> ## Documentation Index > Fetch the complete documentation index at: https://docs.imagine.art/llms.txt > Use this file to discover all available pages before exploring further. # Pro Tips & Best Practices > Habits that lift your output from 'acceptable' to 'performing.' Once you've shipped a few ads, these habits will lift your output from "acceptable" to "performing." ## Strategy * Write down your customer, their problem, and the platform before you generate anything. Every creative decision flows from those three answers. * Pick one job per ad. "Build awareness" and "drive a sale" are different jobs; trying to do both usually does neither. * Match production quality to product price. UGC for impulse buys, Pro Virtual Try On for premium. Mismatched quality kills credibility. ## Creative * Hook in the first second. Not the first three — the first one. If a viewer doesn't feel a reason to watch by frame 30, they're gone. * Show the product within the first 3 seconds, even when the format is story-driven. Brand recognition compounds across impressions. * End with a clear call to action. "Shop now," "Use code," "Tap the link" — pick one and say it explicitly. * Subtitles matter. Most paid social plays muted by default. If the ad doesn't work with sound off, it doesn't work. ## Testing * Generate three variations of every ad. Different hooks, same product. Run them in parallel and learn which hook style fits your audience. * Refresh creative every 7–14 days. Ad fatigue is real — the same ad shown to the same audience loses performance fast. * Keep a winners file. When something performs, save the prompt, the format, the hook, and the avatar. That combination is repeatable. ## Production rhythm * Use URL to Ad for first drafts, the main workspace for refinement. Don't try to perfect inside URL to Ad. * Reuse avatars and products across a campaign. Consistency drives recognition; recognition drives trust; trust drives clicks. * Name your variations clearly. "Ad V3 — Confession hook" beats "Untitled 7." ## Common mistakes to avoid Picking a format because it looks cool, not because it fits the product. Using a luxury setting for a budget product, or vice versa. Setting signals price; a mismatch confuses the buyer. Casting an avatar who looks like the dream customer instead of the actual customer. Shipping one ad and waiting to see what happens. You need three to learn anything; you need ten to learn enough. Treating Ad Studio output as final on the first generation. It's a draft — iterate.