tl:dr; I believe everyone will run ads like this over the coming years. Facebook Ads will operate no different than TikTok’s organic content algorithm. Figuring this out just saved us a ton of money & some of you could still pivot in time for December.

This supplement brand had an AOV (average order value) of $68 with an average customer acquisition cost of $46 using Facebook Ads. No other channels, all spend on FB & IG. This was in Q3 which is their slowest time of the year. Getting 1.47x ROAS was basically break-even. Founder was fine taking that beating to acquire customers with hopes of subscriptions & future sales in January saving the day. Fortunately, we were able to step in & increase ROAS to 3.3x. Here’s exactly how we did it…

How they were running ads before (the old way):

  • Ad account split into “Top of Funnel”, “Middle of Funnel”, and “Bottom of Funnel” campaigns.
  • Each campaign focused on prospects at varying levels of awareness/interest in the product.
  • MOF & BOF campaigns were just running ads with b.s. fake scarcity “Less than 100 units of X left!” or deep discounts “Get 30% OFF today only!”. Founder did this out of desperation to clear space at 3PL due to high fees. I’d argue it had the opposite effect.
  • TOF campaign had 13 adsets (yes, 13). Each with a different interest targeting. “GNC”, “Fish oil”, “Physical Fitness”, “Bodybuilding”, “Crossfit”, you name it. Even some unrelated like “The Single Mom’s Club”.
  • Budgets were set on the adset level so each one was spending a different amount. Some $80/day, some $300/day. Previous media buyer was raising budgets slowly until diminishing returns & would try to find the sweet spots for each audience.
  • Brand had tested a total of… *drumroll* 9 pieces of ad creative in the 3 months leading up to this time. 5 were videos and 4 were static images.
  • This setup was producing conversions for $46 a piece. Each customer spending $68 on average. Very stressful situation for the brand with break-even days followed by a day in the red.

What we did differently (the new way):

  • Created 1 campaign in the ad account with 2 adsets inside it. Turned everything else off.
  • Adset 1 was titled “New creative” and Adset 2 was titled “Scaling”
  • Both are broad with no targeting or exclusions. (Do this unless you sell something that is clearly exclusive to a gender like female hygiene products.)
  • In under 3 weeks we created 13 UGC-style ads, 6 static image ads, 2 GIF ads, and 1 carousel of images. Besides the carousel, every ad was exported in square & tall formats for both feed and story placements.
  • Roughly 60% of these ads were scripted & filmed with “problem aware but solution unaware” customers in mind. They know they want X outcome, but don’t know how or with which product. So the messaging would strictly speak to that avatar. That was represented in everything from the hook to the call to action.
  • 20% of the ads were created specifically for the “solution aware” customers who we know are likely aware of the top 3 competitors. They potentially even bought from them before. The ads here would compare this brand’s product to the competitors. (Those side by side comparison charts work REALLY well. Showing how others have sugar/filler whereas this one doesn’t, etc.)
  • The last 20% of the ads were created with “lurkers” in mind. We think of these people as “on-the-fence” prospects who click on ads, engage with the page, visit the store, but never buy. We basically try to guess the buyer objections preventing the conversion & address these objections in the ad creative. Example objection: “I heard X supplement always tastes chalky, doesn’t matter which brand sells it.” The ad would then have a girl start by saying “Every X I tried tasted chalky but ABC one doesn’t.” For ideas on buyer objections, look up your competitors on Amazon & read the “Customer Q&As” section. There’s some gold there. Another good ad for this audience are customer testimonials either in a video collage or carousel. Videos with customers speaking work best while photos are 2nd best.
  • One key thing in all of our ads is that we try to leave in elements or crumbs that will increase engagement rates & promote discourse. “Who needs to see this?” or “which one is better?” Basically anything to get people to tag friends, share, comment, etc. The algorithm sees this as a positive signal and drastically lowers CPMs. Meta is basically rewarding you for making their users engaged and spending time on the platform. This is great because even when CPMs rise due to competition, that increase is offset by high engagement rates & therefore quality ranking.
  • Now comes the fun part. All of that creative, regardless of WHO it is addressing, goes into the same adset that has no targeting. First, it all goes into the “New creative” adset. We let it run for 3-5 days and let the algorithm do it’s thing. Every ad creative finds it’s own audience because the algorithm tests it in a similar fashion as TikTok does with new organic uploads. They are shown to random pockets of users to see what produces a conversion (or at least a positive signal/engagement).
  • The best ad creative had a $17 CPA, some others were in the mid $20s, and the rest were either $30-$40s or didn’t get a single conversion. Out of the 22 total ads we made, 6 were under $25 CPA so we chose these as our “winners”. We then grabbed their Post IDs and duplicated them as existing ads into the “Scaling” adset. That is where we scale by increasing the daily budget.
  • In summary, we ended up reaching an average $21.75 CPA after a month of scaling which was just above 3.3X ROAS given the new $72 AOV. (AOV was higher because we stopped doing heavy discounts and simply made the brand look better / more reputable through creative.)
  • For December, we’re going to up the ante & test 31 new ads as we head into this brand’s peak sales period. The goal is to have 4-5 simultaneous winners running geared towards the discovery audience, 1-2 for the solution-aware, and 1-2 for the lurkers (retargeting). If this gets some interest I’ll post an update with the numbers of how that goes.

A few common mistakes you’ll encounter along the way:

  • A big one is ending tests too early. When you have new creative, the algorithm will prioritize attributing budget to certain ads over others. If you’re dealing with lower spend and the algorithm oddly allocates 90% of the spend to 1 ad and didn’t even spend $0.03 on others, you may have to intervene and momentarily pause the ad it allocated the most budget to. This forces it to consider the other ads. This is less of an issue though if you have a proper testing budget for your given product & vertical. We allow 5x the desired CPA to spend in those 3-5 days before making a decision to permanently cut an ad. If you’re on a lean budget you may even do it at 3-4x desired CPA. (For this brand we let $125 spend on an ad total before deciding to cut it since our desired CPA was $25).
  • Another mistake is when you transfer the winners from the creative testing adset to the scaling one, and you turn off all but the 1 best winner. That’s a mistake because you should always be scaling multiple ads at once. They each will convert different segments of buyers. So even if 1 ad is converting at $28 and another at $32, you should leave both on and let the algorithm decide when to deprioritize them.
  • Do not skip out on static image ads. I know UGC ads are all the rave (& they do perform very well) but about 40% of our best winners are static image ads. Comparison charts, bullet points, ingredient lists, things of that nature. You’ve seen them on your feed countless times and it’s because they work. Same thing with longer videos. There’s no magic limit to 15 seconds or 20 seconds. We have winners that are 40 seconds and even 60 seconds long. Just make good ads & the rest will fall into place.
  • When launching new tests, do it in the “New creative” adset first. Most people get too excited & think they can shortcut it by putting it directly in the scaling one. This will waste a lot of ad spend and burn a hole in your pocket. Test with smaller budgets in the creative testing adset, then duplicate existing post IDs into the scaling one. Rinse & repeat, and you’ll be golden.

And that just about sums it up. Obviously there’s eons more to profitable ads like having a high converting landing page & a solid offer geared for customer acquisition. But if you are struggling with high CPMs and ads not performing, the above is probably the quickest way to get you out of that. I wouldn’t focus on anything else before this is resolved. Good luck in the rest of Q4 and if you have any questions I’ll try to pop into Reddit & respond. - Simon

  • skrfs@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    Great write up.

    Are you doing one campaign with ABO budgets for each of the two ad sets?

    Are you doing any advantage + / manual bidding?

  • RTizs@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    Thank you for taking the time, very interesting and useful info.

    I have a question, what if the business is completely offline and you can’t really track the return on the ads? Should you then even consider ads of this nature?

    Imagine you’re working on a small family business that does flooring with a very small ad budget. Their website is just a catalogue without any prices, because the final price heavily depends on the size of the job, location and bunch of other factors. Currently, we try to optimize for the best cost-per-lead, but the outcome of each lead is so random that you can’t really tell if what you’re doing is making any positive difference.

    What would you do in this situation?

    • admajesty@alien.topOPB
      link
      fedilink
      English
      arrow-up
      1
      ·
      10 months ago

      Great question! In this case I would setup a landing page with a survey/form so people can get “a free estimate” calculated based on their answers like square footage, etc. Behind the scenes, what would happen is that anyone who fills in their e-mail & contact info would trigger a “Lead” event. However, if they select specific options in that form, they can get redirected to a page with a “Purchase” FB pixel installed. OR, after completing the lead form they could be prompted with a Calendly page to schedule a virtual 25 minute consultation. After booking that consultation, THEN you would trigger a Purchase event from the pixel.

      Now when you are running ads, you are optimizing for booked appointments/consultations instead of just for leads.

      Obviously there are limitations with offline businesses but there is a lot you can still do. Hope this gave you some ideas.

  • lolaspencerz@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    Thanks so much for the incredible write-up. Feels like you wrote it for our brand :) We’re in a similar space and similar AOV, and we’re JUST getting started with our purchase campaign (we’ve ran a lead campaign for a few months before launching). A few follow-up questions if you don’t mind!

    - It sounds like based on what you shared, you don’t use DCT (dynamic creative testing)? Lots of folks pushing for it, and I’ve noticed we don’t get in the learning phase when we use DCTs, but then we’re forced to create new adsets for every test (which means you can’t have everything in one testing adset). What’s your experience with these?

    - When you say you test each ad for 5x the desired CPA, that’s in total, right? Not per day for 3-5x days?

    - How do you find the post IDs for Instagram ads? I’ve found that FB ads are super easy to grab, but going a lil coo-coo looking for how you do the same with IG ads lol

    Thanks again Simon!! Appreciate you for this!

  • mike3394@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    Awesome post. How are you tracking conversions and remarketing? Is there any reliable way to do this anymore?

    • admajesty@alien.topOPB
      link
      fedilink
      English
      arrow-up
      1
      ·
      10 months ago

      If you setup your pixel properly and make sure events are firing, as well as Conversions API that is native to Meta, you will be golden. All of the after-market stuff for 3rd party tracking is honestly overkill and unnecessary. If you want an extra layer of accuracy you can use UTM parameters and then track them straight in your Shopify analytics without any apps or plugins. Even so, with proper setup our Meta Ads are pretty darn accurate right now across the board. Probably 90% accurate.

      Of course, many clients also sell on Amazon, Walmart, & sometimes retail so there are unattributed sales. Especially when you are sending traffic to Shopify but a person opens their phone and buys from the Amazon app. At that point, we handle attribution by finding the correlation of advertising spend to sales from each channel. It becomes pretty easy to calculate the percentage change and ratio of ad spend to “organic” sales on Amazon and such.

      Oh, and this is assuming you have a decently short buying window. If it’s 1 to 7 days Meta will be pretty accurate. After that is where things get rocky. If your average customer takes 10-15 days to make a buying decision, it will be much more difficult to track. For certain special situations like that, I recommend having a smaller conversion first like a free opt-in of sorts. Then at least you have some measurable KPI to strive towards and track.

  • Equivalent-Bonus-712@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    I own a Digital Marketing Agency which mainly caters to e-commerce brands and I can vouch for this post. I am going to screenshot this because I have never seen anyone on internet sharing the actual gems.

    Every single thing from selecting the creatives, to setting up the campaigns, how to introduce a new creative and lastly how to manage high CPMs.

    Moreover, Supplement ecomm brands are the toughest ones to scale and this guy has clearly figured it out.

    Take a bow sir!

  • ball2312345@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    Awesome post! Thank you!! Question:

    1. is this a CBO or ABO campaign?

    2. do you only have one testing adset or can you have multiple different adsets testing different product collections?

    3. What is your structure for running BFCM campaigns?!

    4. How are you retargeting? Separate campaign for retargeting?

    Thank you so much!!!

  • Thedouche7@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    Interesting post. Why do you think this setup is better than multiple adsets with their own budgets? I’m in a situation similar to your client’s initial setup.

    Don’t you have the issue that out of 3 creatives in the test adset, only 1 is spending? Or you never add 3-4 at once?

    Also, do you think this setup would work for Facebook placements only? Btw, you’re using no adv+, right?

  • 12pac@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    Do you set budgets on the Ad set level so it spreads it out over the adset or at Ad level?

  • Left-Singer-3510@alien.topB
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    If I could give you an award I would but seeing as how Reddit removed it here is one from me 🥇 a lot of people don’t give out any information here I am thankful