Online Advertising guide
Facebook and Instagram Ads: Finding an Audience That Buys
Written by Shawn Shellenbarger · selling online since 1997
Why the creative does the targeting now, how to structure a simple account, and the numbers that tell you to scale or shut it down.
What you will take away
- Nobody on Meta is shopping. Your first three seconds earn the stop.
- One campaign, broad audience, three to five creatives.
- Raise budget in small steps or you reset the learning.
Demand capture versus demand creation
Google catches people who are already looking. Meta creates the interest. Nobody opens Facebook to buy your thing, so the ad has to earn the stop before it can earn the click.
That changes everything about the creative. On Google the keyword does the qualifying. On Meta, your first three seconds do.
The offer decides the outcome
Before you write a line of ad copy, look at the offer: the price, the promise, the proof and the next step. A great ad for a weak offer loses money slowly. A weak ad for a strong offer still gets sales.
Ask three questions. Is the price easy to justify against the problem? Is there proof that someone like the viewer got a result? And is the next step small enough to say yes to on a phone, in a feed, with no context?
If any answer is no, fix the offer first. That change improves every ad you will ever run in the account.
The creative is the targeting now
Audience targeting matters less than it used to. Broad targeting with strong creative usually beats narrow interest stacks, because the algorithm finds the buyer from the signal in your ad.
So spend your effort on the hook: the first line of text and the first frame of video. Show the problem, then the result, then the offer.
A creative script that survives the scroll
You have about one second to interrupt and three to earn the click. This structure works for almost any product.
- Hook: name the exact person and the exact problem in the first frame. 'If your Shopify store gets traffic but no sales...'
- Proof: show the thing working, or the result, before you explain it.
- Mechanism: one sentence on why this is different from what they already tried.
- Offer: the price, the guarantee, and what happens when they tap.
- Close: repeat the promise. Do not add a second idea.
A simple campaign structure
One campaign, one broad audience, three to five creatives, one clear offer. Let it run for a few days without touching it.
Resist the urge to build ten ad sets with ten interests. You split your budget and your data, and Meta needs both concentrated to learn.
Test one variable at a time
Winners are usually obvious within 1,000 to 2,000 impressions, which is why the order of testing matters.
- Concept: the angle or promise. Test this first, because it matters most.
- Hook: the first line or frame of the same concept.
- Format: video versus static, square versus vertical.
- Offer: trial versus discount versus free shipping.
The numbers that tell you to scale or stop
Feeling good about an ad is not a metric. These are.
- Hook rate: three-second views divided by impressions. Under 20% means the opening is weak.
- Click-through rate: under 1% usually means the ad is not compelling.
- Cost per acquisition versus your allowed cost per customer.
- Return on ad spend once you have sales, and frequency above 3 means you need fresh creative.
Why your numbers do not match
Almost every Meta account I audit has a gap between what the platform reports and what the bank account shows. Three causes cover most of it.
- Attribution windows: the default credits a purchase to an ad clicked or viewed in the last seven days, even when the customer found you another way.
- Duplicate events: a pixel and a server event both firing, which doubles the conversions in the report.
- Delayed and missing events: people who opt out of tracking, and sales that land after you have already read the report.
Scaling without wrecking your account
Raise the budget in small steps, 20% every few days, and watch cost per acquisition and frequency. Big jumps reset the learning and can double your cost per result overnight.
When one creative carries the account, keep testing new ones inside the same campaign. Creative fatigue is the most common reason a profitable campaign stops being profitable.
Questions people ask me
How much should I spend to test on Facebook and Instagram?
- Plan on $20 to $30 a day per concept for three to four days. That is usually enough for a hook and a click-through rate to declare themselves. Testing five concepts at once sounds efficient and usually just splits the budget too thin to learn anything.
Do I still need interest targeting?
- Rarely. Broad targeting with strong creative outperforms stacked interests for most offers spending under a few hundred dollars a day, because the algorithm finds buyers from the ad's engagement signal. Interests still help for a genuinely narrow audience, such as people who work in one profession.
How often should I refresh my ad creative?
- When frequency passes about 3, or when cost per acquisition climbs for a week while everything else is unchanged. In practice that is every two to four weeks for most offers. Keep new concepts running inside the same campaign so you never start from zero.
Are Facebook and Instagram ads still worth it?
- Yes, if you sell something that benefits from being shown rather than searched for. Costs have risen and the creative bar is higher, but the platform still reaches a large audience at a low cost per thousand views. It rewards strong offers and punishes lazy ones faster than it used to.
Native ads
Want the full system for native ads?
Search and social are only half of paid traffic. The other half is buying placements on content sites, and the course I recommend for that is the Native Ads Master Class by Tom Bell and David Ford. My review covers all 14 modules, the $997 price and who should skip it.
Read my Native Ads Mastery reviewWant a second set of eyes on your account?
Send me where you are stuck and I will tell you the first two or three things I would change.
Book a Call With MeKeep learning
Link to this page