Optimising your Google Shopping feed for margin instead of revenue

Ask an online shop which products sell best and you get a list. Ask which products make the most money and it goes quiet. Or someone pulls an export that three people cannot agree on.

That difference is where most Shopping budget leaks away.

The algorithm does not know your margin

A Shopping campaign gets a feed from you with titles, prices and availability. What is missing is what you earn on each product. So the algorithm optimises for revenue. And revenue says little about whether you are having a good year.

The result is predictable. Your bestsellers with a ten per cent margin get most of the budget, because that is where most of the revenue comes from. Your products with a forty per cent margin sit at the bottom of the feed and get whatever is left. On paper, your revenue per advertising euro goes up. In your accounts, nothing happens.

Putting margin into the feed

The fix is to put your margin in the feed, so your bidding can take it into account.

In practice, that means an extra field with the margin class for each product, and campaigns or product groups split by that class. Products in the highest class get room. Products you barely earn anything on get a cap. Products you lose money on are removed. That last step feels counter-intuitive, because they sell well. That is exactly why they cost you money.

One step further is to send the actual gross profit as the conversion value, instead of the order value. The system then optimises directly for what you keep. That is the purest form. This does require that your margins are correct and that someone keeps them up to date.

What else belongs in that feed

While you are working on the feed anyway, there are three things that almost always pay off:

• Titles that start with what people type in: brand, type, size, in that order. Your internal product name does not belong there.
• Stock levels that are correct. Advertising something that is sold out is the most expensive way to lose a visitor.
• Products you don't want to sell, removed from the feed. Every online shop has leftover stock, display models and items with a long delivery time. Those do not belong in an ad.

None of the three is exciting. All three can be done in a day, and they already pay off the following week.

The win-back flow nobody builds

There is still money to be made once the ad has done its job: with people who add something to their basket and then leave, or who start a trial and don't continue.

You have already paid for that group. The click was bought, the attention was there, and so was the hesitation. Reaching them again costs a fraction of what a new visitor costs. You set it up once and it keeps running.

The cheapest customer is the one you almost had.

With an online shop, we almost always start here, and only then look at the budget. Scaling up a campaign that bids on the wrong products only scales up the problem.