Everything I learned about how to run Meta ads in 2024 from a year of managing more than $1 Million in client ad spend.
This content first appeared in the No Best Practices newsletter on 11.12.2023.
I don’t usually offer “hands on keyboard” Meta ads management as a service, but sometimes I make an exception.
At the start of the year I partnered with an emerging footwear brand on their paid media and creative strategy. Working together we were able to more than double revenue YoY while improving contribution margin % from break-even to consistently in the double-digits.
What follows is an anonymized summary of everything I’ve learned so far about how to run Meta ads in 2024. If you find this content valuable and want my eyes on your media spend, book a 30- or 60-minute call with me or reply to this email for more info on a $1,499 Meta Ads tune-up.
There IS A Wrong Way To Structure Your Meta Ads Account
When I started working with this brand their prior agency had around 12 campaigns running. It was a mix of prospecting, retargeting and remarketing. Only one of the campaigns had a daily budget >$150, and the brand’s AOV was >$300.
This level of segmentation was actively harming the brand’s performance. After iOS 14, more data = better performance. If your campaign’s daily budget is significantly lower than your site’s AOV, you are never going to get meaningful data.
Selecting the 5-10 assets with the best performance from across these campaigns and launching a more consolidated account strategy immediately led to a lift in revenue and an improvement in efficiency.
If you have 100 (or even 20) tiny campaigns running, stop. It is actively hurting you.
Is Meta Ads Account Consolidation A Myth?
If nothing is working in a brand’s ad account and they have at least 500 prior buyers, I like to get them started with this setup:
- CBO campaign with two ad sets:
- A 10% LAL of best customers
- An ad set targeting all relevant interests
- Same set of ads in each ad set–usually best performers to date or best organic social posts
- Daily budget around 2-4x the site AOV
If the business starts hitting our CAC:AOV targets I will gradually scale up the budget. As the budget passes ~$2k/day, I’ll launch a 3rd ad set with broad targeting.
I’ll launch a weekly batch of new assets for testing in a separate campaign, then add winners to each ad set in the CBO as we find them. This is the standard testing/scaling approach, although it’s not the only way to do it.
This worked really well for several months and we got the scaling campaign budget close to $10k/day at a profitable business-level aMER. But we ran into a few problems:
- The best ads featuring the best selling styles would dominate the campaign for weeks, building up so much social proof that almost nothing else would get delivery.
- We’d eventually run out of inventory for those styles, but there was a lag in Meta deprioritizing delivery of the ads, even when efficiency dropped.
Basically, when key styles sold out, we’d get stuck in a 7-14 day rut as the algorithm “figured things out”. Sometimes we ran out of stock of all our top styles at once, and sitewide CVR dropped. Because most of our budget was in a single campaign it took time to walk it back to a profitable level.
To avoid this, we needed to diversify spend across campaigns a bit more. So we started to think of different product categories as “funnels” with their own campaigns. We would still see one or two ads rise to the top, but we were getting more winners across a diversity of styles.
Of course, you can go too far with this. You can’t use this segmented strategy to turn a “stinker” SKU into a bestseller. Nor can you use it to turn a “stinker” ad into a winner.
I also tried out cost caps as a solve. You can read about that here.
What Products Should I Shoot For My Meta Ads?
For a fashion brand, featuring the right product in your ads is almost more important than creative direction, hooks or anything else. And that is because fashion brands usually follow an 80/20 distribution: 20% of your styles will drive 80% of your sales.
Shoot the wrong products and you’ll be out of the game before you’ve even started. BTW, this is why I recommend that fashion brands (and all brands) develop a more nimble production calendar that enables you to shoot new content at least monthly.
These types of products tend to perform best in ads:
- “Core” styles you’ve run successfully season after season.
- Trend-driven products (think: tie-dye sweat suits during the 2020 lockdowns).
- Scroll-stopping products–metallics, sequins, neon colors, etc.
If you shot your entire seasonal assortment on a white backdrop, these styles would perform best.
When you’re running a conversion campaign, you’re targeting in-market demand. The pool of in-market demand for trendy items, and for your own items with a deep reputation, is going to be larger. Larger pool of demand = more potential to scale.
Your best selling items (usually #1 & #2 above) typically have higher conversion rates no matter what channel the traffic is coming from. This makes your Meta spend more efficient, leading to lower CPAs.
Meta is an expensive platform for propaganda. It isn’t the place to push new ideas or try to convince people they should do something different…at least if you expect to make a profit on your ads.
Yes, the new fashion idea of the season is strategically important. Yes, you can and should test it out. But don’t neglect the styles that fall into the three buckets outlined above.
If you can’t accept this (or nothing in your assortment fits into the three buckets), Meta probably isn’t the platform for you.
Related–featuring a single product is going to perform better than featuring multiple products, featuring a category (dresses!) or making it unclear which product you’re featuring. “See it, like it, buy it” is the shortest, highest intent path to purchase.
Single product ads are the equivalent of “bottom of funnel” messaging for fashion brands. They appeal to the most qualified and high intent buyers.
Does this mean you shouldn’t make ads that promote an entire category? No! “Middle of Funnel” and “Upper Funnel” messaging becomes more important as you scale spend. Just make sure you test single product-focused ads first.
What Kind Of Shoot(s) Should I Do For My Ad Creative?
TL;DR go look at the Jenni Kayne ads library. They are a nine figure brand built off a strong combination of merchandising, positioning and performance marketing.
If you want to seriously scale Meta–get your daily budgets past $10k of profitable spend–you’ll follow all the advice in this section to a tee. If you’re ok with $1-2k/day in profitable spend, you can be more selective/more “high fashion” with your creative direction.
But if you disregard everything in this newsletter because it offends your delicate aesthetic sensibilities, you’re going to flop. I’ve seen this happen multiple times.
First: you have to actually show the product you’re selling. Look at a few catalogs you get in the mail this holiday season for examples (I know…)
If you’re selling shoes, you should not be running many (really any) ads of a full model from head to toe. If you’re selling dresses, you should not be getting comments like “where are the shoes from?”
The viewer should be able to discern the details of the product–is it a dress or separates? How long is it? How do I get it on my body? This can be challenging when shooting darker colors–invest in good lighting. This is not the time for crazy camera & Photoshop blur effects.
If you shoot the same product three ways–in an art-directed studio or in situ setting, on a white backdrop, and with an iPhone in a UGC setting–the UGC ad will outperform the others 90% of the time.
If you can swing it, you should shoot all three, because each will appeal to a different customer segment. But if you’re budget-constrained, I would drop the highly polished studio shoot and take steps to make the other formats more branded.
The more you want to spend, the more important creative diversity becomes. Shoots in a variety of settings (and from a variety of sources) create the impression that the world is talking about your product.
You are not going to scale spend to $10k/day profitably with the same set of 10 images you shot six months ago to promote your new collection to wholesale accounts.
I shared advice for creating brand-right UGC-style content here.
Make Sure You’re Profitable, Then Check Your Work
When I kick off a media buying project with a brand, I work with them to fill out this contribution margin estimator sheet. That gives us a business-level target for an acceptable CAC:AOV ratio. Then I distill that target to in-platform metrics, but the business-level CAC:AOV is always the north star for decision making.
Recently I added actual contribution margin stats to the infamous “tracker sheet” to see how valid our original CAC:AOV goal was. There were several months where we overshot our goal, but still made money at a healthy contribution margin.
So I decided to dig deeper: why? It turns out that dollar volume of returns has a big impact on realized contribution margin, and this can swing quite a bit from month to month due to the seasonality of the business. Sales from repeat customers have a smaller but still meaningful impact in padding out contribution margin.
Can’t go into much more detail than that, but I would recommend checking your actual contribution margin at the end of each month to see where you might have some room to push it. Of course, this depends on your risk tolerance; return rate and sales from repeat customers are not strictly within your control.
CPA vs The Dreaded ROAS
When I started managing the account my main in-platform KPI to watch was CPA. This was a target based on our CAC:AOV goal and the brand’s historical AOV.
After some time I realized that we often beat our business-level CAC:AOV goal on days when Meta in-platform CPA was higher than our goal. Sometimes much higher.
I analyzed the correlation between in-platform CPA, in-platform ROAS and business-level CAC:AOV. ROAS actually had a much tighter correlation to CAC:AOV than CPA did. The reason? Sometimes our AOV for Meta-attributed transactions was much higher than the AOV we used to set the CPA target.
The reason for the variability? When you have a broad assortment with a broad range of AURs, there are a lot more combinations of orders a customer can make. Our modal order is one unit at our modal AUR. But some days we bring in customers who buy three or four units. Or an ad featuring a higher AUR style is on a heater.
I decided to start using an in-platform ROAS goal instead of CPA, even though I generally hate ROAS.
How To Run Meta Ads In 2024: Other Miscellaneous Findings
The “learning phase” is real, at least for us. When I added new ads to our largest ad set, or had to change the budget by more than 20% for some reason, we saw a dip in business-level performance for about 24 hours.
This is another reason to diversify spend across multiple campaigns: you feel the effect less if your spend is split across several large campaigns and you’re only making changes to one.
“Just trust the algo” didn’t (always) work for us. When we ran ASC campaigns with a true broad audience (no gender exclusions), we’d get a ton of irrelevant impressions served to men. Men who would catcall our ads in the comments. Engagement doesn’t always mean intent.
“Ads that don’t look like ads” tend to have much lower CPMs and CPCs than campaign style ads. In fact, the closer something is to a polished campaign video, the higher CPMs it tends to get. Is that traffic always relevant? No.
But relevant lo-fi ads + the right product can deliver CPAs that will shock you if you’ve never tried it before. It could change the trajectory of your business.
A final thought: a lot of the advice you read online about Meta ads is written for brands selling a single product and (maybe) some upsell items. The more complex your business becomes, and the broader your product catalog becomes, the less that internet dogma applies.
There are a few mistakes that will absolutely stand in your way of success (a lot of them are outlined above). But beyond that, almost anything is worth a test.