By Bob White, Chief Procurement Officer
A supplier's price increase almost never arrives as a demand. It arrives as a story. Feedstock has moved, freight has moved, a cracker in Asia is down for a turnaround nobody scheduled. The story is usually true. The question is whether it is true by the amount on the quote, and after twenty-six years of buying chemicals I can tell you the honest answer for most teams is: we don't know. We suspect. We push a little, cave a little, and book the variance.
We tell ourselves we've solved this, because we subscribe to a price reporting agency. I used to tell myself that too. What I've learned is that a subscription answers a question I'm not actually asking — and the gap between the two is where the money goes.
Let me put the case for it plainly, because it's real and I don't want to be misread as dismissing it. A credible benchmark gives a negotiation edges — a floor and a ceiling instead of two people asserting confidence at each other. It's worth remembering the other side of the table has those edges already; their commercial desk subscribes too, and information asymmetry has never once tilted my way by accident. A benchmark is also the only thing that makes an index-linked contract possible, which is the single cleanest way I know to stop a volatile input from turning every quarter into a fresh argument. It lets me grade a supplier honestly — an eight percent increase is either outrageous or generous depending entirely on whether the market moved three or thirteen. And run across a full category, it exposes leakage, which almost never sits in the two commodities everyone watches and almost always sits in the long tail nobody has re-priced since the last contract cycle.
So this isn't an argument against benchmarking. It's an argument that the instrument most of us bought was calibrated for someone whose job isn't ours.
The agencies — Platts, ICIS, Argus — are serious institutions and deserve to be described as such. Between them they publish well over a hundred thousand assessments a week. Platts sits so deep in the industry's contractual plumbing that firms who openly dislike its methodology keep referencing it anyway, and for settlement they're right to — there is no real substitute. Argus was first to hold its energy benchmarks to IOSCO principles, and when OPEC producers moved to its sour crude index for US sales, that was the market itself voting on method. ICIS runs a standing public consultation on its methodologies that is, frankly, better governed than most of my suppliers' pricing committees.
None of that is the problem. The problem is the design brief. A benchmark exists to close a contract, so it has to be defensible, liquid and singular — one number a whole market can point at. My job needs the opposite quality: it needs to be specific to a delivery I'm actually making. Those two goals don't sit on the same axis, and past a certain point one is bought at the expense of the other.
Here is the moment the whole thing became concrete for me. A colleague in Jakarta was preparing to negotiate a purchase, and the reference in front of him was a CFR Southeast Asia assessment. But “Southeast Asia” is not somewhere you can buy anything. It's a freight convenience — a bracket drawn on a map. The delivered price in Indonesia, Thailand, Vietnam and Malaysia pulls apart on duty, domestic capacity, port economics, currency and local demand, and it does not pull apart by some tidy constant you can hold in your head. He was, in effect, preparing to argue against the average of four markets he was not standing in. His supplier knew to the dollar what the cargo cleared at in Jakarta. That is not a negotiation; it's a formality with extra steps.
And this is not the agencies failing at their craft — it's their craft working exactly as intended. An assessment lands where trade is liquid enough to defend a single number. Where liquidity thins, the region has to stand in for the country. Sound method. Useless to the person cutting the actual PO. That single mismatch — a reference drawn at continental scale, an order placed at national scale — is the root of almost every other complaint I have.
“We cover methanol” is a sentence, not a number. A commodity isn't a price; it's a grid — grades, purities, incoterms, load ports. The agencies assess the cells that anchor the contracts that anchor the market. I purchase the cells that feed a specific plant. Those two sets overlap far less than a coverage list suggests, and you never notice the missing cell until you go looking for your exact specification and quietly settle for the neighbour.
The question I stopped asking. Not long ago I wanted methanol side by side across India, China, the US, Saudi Arabia and Germany — an entirely ordinary thing to want. It became five separate extractions, a spreadsheet, and the better part of an afternoon, and two of those five countries weren't assessed at all, so the afternoon's real deliverable was a hole in my own analysis. Nobody writes this into a business case, but questions that cost an afternoon simply stop being asked. Multiply that silence across a category team and the unasked-question tax dwarfs the invoice for the subscription itself.
A stack, not a single supplier — because the honest answer isn't “rip the incumbent out.” Where a contract settles against Platts, it goes on settling against Platts. That entrenchment is a feature I'm not paying to fight, and any vendor who tells you otherwise is selling.
For everything the settlement benchmark was never going to reach — the long tail, the country-level view, the five-market comparison I want before the call instead of after it — I've moved to a platform built for that shape of question. ChemAnalyst is the one on my screen. It prices at country level rather than regional aggregate, across a materially wider commodity set; it keeps the demand-supply and trade data next to the price, so a movement arrives with its cause attached rather than as a bare figure; and it turns that five-country comparison into a dropdown instead of an afternoon. It was clearly designed by people who assumed the user was raising a PO, not settling a swap.
I'll name the trade-off too, because I'd want it named for me. A newer platform doesn't carry three decades of contractual gravity, and you won't settle a term contract against it — the market simply hasn't agreed to, and that inertia is real and rational. What it does instead is cover the ninety-odd percent of my spend that no benchmark was ever going to settle in the first place. I stopped asking one tool to be both things.
Three habits, learned the slow way. Fix the reference at the moment of contracting, not in the middle of the dispute — grade, incoterm, delivery point, source, all of it, before either side has a reason to be difficult. Track your delta to the benchmark rather than the headline price, because the delta is the only number that tells you whether you're winning. And before you sign the next renewal on any data subscription, run one uncomfortable audit: write down every country you genuinely take delivery in and every grade you genuinely run, then count how many of them your provider actually assesses.
When I first did that count, the number was lower than I'd have sworn it would be. Most people I've handed this exercise to have the same reaction. The subscription wasn't wrong. It was just answering a question that stopped being mine somewhere along the way.