How do I scale ecommerce ads profitably?

Work out the return on ad spend at which you break even, and you have the only number that tells you whether more spend is a good idea. It comes from your gross margin, not your revenue. Take the price of a typical order, subtract the cost of goods, payment fees, packaging and the shipping you actually pay, and you are left with what an order contributes. Divide the order price by that contribution and you have the return you need just to stand still.

Everything about scale follows from that one number. If you break even at a return of 3 and your campaigns are running at 3.2, there is almost nothing to scale into, because the extra spend will land at a worse return than the average and go straight through the floor. If you break even at 2 and you are running at 4, there is real room, and the question becomes how fast you can use it.

This is why stores that cannot scale usually have a margin problem wearing an advertising costume. Raising the price, raising the average order, or cutting the cost of a delivery does more for your ability to scale than any change to a bidding strategy.

Move budget in increments and give each one time to settle, because every large change sends automated bidding back into a learning period where results are unreliable and often bad.

A workable rhythm is to raise a campaign's budget by a fifth to a quarter, then leave it for a week to ten days before judging it. Two things happen in that window. The system relearns, which drags results down temporarily, and the campaign starts buying auctions it was previously priced out of, which are usually less efficient than the ones it already had. Both are normal. Neither is a reason to reverse the change on day three.

What is a reason to stop is a return that settles below your break-even and stays there. Settles is the word doing the work. Judge on the period after the learning has finished, not during it.

The same logic applies to targets. If you lower a target return to buy volume, lower it once, by a small step, and wait. Stores that drop a target sharply usually see spend jump, results collapse, panic, and put it back, which throws away the learning and leaves them worse off than before they started.

Most scaling attempts fail because the extra budget chases the same searches and the same audience, which means paying more per order for people you were already reaching. Growth comes from finding pockets of demand that your current campaigns are not touching.

The places to look, in the order they usually pay: products in your catalogue that get no meaningful spend because they sit inside a campaign dominated by one bestseller. Search terms your feed does not currently match because the titles use your internal product names rather than the words people type. Geographies you ship to but exclude or under-fund. Devices or times of day that were capped by an old rule nobody remembers setting. Then, further out, a second channel, or the middle of the funnel where people know the category but not you.

Splitting a catalogue so mid-tier products get their own budget is unglamorous and frequently the single biggest source of headroom in an account. The bestseller was never the constraint. The constraint was that everything else was competing with it for the same pool of money.

Watch the blended picture as you scale, because platform reported return can improve while the business gets worse. Each platform counts conversions it had some contact with, and as spend rises those platforms increasingly buy the same orders as each other, and orders that were coming anyway.

The two numbers to keep next to each other are total marketing spend against total revenue, and total marketing spend against gross profit. The second is the one that decides whether scaling is working. Revenue rising while gross profit stays flat means you have bought turnover, not growth.

Also track cost per new customer separately from cost per order. Scaling almost always means paying more to reach people who have never heard of you, so the new-customer figure will rise before the blended figure does. That is expected. What matters is whether it stays inside the ceiling your margin allows.

Give each step a fixed review point, at least a fortnight out, and write down the numbers before the change. Scaling decisions made on memory of what the account used to do are how accounts drift into unprofitable spend without anyone deciding to.

As spend rises, every weakness in the store gets more expensive, because it is applied to more traffic. Scale does not create these problems. It just makes them impossible to ignore.

The common ones: a checkout that loses phone buyers, which costs little at low volume and a great deal at high volume. A free shipping threshold set by guesswork, which quietly eats the margin on every additional order. A product page that is slow because of apps added over two years. Stock that runs out on the product the extra budget is pointed at, which turns paid clicks into disappointed visitors and can pull products out of shopping surfaces entirely.

And repeat purchases. Buying more first orders is the most expensive way to grow. If a reasonable share of those buyers order again, the acquisition cost you can afford goes up, and scale gets easier at every step after. If none of them come back, you are refilling a bucket with a hole in it, faster.

This is why we treat scale as a whole-store question. More visitors is only worth buying when conversion, order value and repeat purchases can hold what arrives.