Everything below is two feeds run against each other: the Queensland Government price feed — the price each station reported it was charging — and the published Brisbane Terminal Gate Price, the wholesale price servos buy at. 352 aligned business days.
TL;DR: Wholesale sets the floor your bowser price sits on, and it takes about a week to get there. But wholesale barely moves your price day to day — retail changes are 3.6 times as jumpy as wholesale, and almost all of that extra jump is the local discount cycle, not the barrel.
Here is the question a driver put on the app reviews, word for word:
“Why is it that prices jump in some instances when the barrel price is still way down?” — Petrolspy iOS 5★, 2024-12-13
I drive a diesel D-Max half the week, so I feel the flip side of this every time the barrel drops and the bowser doesn’t. I wanted to know, with numbers, exactly how much of your pump price is the barrel and how much is something else. The answer surprised me: less of it is the barrel than the anger assumes.
The claim
The folk model is a straight pipe: crude goes up, the wholesale price goes up a few days later, your bowser follows. Rockets up, drifts down — but at bottom, the barrel is supposed to be the thing driving the number.
“But really the wholesale price forms only a small percentage of the retail. So why do prices jump by almost 15% on a weekly basis.” — ‘kilotango’, Whirlpool
That driver is half-right and it’s the good half. Wholesale is a big share of the price level — but it explains almost none of the weekly jumping. Those are two different questions, and the data pulls them cleanly apart.
What we checked
Two series, one grid of business days.
Wholesale is the published Brisbane Terminal Gate Price for ULP — the price before the servo’s own costs. It is published on weekdays only, so every day count and every lag in this post is in business days, not calendar days. Over the window it ran from a low of 147.6 to a high of 182.5 c/L, averaging 162.3.
Retail is a daily Brisbane-metro median rebuilt from actual reported price changes — for each of roughly 473 stations within 30 km of the CBD, the last reported price each day, held until the station next changed it, then the cross-station median. Across the window that panel carries 436 stations a day on average (414 to 440), which lines up with what the cycle post found on the same method: 430-plus present daily.
Then I asked two separate things of them. First: how long does a move in wholesale take to show up in retail? Second: how much of retail’s movement does wholesale actually account for?
One honest wrinkle up front. You cannot just correlate the two raw lines — they share a slow downward drift over the window, and retail also carries a sawtooth cycle that wholesale doesn’t have, so a raw correlation is a spurious-regression trap. So I stripped the retail cycle out with a rolling average and, for the primary read, compared week-on-week changes rather than levels — which kills the shared-drift artefact.
What the feed shows
The lag is about a week. Strip the cycle out of retail and line the underlying trend up against wholesale, and the correlation peaks at r = 0.31 at a lag of 3 business days. Do the stricter test — compare week-on-week changes, which throws away any shared drift — and the single clean peak lands at 4 business days. So the honest answer is a range, not a single day: 3 to 5 business days, call it about a week, for a wholesale move to start showing up at the bowser. That is not a precise number and I won’t pretend it is one — the two methods disagree by a day, which is exactly why I quote a range.
But the coupling is weak. Here is the part the anger doesn’t expect. Once you difference the series properly, that peak correlation is only 0.12. Wholesale explains a small share of what retail does day to day. It is not that the barrel takes a while to arrive and then drives your price — it’s that the barrel sets the slow floor your price sits on, and something else does the day-to-day work.
That something else is the cycle, and the volatility numbers show it in one line:
- Retail median daily change, standard deviation: 6.20 c/L
- Wholesale daily change, standard deviation: 1.74 c/L
Retail is 3.6 times as jumpy as wholesale on a day-to-day basis. Wholesale ambles; retail lurches. And wholesale has no sawtooth in it at all — so that extra movement is not the barrel arriving. It is the local discount cycle, a QLD retail behaviour that rides on top of the wholesale floor. (The cycle itself — its size and timing — is its own subject; the cycle post owns that number, about 45 c/L trough to peak.)
Where people read the lag wrong. When wholesale ticks up and the bowser jumps the same week, it’s tempting to call it cause and effect. But retail was going to lurch that week anyway — it moves 3.6 times as hard as wholesale on any given week. RACQ’s Dr Ian Jeffreys made the timing point during the March 2026 Middle East price spike specifically:
“There is no good reason for fuel companies to be increasing their prices just days after conflict broke out… that usually takes around two weeks to flow through to bowsers here in Australia, not two days.” — Dr Ian Jeffreys, RACQ, March 2026
That was about one conflict-driven spike, not a standing rule — but the mechanism it points at is the one the data shows: a real wholesale move takes a week or so to trace through, so a same-week jump the size of the whole cycle is a pricing decision, not physics.
The verdict
A barrel of oil reaches your bowser like this. Crude and the Aussie dollar move the wholesale price. Wholesale sets a slow-moving floor under Brisbane retail — with about a week’s lag — but it explains only a small share of what your price does on any given day. On top of that floor sits the local discount cycle, and that is where nearly all of the day-to-day movement lives: retail lurches 3.6 times as hard as wholesale, and wholesale has no cycle in it. When the barrel is “way down” and your bowser is up, you are almost always looking at the cycle sitting high on top of the floor — not the floor itself.
So what
For your tank, this splits cleanly by fuel.
If you drive petrol, the barrel is not your lever — the cycle is. Tracking crude will not tell you when to fill; where you sit in the local cycle will. That is worth far more to you than watching the wholesale trend.
If you drive diesel, it’s the mirror image and it’s why diesel is my beat. Diesel barely cycles, so wholesale is most of your story — you watch the slow floor and shop the spread between servos, because there’s no cheap-Tuesday dip coming to rescue you.
“Fuel 2km from me is 157.9/L, yet if i drive my diesel ute 12km down the road… using the app once has saved me $9.80 for 70L of diesel.” — Petrolspy Google Play 5★
Either way, the money is in the spread between the cheap servo and the dear one down the road, and that spread is real and visible. Pinly shows every station’s current price on the map from the Queensland Government feed — the price the station reported, not a driver’s guess — so you can see the spread instead of guessing at the barrel.
One honest limit. The wholesale-to-retail link here is genuinely weak, not just slow — the peak differenced correlation is 0.12, so wholesale is the floor, not the driver, and I won’t overstate it. The lag is soft too: 3 business days one way of measuring, 4 the other, so I quote 3 to 5, about a week, and no tighter. This is period-conditional — 2025 was a quiet wholesale year with too little movement to trace, and the signal is carried by the more volatile late-2025 and 2026 stretches. The gross gap between retail and wholesale over the window averaged about 22 c/L, but that is a raw price difference, not profit — GST, excise treatment, freight and running costs all live inside it and I haven’t split them out, so don’t read it as margin. Crude and the AUD are named here as the macro drivers but they’re not in this feed and weren’t measured — that part is mechanism, not a number I’ve checked. Scope: Brisbane metro ULP, wholesale is the published Terminal Gate Price forward-filled across weekends, and the March–April 2026 QLD data gap is excluded, not smoothed over. Regional and remote QLD carry higher freight and thinner competition, so your town may sit wider than the Brisbane floor.
— Tom