Facebook Campaign Budget Optimization is one of the most important changes to digital ads in years. While it promises to greatly improve ad performance, Campaign Budget Optimization (CBO) does have a bit of a learning curve. In this post we give you a concise guide to get started, best practices and advanced tips.
You’ll need to understand CBO well if you plan to run Facebook ads beyond September 2019. That’s when CBO becomes the de facto way to run Facebook ads. Facebook is taking a “Keep Calm, Carry On” approach to the transition, sending soft warning email notices to advertisers with headlines like, “Your Budgets Are Moving to the Campaign Level”.
No biggie. Just a simple overhaul of Facebook ads. Relax, and let the money roll in.
Here’s the tl;dr of how Facebook Campaign Budget Optimization works — you set budget for a Facebook ad campaign, and Facebook optimizes spend across all ad sets in that campaign.
Your ad sets compete with each other. Facebook phases out the weak ad sets and gives more budget to the best ad sets. Facebook takes more control to give you better results.
This has big implications for your ad tactics, configurations, and performance. Read on to see how deep the rabbit hole goes.
Many marketers are used to having tight control over ad sets performance. Moving budget management to the campaign level, Facebook gives you a sophisticated way to improve results. They ensure your best performing ad sets are given more budget until they are exhausted before giving lower performing ad sets more budget. Here’s what it looks like:
This makes sense. Spend more money on what’s working, phase out what’s not. Facebook wants you to focus on the big picture — how many conversions you get. Don’t miss the forest for the trees. Facebook notes:
To understand how campaign budget optimization distributes your budget, pay attention to the total number of optimization events for your campaign and the average cost per optimization event at the campaign level, instead of each ad set. (source: Facebook)
But what if you really want to spend $X promoting a specific ad sets or ad? You should create a dedicated campaign for it. While Facebook shares that “if you set a minimum spend limit, Facebook will aim to spend that amount,” this isn’t guaranteed.
Set Your Campaign Budget and Bid Strategy
This is where where most marketers will need to change their behavior. We recommend keeping your ad sets and ads in draft mode or paused until everything in your campaign is ready to go.
Making any meaningful changes to the ads in your campaign — whether editing copy or including new ads — will reset Facebook’s Learning Phase and cost you more money.
Facebook needs 50 optimization events to complete its campaign learning phase. Once the learning phase is complete, Facebook can optimize performance to deliver more results at a lower cost. Facebook notes:
When we start delivering our ad set, whether at the start of a campaign or after you edit it, we don't have all the data necessary to deliver it as stably as possible. In order to get that data, we have to show ads to different types of people to learn who is most likely to get you optimization events. (source: Facebook)
Watch out though! Facebook’s learning phase restarts every time you make a “significant edit” to your campaign. Most changes you might want to make are “significant”. At best, resetting the learning phase will just delay your ad distribution. At worst, resetting the learning phase will cost you a lot of money.
Facebook considers it “significant” if you make any of these edits after activating a campaign:
For brand marketing objectives like traffic or engagement, the learning phase cost might be low — e.g. 50 events x $0.10 per click = $5 per learning phase. The learning phase for performance marketing objectives like website/app conversion events can be much more expensive — e.g. 50 events x $10 per signup = $500 per learning phase. In this latter case, making a handful of seemingly innocuous edits over the course of a week could unnecessarily tank your performance and effectively cost you thousands of dollars.
Double check everything before kicking off a campaign. Also, if you plan to roll out many ads over an extended period of time, batch them into separate campaigns. E.g. group all ads that are ready today in a campaign that’s separate from all ads that will be ready in a few days.
It’s easy to jump straight into creating ads, but first consider defining your strategy and tactics for achieving your desired key results — this can pay big dividends. Don’t worry if ad strategy vs. tactics can seem unclear: (1) here’s A Pirate’s Take on Strategy vs. Tactics; and (2) here's an example in warfare terms:
-War = Company Strategy at the Ad Account level; for example: acquire more customers
-Battle = Team Strategy at the Campaign level; for example:
-Skirmish = Team Tactics at the Ad Set level; for example
This planning exercise is similar to setting “objectives and key results” (aka, OKRs) for your team like Google has done since 1999 (great guide from Google Ventures Startup Lab).
Focus on the objective that truly ties back to your team’s end goal. For example, if your team primarily wants traffic, don’t optimize for Post Engagement — instead optimize for website Page Views using pixel event optimization.
It’s common for teams to have mixed goals — e.g. post engagement AND on-site purchases. This is fine, but create dedicated campaigns that use tailored tactics.
We recommend you start with the Campaign Bid Strategy, “Lowest Cost.” It’s high performing, simple, and the best option for most advertisers. It also serves as an excellent performance benchmark.
You can always experiment later with other options — “Lowest cost with bid cap” and “Target Cost.” However, you’ll likely see bad performance if you dive head first into these advanced options before you have the prerequisite knowledge and a performance benchmark.
A well executed “Lowest cost” campaign can easily outperform a weak, “advanced” campaign.
This holds true for Ad Scheduling too. Stick with the default setting, “Run ads all the time”, unless you have a compelling reason to switch to “Run ads on a schedule”. Let the Facebook algorithm do its thing.
If you’re ready for advanced campaign bid strategies, be sure to follow Facebook’s best practices.
Facebook is great at directly driving website and app events — critical for anyone who wants to make money from Facebook and Instagram. You can do so by setting marketing objectives like app installs, in app events, website conversions, and catalog sales.
There’s a catch — you have to properly configure event tracking and conversions based on those events. If you don’t get this configuration right, you’re in for a rough time...headaches, wasted time, and wasted budget. Even smart teams often take a few tries to properly configure conversion events.
Just because your development team tells you things are ready, be sure to check yourself in Ads Manager.
Confirm that the frequency and timing of events reported for a given event matches what you normally see in Google Analytics or other tracking systems you use. For example, if an email newsletter signup pixel event has been fired 100 times in the past 30 days, but you know you've acquired 5,000 newsletter signups in that time, something is amiss.
If you want to optimize for website and app events, Facebook needs each event to fire at least 500 times in the past 30 days in order to run effective ads. Facebook actually recommends reaching 10,000+ occurrences for an event per month for the best results.
If one of your events has fewer than 500 occurrences in the past 30 days, you'll need to collect more data before optimizing for that event. If you don't have enough events yet, don't worry — try optimizing for a more general goal, like Page Views or Traffic until you reach the 500 mark.
Here's Facebook's guide on minimum events needed for conversion optimization.
There's no “one” best audience, but be sure you're in the right ballpark. Your ability to reach a receptive audience can make or break campaign performance.
First come up with 3-5 target audience segments. If you’re looking for some inspiration, check out the PRIZM segments defined by data provider, Claritas — e.g. “Young Digerati” and “Beltway Boomers”. Segmentation will help you craft target audiences.
Next consider your audience behaviors, perspectives, and user journeys to flesh out personas. Personas can cut across segments, and keeping personas in mind will help you refine audience creation and craft more compelling messages.
You want to walk away from this exercise thinking, “I know my audience.” The following excerpt from an interview with the communications team at Sunrun is a concise example of knowing your audience:
In South Carolina we'll focus on words like freedom and control and independence...whereas in California we focus more on the environment and clean energy....I might send an [image of an] eagle and an American flag to South Carolina, and I might send a bouquet of flowers with sunshine behind it in California. - Andy Newbold, Sunrun
Once you know who you want to target, you’ll need to create a Facebook custom audience (uses 1st party data), saved audience (uses Facebook data), or lookalike audience (uses Facebook data, built on top of either a custom or saved audience).
For Engagement and Traffic goals, use a broader audience of 1-10 million people. Here are some examples:
For Website or App Event goals, experiment with targeting both broad audiences (like noted above) and more targeted audiences such as:
Trying many variations of your Facebook and Instagram ads is an easy way to quickly improve performance. Teams from TechCrunch to Travelzoo find that social A/B tests improve ad engagement and conversions by 20% to 100% while also significantly reducing costs. Here are a examples of Facebook and Instagram A/B tests run by Clique Brands:
A Facebook and Instagram A/B test involves taking a piece of content and uploading different variations of text, images or even video cuts to test against a particular audience. We recommend using 6-12 meaningfully different ad variations per ad set. Here’s a short video guide to testing best practices:
You can spend as much as you want, but be sure to at least meet the minimum budget requirements. There are technical minimums and recommended minimums.
Facebook and Instagram technical minimums vary by ad type, bid strategy, and currency — here’s the Facebook minimum budget guide.
Here are recommended minimums per Ad Set:
(a) $50 x total campaign days (e.g. $50 x 5 days = $250), or
(b) the event value x 100 (e.g. $2 per newsletter signup x 100 = $200)
Why the event value x 100? It takes Facebook about 50 events to learn how to optimize for that event. Doubling that to 100 ensures you have a decent baseline of what strong performance can look like.
Multiply the minimums above by the number of concurrently running ad sets and total days your ads will run. For example, if you want to run 10 ad sets in 1 campaign for 5 days, and you’re optimizing for low cost events, set a minimum daily budget of $200 or lifetime budget of $1,000.
Each Ad Set targets one audience, but within a campaign you can target many audiences. It’s generally fine if you target the same or similar audiences across many ad sets if the audience sizes, budget, and target event volume are relatively balanced. Here are two examples to illustrate this concept of balance and performance risk:
If your budget is large relative to your audience size and goal you may compete against yourself, increasing costs. As an extreme example, you may see low performance if you have 2 concurrently running Facebook ad sets or campaigns where:
On the opposite end, here’s an example of where you are unlikely to see audience overlap issues — you have 2 concurrently running Facebook ad sets or campaigns where:
Keep in mind that even if your audiences overlap there may be effectively no performance risk. As illustrated above, performance risk depends on the relative balance between audience size, budget size, and ad goal. If you structure balanced campaigns, you can, as they say when leaving Brooklyn — “Fuhgeddaboudit”
If you're concerned about audience overlap, check out Facebook’s Audience Overlap tool (example below).
You'll typically improve ad event conversion rates and reduce costs by letting your campaign run longer. 5-10 days is a good run time for budgets of high hundreds to thousands of dollars. 2-3 days can work well for smaller budgets.
We don’t recommend running ads for less than 24 hours — Facebook usually takes at least 12 hours to simply start ramping up distribution.
If you pair a large budget with a short run time, spend will be less efficient because you didn't give Facebook much time to optimize bid time or targeting.
Audience size is important to factor into run time too. The recommended times above will work well for audiences that are at least 100,000 people, but for small audience sizes you may need to double the run time. Keep in mind that not only do all the people you want to target need to not only visit Facebook or Instagram, but also there needs to be sufficient ad inventory for your ads to be seen by those people.
There are a few recurring themes in this post: plan well, configure ads properly, know your audience, and experiment.
If you follow these recommendations, you'll be well ahead of most advertisers. Cheers!
This guide was written by Patrick Costello, Naytev's Chief Strategy Officer. If you're interested in using Naytev's advanced social media management and optimization, click here to sign up for a free 14 day trial.