Are you shooting yourself in the foot before you even start walking?
I’m seeing a lot of people making the same mistake on their Facebook/Instagram ads.
The mistake is not understanding how the Facebook/IG machine learning system works.
This was a HUGE problem for me when I first started with digital marketing. Like a helicopter parent, I was analyzing my ads too much and making changes too soon, effectively shooting myself in the foot.
Here are some quick counter-intuitive tips you need to improve your marketing campaigns. Also, the attached photo is an explanation of campaign/adset/ad so we don’t get confused.
1: Under each adset (or targeting option), you should have significant variation in your ads. This means that you have different images, copy, or even offers.
For example if the adset is targeting people that like yoga, you could have different images/ad copy for one offer, but also multiple offers!
I aim for at least 4 ads under 1 adset, but I often have upwards of 16 ads in one adset alone.
2: Make sure you have a proper budget. No, you can’t run ads on $3 per adset and collect significant data.
What constitutes a reasonable budget depends on the # of ads under the adset, the niche, and the offer. If you’re selling a $5 product, a $10 per day budget with 4-8 ads under the adset is okay.
If you’re selling a $1,000 product, well then it’s reasonable to have at least a $500 per day budget, with SEVERAL (like 16) ads under the adset.
This is EXTREMELY important because ad results are often variable in the first few days- Facebook’s machine learning is figuring out who to show what ads to.
3: Do NOT turn off adSETS within 4 days of launching them.
Even at a higher budget it takes time for the ad results to stabilize. On Day I you might profit, on Day II you might make nothing, and then on Day III profit, etc.
You can’t expect consistent results/data until roughly 4 days in.
4. Do NOT turn off individual ads within an adset- yes I’m serious!
Listen, Facebook auto-optimizes ads within an adset. Stop being a micromanager!
After running ads for some time, you’ll notice that some ads within an ad set reached more people and some reached fewer people.
You may even see that one ad “performed better” than another ad, yet Facebook stopped giving that ad reach! This is called the “Breakdown Effect,” which is a topic for another day (Google it for now, or I’ll make another post soon).
It’s okay to analyze what ads are being shown to more people in order to adjust future ad copy, images, and offers but do NOT turn off individual ads.
The way Facebook works is they analyze the entire adset. The adset is like one organism, with each ad being a part of the organism.
If you turn off an individual ad, then Facebook will reset the entire adset.
Seriously, this one is so key. When determining the success of your campaign, assuming your budget optimization is set to “adset level,” only analyze the result of the entire adset & create a new, updated adset version as needed.
In short, Facebook machine learning is complicated and there’s a lot more to understand than just this post… I would encourage you to research it, observe it, understand it.
Most people are micro-managing their ads. They make changes too soon, analyze the wrong data, and shoot themselves in the foot before they can even get running with success.
Good luck! What are your thoughts here?
Thanks – Michael