When the catalogue in Meta replaces manual creatives for furniture
A furniture shop uploads its entire range to the feed, launches a catalogue campaign, and the report in Ads Manager quickly looks better than anything before it. In the warehouse the picture is different — the sofa sets and bedroom suites sit still, while the movement is in stools, shelving units and light fittings. The average order falls while ROAS rises.
The catalogue brings back, it does not persuade
A catalogue ad is a mechanism for bringing people back, not for persuading them. Meta holds the list of your products, and it also holds the signals about which person has been in contact with which item — viewed it, put it in the basket, bought something adjacent. At the moment of display it assembles the ad from these two things. The image comes from the feed. The title and price come from the feed. Nothing in this ad was written for the specific person.
That is enough when the desire has already formed. A person who has measured the wall, looked at several sofas and asked at home recognises the picture at first glance and continues from where they got to. The catalogue saves them digging through their browser history. That is exactly why the result looks so good — this ad collects purchases that were already on their way to happening.
In front of a person who has not thought about a sofa, the same ad does nothing. A product photo on a white background does not show scale relative to the room, does not show how the upholstery looks in daylight, and does not say why this particular model solves anything. It shows an item and a price to a person who asked for neither.
Why the full feed pushes the cheap items forward
The campaign is optimised for purchase. At every impression, the system chooses which item from the permitted set has the highest probability of leading to a purchase right now. When all your products are in the set, the choice is between a stool and a living-room suite, and the probabilities are not close.
The length of the decision goes with the price. A sofa set is measured, discussed at home, and waits for a payday or a free delivery slot, and the person returns to the page several times. A stool is bought from a phone in a lift. The system does not see margin and does not see what you want to sell — it sees events and the frequency with which they happen.
From there the mechanism reinforces itself. Cheap items win the impressions, accumulate events, the model gets confirmation and gives them more impressions. Expensive items are left without data and gradually stop showing. At some point a large part of the feed is not displayed, without anyone having made such a decision.
The ROAS in the report stays decent, because revenue is divided by spend and small orders are many. The average order value falls, while the cost of handling, delivering and returning a small item does not fall in proportion to its value. This is where the number in Ads Manager and the number in the accounts begin to tell different stories.
What a manual creative does that the feed cannot
A manual creative works in the opposite direction — it creates the viewing that the catalogue later brings back. Furniture is bought with the eyes and a tape measure. An interior shot shows the size relative to a room the person recognises as their own. A video shows how the mechanism unfolds, how deep the seat is, how the upholstery creases where you sit.
The second is that a manual creative shows the situation before the product. A living room that a corner sofa does not fit into. Guests sleeping on the floor. An old seat sagging in the middle. The product photo from the feed cannot say this, because it was shot to describe an item, not to describe a life.
The third is purely technical. The creative is itself a targeting signal. A frame of a children’s room, a frame of office furnishings and a frame of solid oak find different people even with the same defined set. The feed carries no such signal — its frames are uniform by design, and the system has nothing to judge which item to show to whom, other than behaviour already collected.
Where the feed breaks for furniture
A catalogue is only as good as the list behind it, and with furniture that list breaks in three ways that I see most often.
The first is variants. One sofa in several upholstery options enters as that many separate items, which fight for the same impression in front of the same person. The random one wins, impressions are scattered and not a single variant gathers enough data. Grouping the variants returns the sofa to a single entry, and the choice of upholstery stays on the page.
The second is availability. Furniture with a distant delivery date or discontinued by the manufacturer keeps being shown until someone removes it by hand. The money goes towards an item that the person will not receive soon, and this does not show in the campaign report — it shows in the emails to customer service.
The third is the link between the feed and the pixel. If the product identifiers in the feed and the values that the pixel sends on viewing do not match, the signal about which person looked at which sofa is never connected. The campaign keeps running, but now as an ad to a cold audience with catalogue photos. That is the weakest possible combination.
The split happens at product set level
The choice is not catalogue or manual creative. The choice is which product in front of which person, and the place where this is set is the product set, not the campaign.
I keep accessories and small items out of the set used to look for new people. They work in front of people who have already bought or already browsed — there, the small order is a second order, not a substitute for the big one. Expensive collections get their own set, so that they do not compete with a stool in the same auction.
The catalogue replaces the manual creative where the desire already exists: returning visitors, abandoned baskets, previous customers, people who know the shop by name. At the beginning of the journey it does not replace it — there is no viewed item for it to build on, and the set has nobody to catch up with.
Who gets the credit
There is one more thing that distorts the judgement, and it is in the reporting itself. A catalogue ad sits closest to the purchase in time. The person saw the interior shot days earlier, came back, then saw the product ad and bought. The credit is recorded where the last touch is.
The report shows a strong catalogue and a weak manual creative. The logical reaction is to move the budget to the catalogue. After that, the flow that fed it dries up, and the catalogue also starts to fall — seemingly for no reason, because the reason is in another campaign that was cut earlier. From the outside it looks like the slow withering of the campaign that passed for the best in the account.
That is why I look at the two together, not separately: how many new people come to the website and what share of them return. A campaign that brings back cannot grow faster than the campaign that fills.
What you can check in your account today
Open Ads Manager for a period in which both the catalogue campaign and the manual one were running, and put two numbers side by side for each of them: revenue and number of purchases. Divide them. That is the average order each campaign brings. Then take the average order of the whole shop from the order system, not from Meta.
If the average order from the catalogue campaign sits noticeably below the shop’s, the choice of which product gets sold is no longer yours — the feed is making it. If it sits close to it, the catalogue works with the whole range and there is nothing to touch.
The second check is in the catalogue itself. In the breakdown by product, see how many items received impressions at all over the period and compare that with the number of items in the feed. The difference between the two is the part of the range that you maintain in the system without anyone seeing it.
Frequently Asked Questions
Can I replace manual creatives entirely with catalogue ads?
In front of people who have already been on the website or have already bought, the catalogue does the job with far less visual production. In front of people who have not yet thought about new furniture, it does not — there is no viewed item to show them. Left on its own, the catalogue exhausts the list of people it has something to bring back.
Why does the catalogue campaign show only some of my products?
The choice of product is made anew with every impression and is guided by the probability of a purchase right now. The first winners gather the data, and without data the other items do not enter the comparison and gradually drop out. Splitting the range into separate product sets is where this is stopped, because the set determines who each product competes against.
Should I remove accessories and small items from the feed?
Not from the feed — the feed is the warehouse and nothing gets taken out of it. The restriction is in the product set of the campaign for new people. Small items stay in the sets for people who already know the shop, where the small order comes after the big one, not instead of it.
Related: Advertising for furniture · Why the advertising that brought the customer to the showroom looks like a loser · Facebook advertising
