Start running ads

Before you build anything, decide what this campaign is actually for. Every ad platform optimizes toward the goal you pick, so the goal is the single most important setting — get it wrong and the system spends your budget finding the wrong people.

Awareness gets you reach. Traffic gets you clicks to your site. Sales (conversions) gets you actual orders. For most SMEs the honest answer is sales, but you can only optimize for sales once your pixel can track a purchase (see the pixel lesson).

1. 1Ask plainly what you want: sales, site visits, or brand awareness.
2. 2Match that answer to the platform objective — don’t guess.
3. 3One goal per campaign; don’t mix several into one.

Don’t pick Engagement just because likes look nice — vanity metrics aren’t sales.

Every platform uses the same three layers: Campaign (the goal + budget strategy), Ad set / Ad group (who you target and where), and Ad (the creative people actually see). Understanding which layer controls what saves you hours of confusion later.

1. 1Campaign controls the goal + overall budget — set once.
2. 2Ad set controls audience, placement, sub-budget — this is where you test people.
3. 3Ad is the real creative (image/video/copy) — this is where you test creative.

Don’t spin up too many ad sets on day one — the budget spreads thin and the system can’t learn.

No tracking, no optimization. The pixel (Meta Pixel, TikTok Pixel, Google tag) is the small piece of code that tells the platform what happened after the click — page view, add to cart, purchase. Without it the algorithm is flying blind and can’t find buyers.

1. 1Install the pixel on every page (via GTM or your store’s plugin).
2. 2Set the key events: at minimum ViewContent, AddToCart, Purchase.
3. 3Use the platform’s test tool to confirm events fire before spending.

Don’t spend before confirming the Purchase event works, or you can’t measure sales at all.

For a first run you have three audience choices: broad (let the algorithm find people), interest-based (target by behavior/interest), or lookalike (people similar to your existing customers). On modern platforms, broad plus good creative often beats narrow targeting — but lookalike from real buyers is the strongest if you have the data.

1. 1No customer data yet? Start broad, or 2-3 interests tied to your product.
2. 2Once you have enough purchases, build a lookalike from real buyers.
3. 3Don’t narrow so far the audience is tiny — the system can’t deliver.

Don’t stack interests until the audience is tiny — “more precise” usually just costs more.

Set a test budget you can afford to lose while learning — not your whole month. The point of the first weeks is data, not profit. A common starting frame for Thai SMEs is a small daily budget held steady for 5-7 days per ad set, so the platform has enough events to exit the learning phase.

1. 1Set a daily budget you can lose without hurting the shop.
2. 2Let each ad set run steady for 5-7 days — don’t poke it daily.
3. 3Judge by cost per result (CPA), not just spend.

Don’t kill an ad because day one had no orders — it’s still learning. Give it time.

You don’t need every metric on day one — just four. CTR (are people clicking?), CPC (what does a click cost?), CPM (what does 1,000 views cost?), and your cost per result. Read them together: a low CTR usually means the creative isn’t landing; a high CPC with good CTR usually means the audience is too competitive.

1. 1Low CTR → fix the creative and hook first.
2. 2High CPC but good CTR → change or widen the audience.
3. 3Write down your baselines so you know if next time is better or worse.

Don’t judge on a single day — read the 3-5 day trend; daily swings are normal.