How to run a Facebook ad that actually sells your product

Every Facebook ad has three layers. Almost all beginner confusion comes from not knowing which layer a setting lives in.

• 1

  Open adsmanager.facebook.com and click the green Create button.

• 2

  Choose objective Sales. Not Traffic, not Engagement, not Awareness.

• 3

  Name it something you will recognise later, e.g. Course — Sales — Jul.

• 4

  Leave A/B test off and campaign budget optimisation off for your first campaign. Fewer moving parts.

Do not over-target

The instinct is to stack ten interests to “find the right people”. Resist it. Meta’s targeting is far better at finding buyers from your Purchase data than you are at guessing interests — but only if you leave it enough room to search. A broad audience with a good Purchase signal beats a narrow audience almost every time.

Set the budget you can lose

Your first campaign is tuition, not profit. Pick a daily number that would not hurt if it returned nothing for four days, because that is a completely normal outcome while Meta is learning. ৳400/day for 5 days is ৳2,000 — a fair price for finding out whether your product sells to strangers.

The creative decides most of your result. A great audience cannot save a boring ad.

Make three ads, not one

You cannot predict which creative works — nobody can. Put 3 different ads in the same ad set and let Meta spend on the winner. Vary the angle, not just the colours: one showing the result the buyer gets, one showing you talking to camera, one showing the product itself.

Hit Publish. Review usually takes anywhere from minutes to a day. If it gets rejected, read the actual reason before rewriting everything — it is often something small, and you can request another review.

Ignore likes, reach and impressions. Four numbers actually matter:

8. Scale, fix, or kill

After 3–4 full days, the diagnosis is usually one of these:

Selecting the pixel in your ad set is the bare minimum. Here is what it can actually do for you once it has data.

One pixel, many products — this one matters

Your pixel fires on every product and course you sell, not just the one you are advertising. So if you optimise for the standard Purchase event, Facebook counts a sale of any product as a win — including ones your ad had nothing to do with. It will happily learn the wrong thing.

The fix is a Custom Conversion, which is just Purchase filtered to one product URL:

• 1

  Events Manager → Custom conversions → Create.

• 2

  Data source: your pixel. Event: Purchase.

• 3

  Rule: URL contains → your product slug, e.g. your-product-name.

• 4

  In your ad set, choose that Custom Conversion instead of plain Purchase.

Stop paying to reach people who already bought

Build a Custom Audience of Purchase in the last 180 days, then add it under Exclude in your ad set. Otherwise you keep spending on customers who already own the thing. Easiest money you will save all week.

The two audiences to build next

Ads Manager → Audiences → Create Custom Audience → Website:

• Retargeting — people who fired ViewContent in the last 30 days but not Purchase. Your warmest possible buyers: they already wanted it and stopped. Usually your cheapest sales.
• Lookalike — once you have roughly 50+ purchases, build a 1% Lookalike from your Purchase audience. This is where a Pixel really starts paying you back.

Which event is for what

Check on it mid-campaign

If a running campaign suddenly stops converting, open Events Manager → your pixel → Diagnostics before you touch the ad. It flags problems like events that stopped arriving. A pixel that quietly broke looks exactly like an ad that stopped working — and people rewrite perfectly good ads because of it.

10. Costly beginner mistakes