The price of putting your product in front of a thousand people on Facebook went from $11.82 to $14.19 over the course of 2025, a 20% increase measured across nearly 35,000 ecommerce brands 1. What came back moved almost not at all. Every dollar spent returned $1.84 at the start and $1.86 at the end, and the cost of buying a single order went from $37.80 to $38.19 1. You paid a fifth more to stand in the same place.
That is the current shape of Facebook ads for ecommerce, and it decides which stores should be adding budget and which should be leaving the channel alone until something changes in the product or the pricing.
The increase landed on all fifteen categories at once
Every one of the fifteen industries in that dataset saw the cost of reaching a thousand people go up 1, which rules out the first explanation most owners reach for, that their ads got tired and need a refresh. When apparel, supplements, home goods, and everything else move together, the thing that changed is the price of the space, and no amount of new creative buys back a market-wide 20%.
The rest of the picture in that data is a store most owners would recognize. The average order came to $71.69, and 1.6% of the people who landed on a product page bought something 1. So roughly 62 visitors have to show up for one to check out, and at $38.19 to produce that order, Meta is charging you about 53 cents of every dollar the customer spends.
Do the subtraction on one order before you do it on a month
Here is the same benchmark run as a P&L line rather than a marketing report, using stated assumptions: the $71.69 average order, the $38.19 cost to acquire it, a 40% gross margin, and no shipping subsidy, payment fees, or returns counted against it.
Forty percent of $71.69 leaves $28.68 after you have paid for the product itself. Subtract the $38.19 it cost to find the buyer and the order is $9.51 underwater before a single dollar of rent or payroll comes out. Add back the shipping and the 3% payment processing you just excluded and it gets worse. Plenty of well-run stores carry exactly that margin structure, and the dataset average walks straight through it.
Turn the same numbers around and they give you a floor. Dividing the $38.19 acquisition cost by the $71.69 order says you need a gross margin above roughly 53% for a first-time customer to break even on their first purchase. Below that line the first order is a loss you have chosen to take, and the only thing that makes it a good choice is a second order.
The margin below which we would tell you to skip Meta entirely
If your gross margin sits under about 50% and a first-time buyer does not come back inside 90 days, do not run Meta ads at all. No audience setting, no creative refresh, and no agency changes that answer, ours included. A store in that position is being asked to fund a $38 customer with $28 of margin, and the fix lives in the price list or the supplier agreement rather than the ad account.
The version of this test with your own numbers in it takes about four minutes in our ad spend calculator, which will tell you not to advertise when the inputs say so. If you have never put your real gross margin next to your real repeat rate, that is the single most useful thing you can do with the next half hour.
What a store keeps is smaller than the margin on the invoice
Across 299 direct-to-consumer brands running $231 million in paid media in the first quarter of 2026, the median contribution margin came in at 29.3% 2. Contribution margin is what is left of an order after you have made it, packed it, shipped it, and paid the card processor, before anything goes to rent or payroll. Twenty-nine cents on the dollar.
Two different datasets, so treat the combination as directional rather than exact. Even so, apply that 29.3% to the $71.69 average order and the median brand keeps about $21 per sale. The cost to buy that sale was $38.19. Which means the median direct-to-consumer brand is roughly $17 in the hole on a first purchase and is running its entire business on the reorder.
That is a legitimate model. Subscription coffee and skincare refills survive it comfortably. It stops being a model the moment nobody has measured what share of buyers come back, because then the loss on order one is a loss carrying an optimistic name.
Two-thirds of the budget sitting in one auction
Brands in the Triple Whale dataset put 68.31% of their advertising budget into Meta 1, and the Common Thread group looks similar, with 58.71% of all spend going to Meta acquisition 2. Putting two-thirds of your customer acquisition inside one company's auction is a concentration risk, and most of this industry sells it to you as best practice. You would not put two-thirds of your inventory with one supplier who reprices without notice and owes you no explanation, which is exactly what a 20% cost increase across all fifteen categories was.
We are not arguing for spreading a small budget thin. Our position on our services page is the opposite, that most accounts should spend less on fewer things. The distinction is between concentrating because one channel is genuinely where your buyers are and concentrating because nobody ever tested the alternative. A store that has never spent a month finding out what a customer costs on search, or through the email list it already owns, is calling Meta efficient on the strength of familiarity.
What has to be true before you add budget
Two things, and both are numbers you can pull without asking anyone for help. The first is your gross margin per order after shipping and payment fees, which needs to clear your cost to acquire a customer with enough room left to cover overhead. The second is the share of first-time buyers who order again within 90 days, because that is the number that lets a store with thinner margins outbid a store with fatter ones.
If both look healthy, scale is the right call and the rising cost of reaching people is a tax you can afford. If the repeat number is unknown, find it before you raise a budget. We do this as the first pass on every ecommerce audit that comes through our Atlanta ecommerce practice, and the answer changes the recommendation more often than the ad account ever does.
The figure worth arguing about at your next planning meeting is the 68.31% 1, and whether you would still be comfortable with it if the next repricing came in at 30% instead of 20%.