Optimize Facebook Ads - By NOT Over-Optimizing
While we make our living by optimizing and automating stuff (and by stuff I’m mainly talking about Facebook and Instagram ad efforts), there’s still a few traps even the smoothest of operators tend to get caught up in.
1: Learning phase vs guessing 💁🏽♂️
The learning phase is probably one of the most powerful concepts (and optimizations) to really respect and utilize when you’re doing Facebook and Instagram marketing, yet, a lot of marketers doesn’t know what it is, “has heard about it” or kind of knows but doesn't care a whole lot.
When you create a new ad set, ad or edit an existing one, Facebook needs to collect a certain amount of data from your audiences through delivery in order to learn exactly what people and which ads to run and in what manner in order to maximize results. The arbitrary threshold for when a learning phase is completed occurs when you’ve completed about 50 conversions in 7 days (or faster). Once the learning phase is completed, Facebooks own algorithmic optimizations get to work, and they are powerful.
The most effective way to ruin the learning phase (and thus keep your targeting pretty random all the time) is to do continuous updates to your ad set settings, like adding a custom audience, changing a few placements or making too big budget changes. When you never allow your ad sets to pass the learning phase and start delivering the real results you’ll always be in the learning phase, guessing, hoping for more low hanging fruits to fall (yes, they tend to fall first and then people panic because the CPA is rising once they’re gone).
Keep a close eye on your learning phase, allow it to learn and then you’ll be able to harvest the full potential of Facebooks optimizations. Not every, but many changes you do along the way will reset the learning phase and you’ll be back guessing - BUT, once an ad set has passed the learning phase and resets due to small or large changes, the learning will most likely go much faster and you’ll be back in the sweet, sweet lane of stability and control. As a spy inside of Facebooks HQ once said, “when people are doing the least amount of changes in their accounts, results tend to be more stable and costs tend to go down”. Don’t quote me on that one, but it’s true.
2: Hacking conversion hours ⏳
I’ve heard all kinds of strategies regarding time limitations. Everything from “I’m only showing my ads every Friday and Saturday since I read that people tend to buy more those days” to “I’ve cracked the system… the majority of purchases happens between 3pm until 9pm, that’s the only time I’m showing my ads now. Saving moneeeeey”.
Those theories (and more of that sort) could work great if your ambitions are to spend a few thousand dollars a month MAXIMUM, but if you’re looking to establish a trustworthy, scaleable and serious account in the eyes of Facebooks algorithms, here’s two problems with utilizing those kinds of strategies:
I: Facebook promotes continuity and predictable behaviour, meaning, an account that always stops, starts, stops and starts are going to have a lower amount of “trust” from the platform. This results in a lower competition rank and lower estimation scores which results in more expensive delivery since accounts who’ve adopted the more common “always on” mindset will be preferred over your account. This especially hurts once you go in to high competition seasons such as Black Week or Christmas.
II: Even though most purchases occur a certain time everybody’s on their own journey. How many times have you seen a brand for the first time and immediately purchases from their website? Not that often, huh?
If you’re a brand that’s globally well known, congratulations, you can then enjoy a bit more flexibility in how, when and what you show off since people already know and trust you (hopefully), but in most cases that’s not the reality. The power of being present with your potential customers throughout the day, whether they’ve just woken up and checks their Instagram or just stepped out from the gym for a quick Facebook scroll before getting in the car, is very important. It can be hard to know exactly when a potential customer is ready to buy but you can be certain that they won’t buy from you unless they’ve seen you, connected with you or trust you in any way.
3: Bid and budget management ♎️
Aside from being incredibly time consuming, changing budgets depending on “what works” is usually a pretty weak tactic. Here’s why:
I: There are tools that does it way better than humans possibly could. Machine learning and algorithms almost always beat the human gut feeling when it comes to making decisions based on data, especially in larger sets. Except being automated and data driven, these kinds of tools should also consider the learning phase in your workflow, doing exactly enough to optimize budgeting and bidding without resetting the learning phase all the time. Facebook got their own Campaign Budget Optimization, we’ve got a pretty sophisticated tool too called Bid and Budget Automation and there’s a few more out there. I really recommend you looking at automating these kinds of tasks, you’ll save loads of time while probably increasing results as a side effect of you doing better things with your time.
II: Continuity, again. As slightly mentioned earlier in this post, continuity is really important for creating optimal delivery prerequisites. When advertisers focus on changing bids and budgets to maximize results they tend to be very short-term in their mindset. They usually look at the past few days or perhaps weeks in order to change budgets accordingly to “what works” right now.
The problem with focusing very short-term is that you pretty much lose all long tail business and strategy, the focus becomes more about maximizing whatever you can at THIS particular day instead of building a great brand with loyal customers. Also, when you’ve found yourself in the day-to-day optimization trap you’ll probably be surprised to notice that things almost never get better than the day before. You’re very likely to be wasting loads of time while creating a work-loop that won’t allow you to grow your business.
III: Bids should be set based on data, not random hunches. Yes, it’s difficult to know what bid strategies to use and what caps to set, and while that is difficult the “Lowest cost” option usually seems like a really tempting solution. Perhaps you might even feel like that’s the very best option, since it’s called “Lowest cost” and all. We’ve been running lots of tests regarding bids and with the shortest of conclusions - “Lowest cost” is often the most expensive strategy long-term. Try out bids through split testing, not setting random numbers based on hunches and “see how it goes”.
4: The “spend $5 and if it doesn’t work turn it off” strategy ⏸
Well, the only reason I’m even touching on this topic is because people (apparently) like to read clickbait topics with “optimization hacks” that either doesn’t work or does the exact opposite from optimizing anything.
If you’re uncertain why this strategy is shit you should read this post again :). Sure, you can harvest some great looking results while spending a budget that equals an expensive cup of coffee but you won’t be anywhere near any shape or form of significance in your data set (perhaps with the exception of optimizing for landing page views), and after all, isn’t data what marketers are supposed to base their decisions on?
Tip: Create a few hypothesises (set of assumptions) on what audiences, ads and settings you think will do the trick, split test them properly and then actually learn what works once you’ve passed the learning phases and have sufficient data to evaluate. Once you’re done, iterate and start over until you’re satisfied.