Six Retailers Named Fuel. None of Them Named the Meter.
Six public retailers flagged fuel as a second-half margin headwind in three weeks of earnings calls. Across all six, the electricity bill at the store never came up.
Between August 27 and September 2, six publicly traded retailers told analysts that fuel or freight would pressure margins in the back half of 2026.
Dollar General's CFO, Donny Lau: "fuel costs continue to be a little bit of a pressure point for us in the back half."
Dollar Tree's CFO, Stewart Glendinning: "the outlook for fuel rates is elevated relative to when we last spoke in May and therefore, an incremental headwind."
Abercrombie & Fitch's CFO, Robert Ball: "freight has been a bit of a headwind for us here. The rates have remained elevated."
Walmart put a number on it. More than $2 billion in incremental fuel-related costs this year, above and beyond original guidance assumptions.
Six companies. Three weeks. One shared cost story. And across all six calls, nobody said what the stores themselves were spending to run HVAC, lighting, and refrigeration.
Freight is legible in a way site energy isn't
That absence isn't an oversight, and it isn't a failure of attention. Freight is simply easier to talk about.
It arrives as one line item. It has an owner, usually in supply chain. It moves with a public commodity price an analyst can look up, which means the CFO explaining it has a shared reference point with the person asking the question. There's an established vocabulary for it, and everyone on the call already speaks it.
Site energy has none of that.
It arrives as hundreds of separate utility bills across a dozen rate structures and as many jurisdictions. It's owned by facilities, which usually reports somewhere other than finance. It gets coded as fixed overhead, which is the accounting equivalent of deciding not to think about it. And there's no shared vocabulary, so a CFO who wanted to discuss it on a call would have to invent the framing first.
So the cost the operator can't control gets named, and the cost the operator can control doesn't get named anywhere.
The one you can't control versus the one you decide
Diesel in a truck you don't own is a rate. You absorb it. You can hedge a little, route a little smarter, renegotiate at contract renewal. But the number is set somewhere else, by markets you don't participate in.
HVAC runtime at a store you operate is different. It's a decision, and it gets made and remade every fifteen minutes of every day, whether anyone is watching or not. Somebody set that thermostat, or nobody did and the installer's default is still in there from opening week. The compressor that short-cycles all afternoon is making a choice on your behalf. So is the rooftop unit that runs full-bore for four hours after the last customer leaves.
None of that shows up on an earnings call. All of it shows up on the bill.
And the underlying rate is moving too. Commercial electricity averaged 14.19 cents per kilowatt-hour in June 2026, against 13.54 cents in June 2025. That's up 4.8% year over year, on power you're already buying.
Absorbing cost pushes the pressure straight into opex
There's a second thing worth pulling out of these calls, because it changes where the pressure lands.
Dollar Tree's Glendinning, on the same call: "tariff rate is absorbing inflation and helping us to maintain value across key categories."
Read that as an operating decision rather than a pricing one. When a retailer chooses to absorb input cost instead of passing it to the customer, the margin has to come from somewhere else. It comes from operating expense. Labor is already tight and cutting it has a visible cost at the store. Occupancy is contractual. Which leaves the lines nobody has instrumented.
Site energy is one of the largest of those lines and one of the least examined. It's being asked to absorb pressure from decisions made three levels up, and it's being asked to do it without anyone having established what it should be in the first place.
What a legible energy number looks like
The reason site energy stays out of these conversations is that nobody has handed finance a number worth naming. Fixing that is less complicated than it sounds, and it starts by changing the unit.
Give it a unit. A dollar total isn't a unit, it's a lump sum, and a lump sum can't be compared across sites, seasons, or years. Kilowatt-hours can. Until site energy has a unit attached to it, it stays filed under overhead, and nothing filed under overhead ever gets interrogated.
Report it per site and per operating hour, not per month. A monthly portfolio total is unmanageable by construction. Kilowatt-hours per site per operating hour is a number a district manager can actually be held to, and it survives comparison across stores of different sizes and hours.
Separate occupied from unoccupied. A retail box is occupied ten or twelve hours a day and energized for twenty-four. The unoccupied window is where controllable load hides, and it's the one place where a change creates no customer-facing risk at all.
Name an owner. Part of why freight makes the call is that someone's name is attached to it. Site energy rarely has that person, and unowned numbers don't improve.
Where to start
Pull last twelve months of utility data for ten sites. Pick a mix: your best performer, your worst, and eight in between. Split each bill into consumption and rate. Then look at what the buildings were doing during the hours nobody was in them.
That exercise takes about a week and it produces the first defensible energy number the finance team has ever seen. That number is the entry ticket. Once site energy has a unit, an owner, and a trend, it can be discussed in the same rooms where freight already gets discussed.
None of this touches your freight costs. Fuel is going to do what fuel does, and GlacierGrid has nothing to say about diesel. The point is narrower: while six operators spent three weeks discussing a cost they can only absorb, the one on their own roofs went unmentioned. It's still there, it's still moving, and it's the one that answers to a decision.
Run a free 90-day pilot. Pick a handful of sites, set success criteria you could actually fail, and find out what your unoccupied hours are costing you. Start here.
