Everything you need to know about Meta Ads Testing: How do you structure your first Meta ads campaign, and how do you know it’s working?
This content was first published in the No Best Practices Newsletter on 4.14.2024.
I’ve spoken to and consulted with a lot of folks who want to launch a brand with Meta ads, or who want to launch Meta as a growth channel for their brand. The top questions I get: how do I structure my first campaign, and how do I know my ads are working?
Here are my answers to those questions.
Scaling Meta Is Not Just About Ads
First, let’s get one thing out of the way: not every brand can run Meta ads successfully (more on that here). Here are the steps you should take before you hand Zuckerberg a single cent…
If you are starting a brand from scratch:
- Set a hard limit on what you’re willing to spend to validate your brand and/or find winning, scalable ad creative.
- Better yet, try to make 50-100 sales in person, and use the feedback you receive to develop your product assortment, brand, offer and ad creative.
If your brand already has a DTC selling history:
- Download Lifetimely and figure out what percent of your new customers make a repeat purchase. If it’s below 20%, figure out why and get it up to 25% before moving forward with ads.
- Figure out what percent of your first purchase AOV you can spend on marketing while remaining contribution margin-profitable on your first order. If this is below 25%, fix your unit economics until you can spend at least 25%.
- Calculate your AOV for the last 6 months. If it’s below $75, try to get it up.
- Figure out what product or category you’ll “hero” in your ads. Your best seller is usually a good choice. Analyze the current path to purchase for this product and determine how you can remove optionality (prevent analysis paralysis).
- If you’re selling a product where the audience is skeptical, lots of education is required, or switching costs are high, consider building out a landing page before you launch your ads.
One more thing: the recommendations here will only work for brands where the consideration period is one to seven days. If you’re selling something expensive or complex, read this instead.
How Much Do I Need To Spend To See It Working?
Meta ads are “working” if they’re driving a lift in sales from new customers and your ad spend as a percent of new customer sales is hitting your target. Click here for more on how to set your target CAC:AOV ratio.
If Meta is your first marketing channel and organic sales are minimal, it will be easy to see your sales go up once you find a few winning ads. You’ll increase the budget in Meta and you’ll see eCom sales numbers increase.
If you have at least $1-2k in daily sales coming in from other channels, you’ll have to spend more to see a noticeable sales increase.
Hypothetically, you cout test Meta ads with as little as $100/day. That might drive one or two conversions. But it’s hard to discern signal from noise at that spend level if your daily new customer sales regularly vary by 5-10 conversions, or more.
How to solve this without breaking the bank? Test creative at a lower daily budget ($100-300/day). When you have ads you think are winners, scale your daily spend enough to see a noticeable lift in daily new customer sales.
Here is an example:
A brand’s current new customer sales are between $3,500-4,500/day. Their AOV is $100, so they get between 35-45 new customer orders per day.
The brand starts testing ads, spending $250/day. This drives an average of 5 incremental new customer orders per day. Because the average variance in new customer orders before Meta was introduced is 10 orders, it’s hard to say if the $250/day is driving a lift in sales.
To prove a lift, you’d want to scale daily spend to $750-1,000/day. This will drive 15-20 incremental orders, which will be a noticeable lift vs the prior average.
There is no perfect formula here. You’ll have to pick a spend target based on the variance in your own business.
How Do I Know If An Ad Is Good?
Can you trust Meta’s in-platform analytics? Kind of. In-platform reporting doesn’t perfectly represent the incremental impact of your ads. But it is the best way to make decisions about which ads to pause or scale.
How much do you need to spend against a single ad before you have a true understanding of its performance? The frustrating answer is “enough budget to drive 50 conversions over 7 days” i.e. get the ad out of learning mode.
But some ads never drive a conversion, so you would be spending infinite money. That isn’t realistic. Meta is powerful because it provides tools to quickly suppress ads with low performance potential.
There are three ways you can structure an ad creative test:
- Put all your ads into a CBO campaign. Meta will automatically spend against the ad(s) with the best performance potential. The others will receive minimal spend.
- Put all your ads in a cost cap campaign and set the maximum bid to 125-150% of your target CPA. Meta will only spend against the ads that have a chance of hitting your CPA.
- Put each ad in its own ad set in an ABO campaign, so each ad in the test receives a minimum guaranteed budget delivery. Turn off under-performers after a few days, or after a certain spend threshold has been reached.
If you’re just starting out with ads I would start with method #1. Method #2 works best for impulse buys or other products with a lot of in-market demand. Method #3 is best if you feel confident in your creative OR you want to run a structured test OR your consideration period is longer.
Ads should pass three tests to scale:
- Can this ad drive one conversion at <3x my target CPA?
- Can this ad drive multiple conversions at <1.25x my target CPA?
- Can this ad continue to drive conversions over a week at <1.25x my target CPA?
The CPA of your first conversion isn’t always indicative of the CPA your ad will drive at scale. But if you’re still way off target after 4-5 conversions, it’s unlikely your CPA will come down enough to scale the ad.
If your AOV is <$100 it will take about $500 total spend per ad to answer all three questions OR about 80-100 clicks/visits to your website. If your ads are long form video with a lower CTR, you might want to spend more.
CBO is good for new brands because it prevents you from holding out hope for ads that won’t hit your KPIs. Your most scalable and successful ads will usually siphon up 80-90% of the CBO budget within 24 hours. If you launch five ads into a CBO and the spend is evenly split between them, they’re probably all duds (or all winners!…but probably not).
Be sure to wait at least two to three days before passing judgment on ad performance, and let your test run for five to seven days before turning anything off or making any final decisions.
You need to spend enough to give Meta meaningful data. A single conversion that comes in after a few hours might be a fluke (lucky or non-incremental). That is why I suggest spending at least $500 against ads that pass tests 1 & 2.
However, if you have a truly exceptional ad on your hands–the kind that changes the trajectory of your business–you’ll know within 24-48 hours of launch.
What Else Should I Know About Meta Ads Testing?
If you’re just starting out, it will take a lot of creative iteration to find one or two winning ads. Build up a bank of at least 20-30 assets to test before you get started.
Static ads can be good for quickly getting a read on which products, positioning and offers resonate best with your market. That said, static ads work best when there is a lot of educated in-market demand for your product. If you’re selling something more complex, or selling to a skeptical market, you’re probably going to need long(er) form video to achieve optimal CPAs.
In the beginning, you want your creative assets to cover a wide spread of products, personas and positioning.
If I told you I was thinking of a random city in the US, how would you identify the city using the fewest questions? You would start by splitting the country in half–is it east or west of the Mississippi River? And then continue to split the remaining territory in half until you found it.
A similar concept holds true for finding winning ads. If you have no information, you want to cover as much ground as possible in the beginning. One or two concepts will rise to the top–you can iterate on and refine those.