Ad trends 2025: what worked in paid advertising

Why most people work hard on the wrong things

This is Curve's trend report for 2025, a look back at what actually worked in paid advertising that year. The principles below are taken straight from how we worked with our clients.

Most businesses do not struggle because they are not working hard enough, they struggle because they are working hard on the WRONG things. They focus on small, insignificant details while ignoring the big, fundamental problems that are actually stopping them from growing.

What got you from 0 to 1 million in revenue, or from 1 to 10 million, or from 10 to 100 million, is NOT what takes you on to 200 million. Every growth stage requires you to solve a completely new set of problems, and the biggest problems are often the ones you do NOT see yet.

This guide does not give you yet another random tactic or a "hack" promising quick results. It shows you how to avoid the fundamental mistakes that strangle your growth, and how to build a system that scales, no matter how big you want to get.

Video walkthrough of the whole document: watch the walkthrough on Loom.

5 common mistakes that ruin advertising, and how to fix them

1. No intention when you test. Are you testing different concepts? Are you testing different angles? You need to do all of this to have predictability when scaling. It is part of a bigger strategy, and it is the foundation of a well-functioning system.
2. No iteration, you do not do MORE of what works. Every time you have an ad that works, you want to multiply it tenfold with many new variations. A little preview: you will test an angle (for example "quenches your thirst", "tastes great" or "the healthiest on the market"), and if the angle works, you run it through the 12 concepts further down in this document.
3. Not knowing how the algorithm works.
4. Not knowing what works (poor tracking).
5. Prospecting and retargeting are mixed together.

The difference between an angle and a concept

• An angle is the reason someone buys.
• A concept is how the ad communicates the angle.

Far too few people know this distinction. It is valuable because it opens up an ocean of possibilities. Biggest of all: running winning angles through many different concepts. That way you can easily multiply your volume, with a far higher probability of success.

How the algorithm works

1. Understand the business model of the social platforms, especially Meta. We advertisers are the customer, but the product (most important) is the user. If you create a bad user experience, the user leaves the platform. As a result there are fewer users on the platform, and fewer people want to advertise there. That is bad business for Meta, and not what they want. So focus number one is: keep the user on the platform, do NOT scare anyone away. (That is exactly why LinkedIn does not like links in posts, you are sending users out of their platform.) A secondary focus for the platform is creating sales for the advertiser. But they know this is best achieved through a good user experience. The smoother a buying experience is, the more users stay on the platform, and the more advertisers get sales. Everything is connected.

2. The content does all the targeting. If an ad says "how busy moms can make healthy food for the whole family", the ad will be shown to busy moms who care about healthy food for their family. We even run some ads for the agency, and they are completely BROAD, no targeting. They talk about advertising for businesses spending 200,000 kr/month on ads. It surprised me: the ad converts well with highly qualified businesses.

3. The algorithm knows what you, I, your aunt and your mom prefer to see in our feeds. I almost only get video ads, because I stop to watch them more often and click more often. That is good business for Meta: their customers (advertisers) make good money showing me video ads. Because I do buy! This is why you want images, videos and carousels. But not only that, you also want many different angles, concepts and twists on your ads, because there is no ONE SIZE FITS ALL. When the algorithm knows what each individual person loves to see, it will manage to find it for the user. But not if you are running a few generic ads.

4. If you do not have a wide range of ads, your profitability goes down. You give the algorithm less to work with. Fewer ads = fewer data points = worse matching to user preferences. The algorithm has one goal: show people what they like to see, what keeps them on the platform, but also gets them to buy from you. With few ads you paint the algorithm into a corner where it cannot deliver relevant ads to each user. BUT: do you have 10, 20, 50 variations? Then the algorithm finds the one that matches that specific user. Boom! High relevance. Low cost.

More variation = more relevance = lower ad cost = higher ROAS. You are letting the algorithm optimize for you, but it cannot do that if it only has one card to play.

Prospecting best practice

• Exclude your customer list from all TOF prospecting campaigns.
• Set your Advantage+ shopping campaign to a 0% existing customer cap.
• Define Engaged Audience and Existing Customers.

Retargeting best practice

Run a MOF campaign (middle of funnel) targeting pageviews, engagement and so on.

• Here you need to uncover 10 questions people at this stage have. Simple questions you might think are self-explanatory. Examples: Who is the product best suited for? Who made the product? How does the product work? Are there any guarantees?
• Answer each question uncovered above in its own ad. 10 ads in total.
• Create 10 ad sets with a hard frequency cap of 1, and give them a low budget.
• Let each ad set contain one single ad.
• Run this as MOF.

Pro tip: Hire someone to act like a person considering buying your product. Have them go through reviews, different websites and your own site, and find out which questions come up. Answer those questions! You cannot spot every objection yourself when you work with your product every day.

Run a BOF campaign (bottom of funnel) targeting carts, trials and so on.

• Here you need to uncover 10 questions people at this stage have.
• Focus on ads that trigger action. Tailor the message. Social proof.
• Answer each question uncovered above in its own ad. 10 ads in total.
• Create 10 ad sets with a hard frequency cap of 1, and give them a low budget.
• Let each ad set contain one single ad.
• Run this as BOF.

The 3 phases of scaling, from idea to 10,000 kr a day

1. Testing system
2. Iteration
3. Scaling (technical)

What scaled in 2025: a thorough review

VOLUME. It is not one single thing that will move you up and forward. Now you know what the big trend is. What we spend 200,000 kr a day on.

You do not need to look for some magical one-time fix. You have the most fundamental factor in place. You know what the very best are doing, and you can do the same yourself. You know you are on the right train. You are on the right track.

The last thing you want is to have something fundamentally wrong while obsessing over lots of small things that do not really drive big growth. Because, remember: the longer you sit on the wrong train, the more expensive it gets to come back home. The longer you have underlying problems in your marketing, the more money you will bleed.

Next steps

Want to see what these principles look like in practice? See the results in our cases, including Collagen for Hund. Whitelisted ads and creative volume are at the core of our street interview ads and UGC production. You will find more guides and reports on the blog.