| Hey, Facebook made major changes to its ad platform throughout 2023 and 2024, leaving many authors struggling to match the performance of old – even experienced advertisers. It has been even more challenging for those trying Facebook Ads for the first time because creating a campaign right now is like running the gauntlet where one wrong choice on a random setting could totally tank your ad. And the choices which lead you astray are often recommended by Facebook. Yeah. So much for AI simplifying everything. Despite those challenges, many of us have been able to maintain or even improve performance this year. How are we dodging all these landmines? Which settings do we choose to lean into the strengths of all the AI-branded Advantage features which Facebook is constantly touting? How do we account for the system's (considerable) limitations? And what preparations are we making for the next wave of changes? Over several emails, we are going to break down Facebook's Advantage features – which are a mishmash of AI, machine learning, algorithms, and plain ol' data crunching so you have and understanding of: - how it works
- where it impacts your ads
- why you need to disable large swathes of it or else your ads will implode
- what you should do to plan for the even more radical future that Meta is actively building towards.
And no, that doesn't mean mindlessly embracing all this stuff – handing over the wheel completely to Facebook's overconfident robots. It means grabbing back as much control of your ads as possible while also being properly educated about the changes around things you can't control; whether you like them or not, you must account for them! And to help you understand how the system, works, and especially what it prioritizes, we'll first look at a straightforward example of how Facebook's algorithms work and what it values most of all. (Relevant eminder: The internet is ruled by engagement.) And, before I get emailed by any more robotics prefers, just note that I'm using the phrase "Al" when most of the Advantage stuff that Facebook is pushing is no such thing. It's machine learning. It's data crunching. It's algorithms. But everything is branded as "AI" these days – which is kind of funny when AI's brand is in the toilet right now, along with crypto and NFTs, but I digress. a simple example This example might sound too simplistic, but it genuinely gives you a great insight into Facebook's algorithms, its machine learning - the overriding biases baked into the system and how it operates. Let's say you have an ad set with two ads inside it. I often do this for a whole bunch of reasons which aren't relevant right now. What is important is this: Facebook won't serve both ads equally. In fact, the system often picks a winner very quickly indeed, barely serving the alternate. And the way it does this reveals a lot about the strengths and weaknesses of all the Advantage features being increasingly forced on us. (Not hyperbole – you can't switch some of it off anymore, and the interface has been redesigned to funnel you towards leaving all of it switched on.) Let's say my hypothetical ad set has one ad with a square graphic – like a typical Insta image – and the other is rectangular, like a letterbox, more of an old-school Facebook image. I often set my ads up like this; the letterbox shape does well with readers using Facebook on a desktop device and then the square ads slay on mobile because they dominate the viewport of a phone. Unless I have really messed up the graphics, I would expect to see maybe 80% of the serving going to the square ad image – right from the get-go. Maybe even more than that because Facebook's system has some pretty clear attributes. 1. It is lightning quick. Facebook's algorithms will make decisions in a split-second. It doesn't need to ponder something, weigh its options, reflect on the updated data, or even notice that it has updated – it makes decisions on the fly, without prevarication. The downside is that Facebook's system can seem to make decisions a little too quickly – for example, picking a winning ad after a couple of hundred impressions, which arguably isn't a sufficient sample size. A flesh-and-blood statistician might prefer a sample of at least 1000 impressions, for example. 2. It is single-minded. Facebook's algorithms don't feel doubt. Once it decides one ad is performing better than another, it won't revisit that decision unless something substantial changes. The flipside is that, unlike an anxious meatbag, Facebook's system doesn't go back and check if it made the right decision – and won't catch a misstep. 3. It goes all-in. Facebook's system doesn't really hedge its bets. If it thinks one ad is the winner, it will back it completely. This can be beneficial when the ad is shifting books at a decent clip but the downside of being utterly single-minded, making quick decisions, and being without any doubt, is that when an ad goes off the rails the disaster can be spectacular. And you won't know until it is too late. And that happens frequently enough because probably the most relevant aspect of Facebook's algorithms is that they are blindto anything that happens outside of the Metaverse. This is key to understanding why Facebook's new Advantage features often cause authors more harm than good because we generally don't send readers to a website we own and control, but to retailers like Amazon. Facebook has no way of knowing if the people who click on our ads actually complete the sale on Amazon, so the system cannot optimize for conversions. If we were sending readers to our own website, and we had the Meta Pixel correctly installed, and we had gone to the considerable time and trouble to set up a direct sales business on our sites, then we could potentially find success with all these Advantage features, as the system would be able to optimize on the fly, as above. Facebook would see which types of clickers are turning into purchases, and which are not, and then serve to the most responsive segment of our audiences. But when we are running regular ads, sending readers to Amazon, all Facebook can see are the clicks. That's all the system can optimize for. It looks at who is clicking on our ads, and then serves it to more people like that. Whether they are purchasers or not. disaster strikes You might be thinking "big deal," but any experienced advertiser will tell you that some segments of your target audience will indeed click in huge numbers… and not complete the purchase. One common newbie error is to take Facebook's rather heavy-handed advice and leave on Advantage Placements – letting the system decide where your ads are shown – i.e. Facebook or Instagram or Messenger or the Audience Network. And it will recommend all these options in glowing terms, throwing some rather cherry-picked stats your way about how it will reduce your CPC. Problem is, the most valuable Placement by far is the Feed, and those clicks are the most expensive. Whereas clicks in the Audience Network placement are very cheap because they are indeed literally worthless. Those clicks don't convert. But the system can't see that, so it can move all your budget onto worthless Audience Network clicks – leaving you out of pocket with few sales to show for it. Or maybe you'll run into a completely different issue – like some ambiguity in your ad image or text leading Facebook users to not realize it's an ad for an ebook, leading to lots of wasted clicks. Facebook can't see that these guys abandoned their shopping cart in droves, and just keeps serving your ad to more and more people like that. An ad which attracts clicks but fails to generate sales is exceedingly common and that's kryptonite to authors who don't adjust their targeting or turn off the appropriate Advantage settings. Otherwise, you are in danger of falling into a horrible feedback loop which completely wastes your budget. why is facebook like this At this point you might feel like you have stumbled into a Saw sequel, featuring an elaborate torture machine designed to make advertising books as excruciating as possible. And because we tend to focus our advertising time and dollars around major events like a book launch, when something goes wrong, the stress can be considerable. I'm reminded of the scene in Catch-22 when Yossarian is asked why he's always trying to get out of flying any more missions and he says, "because they're all trying to kill me." When it's pointed out to him that the enemy are trying to kill everyone, he rather reasonably points out that this doesn't make him feel any better. What I'm saying is, it isn't personal. Which doesn't help… so let's actually help. Errr… next week, when I'll help you Regain the Advantage by showing you exactly which settings to change, so you can get Facebook's algorithms working for you – instead of against you. Dave P.S. Writing music this week is "Blue" Gene Tyranny with Next Time Might Be Your Time. | 
![[THE LATEST] Inside hell's kitchen 👩🏽🍳](https://lh3.googleusercontent.com/blogger_img_proxy/AEn0k_sjYsDTswHF8FVXtlRNbhfbp8ZFhv1gvU7E7tPuCJFHAEZVZx-ZlNfT5bNVfKrASXCCNQjcw5jiNnFWP20Oja9c9zSAK4WLKGBVBrcsjuXTgR3g49TlZP04bRgGlNQBE0fgtPSoy759bOAbKf-b29Tw_XKpJYFAv8Eo8UajayuZRAUDFD48MpAHkjaGlxW9N4S6mbPzvvpTlXLnK_FgfZetNVs_Uc6yujkm3VQAR-mIoy4iXfTbPrjAbNGRPpFOor_VclTaZPCFOYSdTfhC97M=w72-h72-p-k-no-nu)
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.