Creative is the new targeting

0:00 Creative is the new targeting

Here's a question that changes how you think about ads. If the platform decides who sees your ad, how does it decide? Increasingly, by looking at the ad itself. The visuals, the words, the sound, and who engages with it. Which means your creative is now your targeting. In this lecture you'll learn how to build a genuinely diverse creative portfolio, using concepts, angles and variants, a simple matrix to find ideas nobody else is running, and a prompt that turns real customer language into a dozen distinct concepts.

0:39 Why it matters

Why does this matter? Because when the platform stopped letting you choose audiences precisely, creative became the steering wheel. Here's an analogy. Think of fishing. In the old days you chose exactly which pond to fish in, that was interest targeting. Now the platform chooses the ponds. What you control is the bait. Different bait attracts different fish. If you only bring one kind of bait in ten colors, you catch one kind of fish. Bring genuinely different bait, and the platform can find you very different fish, including ones you didn't know were there.

1:20 Same product, different doors

Picture a lunchbox brand. Ad one shows a mother packing school lunches at six in the morning. Ad two shows someone meal-prepping after the gym. Same product. But the system will show those two ads to very different people, because it learns from the content and from who responds. Now picture ten versions of the first ad, with different button colors. To the system, that's basically one idea. You've given it one door to walk through instead of ten.

1:54 Three levels

So let's be precise with words. A concept is a distinct idea about who has what problem. An angle, or hook, is the specific way into that concept. A variant is an execution detail, like format, length or the first frame. Diversity that matters lives at the concept and angle level. Variants are for polishing a winner, not for finding new audiences. If your testing plan is mostly variants, you're optimizing the wrong layer.

2:26 The angle matrix

To find concepts, use an angle matrix. Down the side, your personas, the different people who buy or influence the purchase. Across the top, motivations: save time, save money, status, health, safety, belonging. Then layer in objections, like price or trust, the proof type you'll use, like a demo, a testimonial or an expert, and the format. Fill the most promising cells, and you'll spot combinations your competitors have never tried.

2:57 Simple example: Karachi tea brand (illustrative)

Let's do a simple worked example with the matrix. A tea brand in Karachi sells premium loose-leaf tea. Personas: office workers, new mothers, and gift buyers. Motivations: energy, calm, and impressing guests. That's nine cells. Office workers plus energy gives a hook like, your three o'clock slump, fixed without the sugar crash. New mothers plus calm gives, five quiet minutes, one perfect cup. Gift buyers plus impressing guests gives, the tea your guests will ask about. Three very different ads for the same product, each opening a different door. Fill the grid, then pick the most promising cells.

3:40 Example: Abu Dhabi fitness app (illustrative)

Here's an illustrative example. A women's fitness app in Abu Dhabi launched with three concepts. After building the matrix, they ran ten. Twenty-minute workouts between meetings. Postnatal-safe plans reviewed by a physiotherapist. Training around iftar during Ramadan. Working out at home with no mirrors and no audience. Native Arabic versions written by a person, not just machine translated. The winners were the Ramadan and privacy concepts. Interest targeting would never have found those audiences.

4:12 Anatomy of a short video

Now, the anatomy of a short video that performs. In the first one or two seconds, interrupt the scroll and show relevance: the problem or the result. From two to about eight seconds, make a specific, believable promise. In the middle, demonstrate or prove it. At the end, a clear next step. Make it readable with the sound off, burn in captions, and keep key text out of the areas where the app's buttons sit. And on short-form platforms, native phone-shot styles usually beat polished TV edits.

4:50 LLM as creative partner

You can use an LLM as a brainstorming partner here, and the lesson has a prompt you can copy. The secret is the input. Paste in real customer language: reviews, support tickets, sales call notes. Ask for twelve concepts in a table, each differing in persona or motivation, with a hook of twelve words or fewer, the proof type, and why it would reach a different audience. Tell it not to invent statistics or testimonials, and to flag anything that needs legal review. Then a human picks and refines.

5:29 Two things to watch

Two things to watch. First, patience. Don't kill a concept after one day; give each a fair amount of spend before judging. Second, compliance. Health claims, financial promises and before and after images are restricted on most platforms, and regulators like the ASA in the UK and the FTC in the US enforce advertising rules. Real testimonials need real, consenting customers and typical, honest results.

5:57 Mistakes + try this now

Here are the common creative mistakes. Calling ten color variants diversity. Writing every hook about the product instead of the customer's problem. Killing concepts after a day or two of spend. Translating Arabic or Urdu ads word for word instead of writing them natively. And making claims you can't prove. Try this now: take your current live ads and label each one with its persona and motivation. If most of them share the same pair, you've found your gap. Write three new hooks for pairs you're not covering.

6:35 Watch me do it: from quotes to concepts (illustrative)

Watch me do it. Let's turn raw customer language into concepts for an illustrative Manchester dog-food subscription. I paste forty review snippets and twenty support tickets into a spreadsheet. First, I tag each quote with a motivation. Words like finally eats without fuss go to picky eaters. Delivered before I run out goes to convenience. Vet recommended the ingredients goes to health. Cheaper than the pet shop goes to price. Second, personas: first-time puppy owners, owners of older dogs, and busy professionals. Third, I build a three by four grid and count quotes per cell. The biggest cluster is older dogs plus health, which our ads have never addressed. Fourth, I run the LLM brief prompt with these quotes and ask for twelve concepts. I keep five, drop four that are generic, and rewrite three that made claims we can't prove, like adds years to your dog's life. Fifth, hooks: for older dogs plus health, joint-friendly food your senior dog will actually finish. That concept came from real words, not a brainstorm, and it's the one the platform will match to an audience we weren't reaching.

7:56 Recap and next step

Recap. Under broad delivery, creative is targeting. Build diversity at the concept and angle level. Use the matrix to find new ideas and an LLM to brainstorm from real customer words. Measure each concept's cost per result, hook rate, and whether it reaches new customers. Your next step: build a matrix with at least twelve cells for your product, pick six concepts that differ in persona or motivation, and write a twelve-word hook for each.