Because you’re doing it wrong. Okay, so the title was slightly misleading but if it jolts you to reality that’s good enough for me. I want to quickly cover a small but super important aspect of Facebook ads that a lot of people gloss over in the short run that ends up costing them so much money down the road – Adset Optimization.
Now, for the simplicity of the article, I will be talking about ads only related to the Website Conversion objective. I”ll also try to keep this much shorter than my last wall of text on custom conversions (1500 words..ouch), but knowing my penchant for going in depth, no promises!
Adset Optimization 101
Introduction To Facebook Ads Optimization:
I’ve heard it all before. You launch a new adset advertising your product or service, set it on a test budget of $5-$20 and then one of these happen
- You’re not getting conversions
- Even if you are, they are quite inconsistent
- You get a lot of conversions
In the case of the first 2 situations, you have no idea if you should keep testing or kill the adsets. Conversely, for the last situation, you aren’t sure how to scale it! Jeez, what a bother right?!
The number one thing that will kill your business is not knowing your numbers. If you do not have a clear idea of the kind of ROI your advertising campaigns will generate, you might as well pack up.
Okay, so let’s start from the top!
Defining Adset Optimization:
Understanding the Facebook advertising algorithm and leveraging that in your favor has the potential to unleash your business to the next level. There is no other advertising platform in the market right now that can match the targeting and optimization capabilities that Facebook offers. It is probably the most powerful advertising tool available to a media buyer.
You can absolutely use Facebook ads for your local business to DOMINATE your zip code or town or scale your products and/or services to millions of highly targeted prospects.
The part where businesses slip up is in the technicalities of setting up campaigns and systems that allow for optimization to take place. A lot of businesses just don’t know how to leverage it and it’s quite disappointing knowing the amount of money they are leaving out.
Facebook’s advertising algorithm is their best-kept secret. No one really knows how it works, to the extent that different Facebook Ads Reps might give different advice. However, we do know this, the algorithm is primarily machine learning (think artificial intelligence) and it learns based on the amount and quality of data it receives. Adsets need data to optimize. Remember this.
Now before I delve deeper into this, I want to highlight some key points:
- Optimization happens at the adset level
- Duplicating an existing adset won’t transfer the data.
- Any and all adsets start approximately from 0 Optimization
- There is no documented evidence from Facebook stating that the pixel will auto optimize any adset
The Magic Window (No, Not The Conversion Window):
This is one of the worst-kept secrets in Facebook marketing. Facebook has extensively covered it, and yet, a lot of marketers don’t have the slightest clue. Your Adset needs a minimum of 50 conversion events to register on the adset level within a week to have a shot at optimizing correctly
Ideally, you should be getting 50-100 conversions per week to really leverage the optimization capabilities of the Facebook algorithm.
So How Do I Optimize:
Since we know that the basic requirement to getting a well-optimized ad is 50 conversions a week, you will want to make sure that your ad gets at least 7-14 conversions a day. This is the minimum.
Next, you will want to calculate the cost per conversion of each pixel optimization event and set the budget of your ad at 7 to 14 times the cost per conversion. For example, if you want purchases, and your historical cost per purchase is $10, your starting budget must be $70/day at the very least to even have a shot at optimizing.
TIP: If you do not know your historical cost per conversion, or are not able to budget for it, try to optimize one step higher in the funnel. For example, the cost per Add To Basket will always be exponentially cheaper than the cost per Purchase. Similarly, cost per view content is cheaper than an Add To Basket. While, I personally don’t use this method as it takes longer to optimize for the final event I want, It is definitely a viable way to optimize an adset for a higher funnel event at lower budgets to get trickle down data for the final event.
Advanced Tip: Create Custom Audiences and Lookalike Audiences off the higher funnel event audience and retarget them with ads optimized for the final event. As these audiences are warmer, they will almost always convert better than cold audiences.
Facebook Ads Conversion Window ( 7 Day vs 1 Day):
This is actually quite an interesting feature for optimization and can be a standalone topic itself. I’ll try and explain it in the simplest terms for the sake of the article.
As we have already covered, the algorithm learns and optimizes only when a conversion event is attributed to an adset. Conversion windows are simply that window during which any conversion event triggered from an adset will get attributed to that adset.
For example, If you set a 1-day conversion window and a prospect clicks through the ad but converts after 24 hours from viewing/clicking the ad, that conversion will get attributed to that ad and adset however for optimization purposes it will not be counted thus making it losing out on a chance to optimize.
Conversion windows allow Facebook to learn from the data generated during the selected conversion window and optimize accordingly.
On the surface, it seems that selecting a 7-day conversion window would be the obvious choice, however, due to causal vs coincidental conversions this should always be tested (causal vs coincidental conversions will be covered in another topic)
How To Not Suck At Adset Optimization
The simplicity in this section will amaze you. Know your numbers and feed data to your adsets.
For example, if you choose to optimize for the purchase event and your cost per purchase is $10, setting your adset at anything less than $70/day is a losing battle. You can’t expect the adset to collect enough data at $5 to give you the required 7 conversions a day when each conversion itself costs $10.
Facebook wants you to spend money. SHOCKER! But spending money is a good thing since you are buying data. Data=Adset Optimization. Being successful with Facebook Ads is all about buying data.
A general guideline is to have a test period to find out what your stable cost per conversion is. I would advise running an adset at a minimum of $10/day.
Once you figure out a general baseline for the cost per conversion, duplicate that ad at a budget that is at least 7x the cost of that conversion plus an additional 20%-30% buffer room. Alternatively, you can increase the budget on the original test adset itself.
If your targeting, copy and creative are on point then you should very well be getting the minimum number of conversions required to get your adset properly optimized. Best practices are to allow adsets to run for 48 hours before expecting decent results.
Closing Thoughts Regarding Optimization
Optimization of an adset goes hand in hand with testing and scaling. However, trying to cover all of that is beyond the scope of this article. I will cover both testing and scaling strategies in depth on their own in separate articles.
Optimization plays a crucial role if you actually want a positive return out of Facebook. It can possibly lower your cost per conversion and click. There really isn’t any trick to this. Split test different creatives and strategies till you find a winning combination. Once you find a winning adset just add sufficient budget increases to it so that it keeps winning.
Don’t be afraid to spend money on Facebook Ads. But it has to be smart money. You can’t increase the budget on a losing campaign and expect it to work. That is why having an adequate testing budget is of paramount importance, something I repeatedly drill into my clients till they say it with me. When you are testing you are usually not making any money. It’s more common to lose money on the front end testing all the angles. But once you nail a combination you need to make sure it gets that DATA!
Oh, and before you forget – Test Everything.