Video, Image, Carousel or Catalog? What Works on Meta With New Customers

Created

November 16, 2026

|

Updated

November 18, 2026

|

Needle

“Should we be doing more video?” It's one of the most common questions founders ask about Meta ads. Video feels like what everyone says you should do. Images are quicker and cheaper to make. Carousels and catalog ads often get forgotten, or get used only for people who already know the brand.

So we looked at one specific job: finding new customers. We took the Meta ads our DTC brands ran from October 2025 to October 2026 and kept only prospecting ad sets: broad targeting, interests and lookalikes. We left out retargeting (ads shown to past visitors, engagers or customers) and mixed ad sets (ones that include warm audiences, even as a suggestion to Meta). We checked each ad set's actual audience settings in Meta, not just its name.

Then we sorted every ad into video, image, carousel or catalog (the ads Meta builds automatically from your product feed), and compared what each format cost and what it brought back for every $1 spent.

The short answer: with cold audiences, carousels did best, even against other ads chasing the same people. Video and images came out about even. Catalog ads looked strong at first glance, but not once we compared them fairly.

The numbers

Same audience, same budget. Carousels sold more.Sales for every $1 the other ads in the same ad set brought in. New customers only.Carousel$1.21Image$1.06Catalog$0.97Video$0.96$1.00 = the other ads

Every ad here ran in a prospecting ad set next to other ads, shown to the same new audience, on the same budget, at the same time. For every $1 the other ads in the same ad set brought in, carousels brought in $1.21 and images $1.06. Catalog ads and video came in just under, at $0.97 and $0.96. Comparing a format with its neighbours is the fairest test we have: it takes the audience, the brand and the season out of the picture and leaves mostly the format.

Here's why each format won or lost:

Why: video gets the click, carousels get the saleNew customers only. Orange = best in each box.Cost to show to 1,000 peoplelower is betterVideo$14.36Image$7.61Carousel$11.88Catalog$14.63Share of people who clickhigher is betterVideo2.1%Image1.2%Carousel1.3%Catalog1.0%Share of clicks that buyhigher is betterVideo2.1%Image1.7%Carousel3.9%Catalog4.0%Cost to win one salelower is betterVideo$34Image$36Carousel$24Catalog$37

All prospecting spend in each format combined. These totals mix brands, so they partly reflect which brands use which format. The first chart compares like with like.

What the data tells us

1. Carousels win, even with people who've never heard of you

A common objection to carousels is that they only look good because they're shown to warm shoppers. Not here. Every ad in this analysis was shown to a cold audience, and carousels still came out on top. Against the other ads in the same ad set, carousels brought back 21% more per $1. Against each brand's own average, they did 45% better, and they beat that average in most brands that ran them.

A carousel sale cost about $24, the cheapest of any format. People who clicked a carousel were almost twice as likely to buy as people who clicked a video or an image. Yet carousels got just 3% of the prospecting budget.

Why might they work? A carousel lets a new customer swipe through several products, colours or reasons to buy without leaving the feed. They arrive on your site already knowing what they want. Carousels were still a small part of the data, so treat this as a strong reason to test more of them, not a guarantee.

2. Video isn't the magic answer

Video got more than half the prospecting budget, but it didn't beat images. Against the other ads in the same ad set, video came in 4% below and images 6% above. Brand by brand, video beat images in most brands that ran both, but not all, and several were close calls.

If you've been told “you need to do more video to reach new people”, the data says: maybe, but not because video always wins.

3. Video gets the click. It also costs almost twice as much to show

People clicked on video ads about 65% more often than on images (2.1% vs 1.2%). But video cost almost twice as much to show to the same number of people ($14.36 vs $7.61 per 1,000). Once someone clicked, video turned it into a sale a little more often. Add it all up and a sale cost about the same either way ($34 vs $36).

4. Catalog ads aren't the shortcut they look like

On the raw totals, catalog ads looked great: $2.30 back per $1, and the highest share of clicks turning into sales (4.0%). But most of that catalog spend came from a single brand whose ads perform well across the board. When we compared catalog ads with the other ads in the same ad set, their edge disappeared: they came in about 3% below their neighbours.

Catalog ads can still be useful, especially if you sell a large range and keep your product feed tidy. Just don't assume they'll outperform your best creative with new audiences. Test them alongside it.

5. The format matters less than what's in it

The biggest gaps weren't between formats. They were between good and bad ads inside the same format. The typical video and the typical image brought back almost the same ($1.14 vs $1.08 per $1). But among videos, the best quarter brought back $1.68 or more, while the weakest quarter brought back $0.76 or less. Images were even more spread out ($1.85 vs $0.61). A great image beats a mediocre video, and the other way round. The format is the container. The message, the hook and the product shown are what new customers respond to.

How to choose formats for your prospecting

  1. Look at your prospecting campaigns on their own. In Ads Manager, separate ad sets that target new people (broad, interests, lookalikes) from those that target past visitors or customers. Mixing them makes every format look different from what it really is.
  2. Check where that budget goes today. Sort your prospecting ads into video, image, carousel and catalog for the last 3 months. Most brands find one format takes almost everything.
  3. Compare formats inside the same ad set. That's the fairest test you can run in your own account. Look at sales for every $1 spent and the cost to win one sale.
  4. Add carousels to your prospecting ad sets. Start with one or two next to your best ads: several products from one collection, one product in different colours, “3 reasons to buy” with one per card, before and after, or customer reviews with one per card.
  5. Don't drop images to chase video. Images are cheaper to make, about half the price to show, and performed just as well. Keep a steady supply of both.
  6. Use video where attention matters most. Video earns more clicks, so use it to introduce your brand, show a product in use, or have the founder explain why it exists.
  7. Test catalog ads, don't assume them. If you have a product feed, run catalog ads alongside your creative and let the results decide.
  8. Test the message, not just the format. When you try a new format, keep the idea the same as a winning ad, so you can see what the format changes. Then test new ideas inside the format that works.

How we did this

The right mix, made for you

Most brands don't have a format problem. They have a supply problem: not enough good ads, in enough formats, to find out what works with new customers. That's what we do at Needle: strategy, video, image and carousel creative made from your existing photos and videos, campaign setup, daily optimisation, and a clear review of what worked.

For peak season, we're offering a 30-day test for 10 brands:

Claim your spot →