Meta Ads 2026: My Full Strategy Step by Step

Short answer: my whole Meta Ads system in 2026 comes down to three decisions. Decision one: the customer avatar. You build it before you launch the first ad, because a creative with no specific person in it has nobody to talk to. Decision two: awareness level. Five stages, one ad set per stage, inside one CBO campaign, with copy matched to what that person already knows. Decision three: destination. Product page, landing page or quiz, each in its own ad set, because they differ in click cost and conversion quality. New creatives go into existing ad sets. I show the full system live in the video Zero to $40k in 3 Days, Full Meta Strategy Step by Step, and this post is its written version.

The store I recorded this on did almost $40k in three days by my own numbers. I give that figure as context for my own results.

This is not a promise of results. Outcomes depend on product, budget, execution, market and experience.

What changed: Andromeda and the end of weekly ad sets

Meta described Andromeda in December 2024 as a retrieval engine that personalizes ad delivery and requires creative diversification (accessed 2026-08-25). That official material prescribes no account structure. Everything below is my operating blueprint, not Meta guidance.

The practical change on my side looks like this. I used to open new ad sets every week for new batches of creatives. Every set started collecting signals from zero, and data scattered across dozens of small piles. Now a new creative goes into an existing ad set when the persona has not changed. Signal history stays in one place and the new message stacks on the old one. I open a new structure only when the persona or the destination changes. A new creative alone is not a reason.

Decision one: the customer avatar

Before you switch anything on, you answer one question: who is supposed to see themselves in this ad.

Take beauty as the simplest case. The same cream can be promoted by a 40-year-old brunette with five minutes for makeup in the morning. It can be promoted by a woman in her sixties looking for mature-skin care. It can be shown by a young person trying this kind of product for the first time. Those are three different avatars and three different ads, even though the product stays the same.

The most common mistake I see from people starting in clothing: they shoot creatives with models because it looks good. That ad sells worse, because the viewer does not see themselves in it. You are not selling to the person YOU think is the customer. You are selling to the person who is scrolling and suddenly sees their own problem in your product.

How many avatars should you build? For one product you can easily map out dozens; research with Claude takes half an hour. Three is enough to start. Which one wins, Facebook will tell you, more on that below.

Decision two: five awareness levels in one campaign

My ad structure looks like this: one CBO campaign, five ad sets inside, each tied to one awareness level. Five ads per ad set. All copy in a given set speaks to one level.

Ad set	Who the viewer is	What they already know	What the ad does
1	Unaware	Does not know the problem exists	Mirrors their situation, educates
2	Problem aware	Knows the problem, not the solutions	Names the pain and its source
3	Solution aware	Compares methods	Explains why alternatives failed
4	Product aware	Reads reviews and comparisons	Removes one objection with proof
5	Most aware	Looks for offer, code, where to buy	Gives the offer and a direct CTA

An example with the same cream. The ad for the unaware says: you did not know this product existed. The ad for the problem-aware says: you have dark circles, this product works on that problem. The ad for the most aware says: offer, price, buy now. The same offer needs a different ad at every stage, otherwise you talk past what the viewer has in their head.

Another avatar is a copy of the same campaign. You duplicate the structure, swap the people in the creatives and rewrite copy for the new avatar. You do not mix two personas in one campaign.

Decision three: destination. PDP, landing page or quiz

The third axis is where the click lands. In my setup individual ad sets route to different paths:

Destination	What it is	When it makes sense
PDP	straight to the product page	Product needs no explaining
Landing page	dedicated sales page	Product needs context or an offer
Quiz	a few questions, matched offer at the end	Viewers with low product awareness

The quiz deserves its own paragraph, because it is currently my strongest path. The customer answers a few questions about their situation and gets an offer assembled for their answers. The matching runs on Claude on logic you write down. Quiz conversion comes out around 70 percent for me. That is my result on my products, not a market average, and your numbers may differ. The mechanism is easy to understand though: a person who described their own problem gets a page that answers their answers.

Each destination lives in its own ad set, because click cost and traffic quality differ between them. In one campaign Meta splits the budget itself and the data shows you which path wins.

Creatives: one scene, three people

I produce ads in Higgsfield connected to Claude. I generate UGC-style video: a specific person records themselves, talks to camera and shows the product in use. The product must be in the frame, that is a hard rule.

I shoot the same scenario with three different people and that gives three different ads for three avatars. Same scene, same message, only the human changes. The video model I currently work with is Seedance 2.5. I consider it the best tool for this kind of creative right now, but the whole system does not depend on any specific model. UGC shot on a phone with a real person works on the same principle and is often enough.

You add new creatives weekly to the set that is winning. I write more about that process in the creative testing glossary entry.

Facebook will tell you who your customer is

You do not have to guess which avatar and which awareness level wins. Watch CTR, click cost and add-to-cart in every ad set. When one level clearly pulls ahead, after about four days you move budget there and make new ads only for the winning avatar.

I judge a campaign after four to five days, not after the first hours. Killing early is the most expensive mistake in this system, because you cut a set in the middle of its learning.

Offers and page testing

The offer is a separate layer and I keep testing it. I watch what competitors do: who gives a second product for a token amount, who throws in a third for free. Shopify now has built-in A/B tests on the product page and the cart, so I put up two or three variants and watch which one sells. I assemble product pages with Claude in about fifteen minutes, so the cost of a test is render time, not a week of work.

Scaling rules

Scaling on my side looks the same as it has for a long time. I wait until blended ROAS holds above my profitability threshold for seven days in a row. Then I raise budget by at most 20 percent and no more often than every three days. I do not touch a winning campaign without a reason: every change has a written hypothesis and a rollback condition. These are my operating rules, not Meta guidance.

Live implementation

I implement this structure in front of viewers in the Building a Store With You series. Every day I show numbers and progress on Instagram and TikTok. Implementation materials and further teardowns go to Discord, entry through the community page.

What this system will not do for you

The video shows an account structure, not a result. Three things stay on your side: product choice, budget, and patience for a four-day test. I do not know your account, so I cannot tell you which awareness level will win for you. You will see that in your own data after the first run.

If you are still pricing a start, begin with how much does dropshipping cost, and I break down ad costs in how much do Facebook ads cost for a dropshipper.

FAQ

Does this structure work on a small budget?

The structure is the same, but five ad sets with five ads each need traffic before results are readable. On a small budget start with two or three awareness levels and one destination. Add more once the data starts saying something.

What is CBO and why one campaign?

CBO is a campaign where Meta distributes budget across ad sets itself. In this system one campaign serves one persona: five awareness levels inside, and budget flows where conversions are. Another avatar is a campaign duplicate, not another ad set in the same one.

Do I have to use AI for creative production?

No. The system rests on matching the message to the avatar and the awareness level. Higgsfield speeds up production for me, but UGC shot on a phone, with a real person showing the product in use, works on the same principle.

How long should I wait before scaling?

I judge a campaign after four to five days. Scaling starts after seven days of stable blended ROAS above the profitability threshold, then at most plus 20 percent of budget and no more often than every three days.

Where can I see this system implemented live?

In the Building a Store With You series on YouTube and in the community, where I post implementation materials and answer questions. Entry is free through the community page.