Anatomy of a Meta ad: the surface area of experimentation

Yesterday we established that creative is the last lever you own. Today we open it up and count the knobs — because you can only improve the parts of a thing you can name.

1One "ad" is a stack of independent variables

On Day 1 we landed the inversion: targeting and bidding have been absorbed by Meta's automation, so creative is the only lever left with real human leverage. Good. But "work on the creative" is useless advice until you can say which part of the creative. A founder who says "the ad isn't working" has said almost nothing. A founder who says "the hook holds but the click rate is dead" has located the problem to one knob — and knobs are fixable.

So start from first principles. Sit in front of a single Reel or Feed ad and ask: what here did a human decide, and could a different human have decided differently? Every answer is a variable. Here is the full list of what you control:

• Hook — the first 3 seconds of video, or the first frame of a static. The single highest-leverage variable on the surface (Day 3 proves why).
• Visual / footage — the core imagery: the clip, the photo, the scene, the product shot.
• Format — single image or video, carousel, or collection. (Reels is a placement, not a format — your 9:16 video runs there.) A structural choice, not a cosmetic one.
• Aspect ratio — 9:16 (full-screen vertical), 4:5 (Feed-optimal), 1:1 (square). Each placement wants a native shape.
• On-screen text — words burned into the creative itself: the opening title card, caption overlays, the short label or strapline pinned over the footage.
• Voiceover / audio — the spoken script, the music bed, the sound design — and the brutal fact that a large share of feed viewing happens with sound off (assume sound-off until your own retention data says otherwise).
• Captions — the subtitle track that carries the message when sound is off (which is most of the time).
• Primary text — the body copy above the creative in the Feed. (Instagram users call this the "caption" — never confuse it with subtitle captions, knob 7.)
• Headline — the bold line beneath the creative, next to the button.
• Description — the optional sub-line under the headline.
• CTA button — Shop Now / Learn More / Sign Up / Get Offer. A dropdown, but a meaningful one.
• Thumbnail — the poster frame shown wherever video doesn't autoplay (data-saver mode, some placements, search) and in your own Ads Manager previews. Meta picks one automatically unless you set it — set it.
• Destination — where the click lands: a product page, a landing page, a lead form, the app store.

That is thirteen variables, and a real one would add a few more (the page name and avatar, link preview, profile context). The point is not the exact count. The point is that "the ad" is a noun that hides a dozen decisions. And above all thirteen sits a zeroth variable — the angle: the claim or offer the ad makes. It isn't a field in any panel, and it outweighs every knob on this page; Week 2 is where we engineer it. Today is the chassis, not the engine. Below, the same Reel, exploded into the parts you actually set.

2Each knob is independent — so the combinatorics explode

Here is the consequence that matters. Because these variables are independent, they don't add — they multiply. Change the hook and hold everything else constant: you have a new ad. Swap the format from a single image to a carousel: another new ad. The space of possible ads is the product of your options on every axis, not the sum.

Make it concrete. Take a generic skincare brand running one product. Say you have 3 hooks you believe in, 2 visual treatments (a studio shot and a phone-filmed UGC clip), 2 aspect-ratio cuts (a 9:16 vertical and a 4:5 feed cut), and 4 CTA buttons worth trying. That is not 3 + 2 + 2 + 4 = 11 ads. It is:

3 × 2 × 2 × 4 = 48 distinct ads

From four small decisions. Add three primary-text variants and you are at 144. This is the surface area of experimentation, and it cuts both ways. It is exhilarating — a near-infinite test space where the next breakout is always one combination away. And it is a trap — because nobody can make, judge, or learn from 144 ads by hand without a system. That tension is the whole reason this course exists: Week 2 turns the explosion into something you produce on purpose, and Week 4 turns it into something you learn from.

Notice too that the knobs are not equal. The hook (first 3 seconds) decides whether anyone sees the other twelve variables at all — most of the audience scrolls past before the second clip plays, and Day 3 gives you the exact metric (hook rate) to see this in your own account. The CTA button is a four-option dropdown. Treating them with equal effort is a mistake of proportion; we'll grade each knob's leverage precisely tomorrow.

3This is your shared vocabulary for the next nineteen days

There is a second, quieter reason today matters as much as any lesson in the course. This list is the vocabulary the rest of the course is written in. When Day 3 talks about a "weak first frame," it means the hook. When Week 2 builds one concept into fifty assets, it is permuting these exact knobs. When Day 4 introduces the Creative Genome — the nine-axis tagging schema that makes every later lesson's learning loop possible — several of its axes are the variables on this page: format, visual treatment, hook type, CTA.

So today's list and Day 4's genome are two views of the same object. Today is the physical anatomy — every field you fill in to ship an ad. Day 4 is the diagnostic anatomy — the labels you attach so that, months later, you can ask "which of these parts actually earned the result?" and get an answer. You cannot have the second without first being fluent in the first. A surgeon names every structure before cutting; a buyer names every knob before testing.

Internalise the names now. From here on we will say "hook," "treatment," "primary text," "thumbnail" and assume you can point to each one on a live ad without hesitating.