (divert feed, accelerate reclaim from a cleaner face, re-plan the blend).
This is where most operations see their first genuine prevention event: a cargo that would have been at-risk is caught three or four days before load, and the blend or loading plan is adjusted without any commercial exposure.
The load port surveyor is contractually independent, as they should be, but they can be operationally informed. Stage three is ensuring that the surveyor’s pre-load preliminary sample is automatically compared to the specific tolerance band for the contract covering that cargo, not a generic grade tolerance or the tolerance for the last cargo of that product type.
Practically, this means the shipping programme holds the contract spec for each nomination, the surveyor’s preliminary result feeds into the same view, and any deviation outside the contractual tolerance triggers a hold-and-review rather than proceeding to the final certificate. The surveyor remains the certificate’s author; the logistics team can stop the clock before the certificate is issued.
The test for whether stage three is done: when a preliminary sample comes in outside tolerance, does the system pause loading and request acknowledgement, or does the cargo continue to load while an email thread chases confirmation? If the latter, stage three is not done.
The discharge sample is the ground-truth signal that the supply chain spends weeks producing. Most operations receive it, file it in the commercial folder, and forget it. Stage four feeds the discharge result back into the stockpile and blend model, letting the next cargo’s calculation incorporate what actually happened with the last one.
This is where the quality system starts to self-correct. Patterns that would take a human quality manager months to spot, a slow drift in a particular seam’s ash, a recurring moisture gap at a specific port transhipment, become visible within one or two cycles. The business value here is not the individual correction; it is that the operation starts producing its own quality intelligence rather than buying it back from customers as off-spec rebates.
The final stage, and the one with the highest commercial upside, is sharing the quality thread with the customer. Not the raw data, not the internal variances, but the confidence signal: for each shipment, a pre-departure quality confirmation that the customer can see alongside their own port-arrival sample. It changes the commercial conversation from defending an off-spec result to explaining variance against a baseline both parties already trust.
It also does something the CFO will notice: it moves supplier differentiation from a sales-deck claim to an operational artefact. A buyer comparing two suppliers in a renewal conversation, one of whom can produce a traceable quality thread from mine to vessel and the other cannot, is not making a price decision anymore.
It can help to see the thread working in specific situations. The numbers here are composite, and the names are disguised, but the mechanics are drawn from real operations.
A Pilbara iron ore producer running a weekly Capesize programme had historically averaged two off-spec events per year, usually caused by a soft seam boundary that blended into the product for a few days before the geology team flagged it. The seam had been known for years; the blend plan was designed around it; the shift-by-shift tracking of which seam was active was not reaching the logistics team’s blend model.
After stage two of the playbook, the stockpile model ingested grade-control analysis in near real time. The first time the seam boundary was crossed under the new system, the alert fired four days before the Capesize load date. The blend was re-plotted with a higher proportion from a different stockpile. Loading proceeded on schedule and on spec. No conversation with the customer.
Cost of building that capability: roughly six weeks of integration work and a training day for the coordination team. Cost of the event it prevented, if it had gone through: the typical two-to-three per cent rebate on a 170,000-tonne cargo, plus associated demurrage, which is a multi-hundred-thousand-dollar figure at current commodity prices.
A thermal coal exporter had seen a steady, unexplained creep in customer-side moisture results over a four-month period. Load-port moisture was within tolerance every time. No individual sample triggered a review. In aggregate, the trend was clear once plotted, but no one was plotting it.
After stage four, the discharge sample data were being fed automatically into a cargo-by-cargo comparison view. The trend became visible within three shipments under the new system. Investigation traced the drift to a specific stockpile that had been sitting through a wetter-than-usual season, losing heat and gaining surface moisture during extended storage in the buffer. The corrective action was a change in stockpile rotation and a minor tweak to the reclaim sequence; the downstream moisture numbers stabilised within a month.
The important part of this example is not the correction itself. It is that the operation had been losing a small but consistent margin on every cargo for a full season without anyone noticing. The quality thread, once connected, revealed a loss that had been invisible because no single event was large enough to be flagged.
Every senior logistics manager has heard versions of this conversation before, and most carry a reasonable scepticism. A few of the objections are worth addressing directly.
If this article has resonated, the single most useful thing to do in the next working week is not to commission a platform, not to write a business case, and not to schedule a vendor meeting. It is to run a one-hour audit of the thread’s current state in your operation.
Take the most recent Capesize cargo. Trace the quality signal from the pit to the discharge port. Write down, for each stage, where the data lived, who would have seen it, and whether it was in a form the next stage could have used without a human reading it and re-typing it. Count the number of manual hand-offs. Count the number of systems involved. Count the number of email threads.
That audit, on a whiteboard, in your own words, with your own data, will tell you whether the thread is continuous or intermittent. In most operations, the count of manual hand-offs alone is eye-opening. It is also the first honest artefact to take to a conversation with the CFO about why the current state is an exposure, not a cost centre, an exposure, and what the staged path to closing it looks like.
The goal is not to rebuild the operation. The goal is to make sure that the next time a buyer’s procurement team asks, “How do you guarantee the quality of this cargo?” the answer is not a sales-deck sentence but a living, traceable, queryable artefact that shows them exactly how.
The contamination call still happens sometimes. Supply chains are complex, and quality is probabilistic. But the operations that have closed the thread take that call once every two or three years, apologise properly, and walk the customer through the data that explains what happened and what has been changed. The operations that have not closed the thread take that call more often than they admit to each other, and every time it happens, it quietly costs them the customer relationship they thought they had built.
You already have the data. The people on your team already know what good looks like. The playbook is the set of connections between them that turns individual expertise into a system that cannot fail the way it used to.
For more on how SCIAR approaches the continuous quality thread in bulk commodity operations, see our related pieces on why stockpile blend plans fail and building a single source of truth for your shipping programme. External authority on the underlying audit expectations can be found in ICMM’s guidance on traceable supply chains and the LME’s chain-of-custody requirements.
Nick Ogle has over 30 years of experience in Enterprise IT, spanning engineering, sales, and marketing roles across Australia, the USA, and APJ for various IT vendors. Nick has also founded his own consulting business.
Nick is passionate about entrepreneurship and Software innovation that drives positive change. Currently, he is the Sales & Marketing Manager at SCIAR Systems, a Newcastle-based SAAS startup, where he is helping commercialise their groundbreaking Bulk Commodity Logistics solutions.
For more information on Nick and to find articles that have been written on the IT sector in the past, feel free to look at his LinkedIn profile or browse some of the additional articles Nick has written for SCIAR.