There are two keywords that advertisers use most commonly when discussing Facebook ads. The first is ROAS, which is the return on ad spend investment. The second keyword is SCALE. The scale is defined as increasing your advertising spend to increase your revenue.
The main objective of most advertisers is to get the best ROAS and scale their spending to get the highest profits possible. While the goal above is agreed among all advertisers, the way to scale your ads remains a controversial subject.
The root of the controversy dates back to when Facebook ads started. Because it was a new platform, very little guidance was given to advertisers as to when and how to scale their budgets. The lack of transparency to how the optimization algorithm works sent advertisers in a quest to figure out the perfect formula to increase their campaign budgets without harming performance. An industry was created around the concept of scaling ads and like mushrooms after the rain, “gurus”, 3rd platform companies, and, of course, marketing agencies all swore that they have the secret sauce.
Very quickly myths started appearing everywhere around the subject of scaling ads. Statements like “ Never raise your budget more than 10% a day”, or “ If you raise your budget too quickly, your campaign’s performance will dramatically drop” and many other unsupported arguments arose.
Before we rip some of these myths to pieces, it’s important to examine the following question:
If Facebook’s objective is for you to spend as much money as possible, so they could make more money off of ads, why would they harm your campaign’s performance by increasing it too quickly?
The three most important factors that involve scaling your ads are the learning phase, the time frame optimization and the timeframe to properly evaluate the results.
Ads Learning Phase
To understand when you can increase your campaign’s budget it’s important to understand the ads learning phase.
The learning phase is the period when the optimization algorithm still has a lot to learn about an ad set. During the learning phase, the delivery system is exploring the best way to deliver your ad set – so performance is less stable and cost-per-action (CPA) is usually worse. The learning phase occurs when you create a new ad or ad set or make a significant edit to an existing one,
The learning phase is important because it’s the most important factor in scaling your ads.
If you increase your spending during the learning phase, performance will be volatile. Initially, the campaign performance might be good and you might be picking up a few cheap conversions which might push you to increase your budget aggressively. It might then result in a decrease in performance which might make you think that the sudden budget increase is to blame. It’s important to understand the importance of statistical significance in marketing. If your campaign starts good with a few very low-cost conversions or starts with zero conversions at all, your results are not statistically significant. Scaling your campaigns when you don’t have statistically significant data is not recommended.
Here are the best practices for increasing your budgets and bids.
- Wait to edit your ad set until it’s out of the learning phase.
- Avoid editing your ads as it might reset the learning system.
- Use realistic budgets. If you set a very small or high budget, the algorithm has an inaccurate indicator of the people for whom the delivery system should optimize. Set a budget that is likely to generate enough conversions for learning to be complete. Facebook recommends enough budget to generate 50 conversions, but it’s based on how much data you have from other campaigns. I’ve seen learning complete with as low as 10 conversions in 3 days and as high as 50 conversions in two weeks. So several factors will impact the duration of the learning phase.
The key takeaway from the information about the learning phase is to understand that your budget and bidding increase strategy should take place after this phase is complete. Start with a decent size budget and bid, and once the learning phase is complete, you can start increasing your bids moderately or aggressively depending on your budget.
Don’t scale based on one day.
99% of advertisers who fail to run effective ads on Facebook fail because they optimize their campaigns by same-day performance. Remember, Facebook ads should be viewed more like a marathon, not a sprint. Longer time frames outweigh shorter time frames in terms of importance in optimizations. Successful advertisers understand the value of the learning system and the value of data feedback that improves the optimization in the long run.
When you optimize your campaigns, you should always do it based on at least 7 days of data. That means, that you can optimize your bids and budgets every 2-3 days, but your bids and budget changes should be based on at least 7+ days of data for best results.
Large advertisers evaluate their performance on longer time frames. Some advertisers use 14 days, 30 days, and some even lifetime results if their marketing budget permits that. When looking back at the campaign’s results and evaluating it’s KPI’s, it won’t matter if one day you lost and made money, it would matter how successful the campaign was in total.
How and when to increase bids and budget.
Now that we covered the basics, let’s examine my suggested method for scaling campaigns. The biggest factor that will determine how fast you can scale is your budget.
- If you market by the revenue to ad spend method (that is, you fund today’s ad cost with yesterday’s revenue), it might be more challenging to use this approach. For advertisers who are limited in cash flow, you can start with very low daily campaign budgets that will allow you to run your ads for at least 5-7 days on a very low budget without interfering with bid changes, pausing, and resuming and heavy editing. Your best bet is to create a limited amount of campaigns on a low budget and let the optimization algorithm deliver your ads most effectively. Then, increase/decrease your campaign budget slowly based on the results that you get. Ensure that after the increase, you will be able to allow at least a couple of days for the campaigns to deliver before optimizing again. Don’t increase heavily if you are concerned with the higher ad spend which may cause you to pause your ads as a result. Understand that patience is the key here. If you increase bids and budgets only to lower them back again you would be doing more harm than good. Your goal is to create low budget campaigns and let them run without interference. Evaluate results based on 5-7 days and try to limit your increases to 2-3 days with slow gradual increases that won’t put you at risk of being low on funds after having 1-2 bad days.
- If you have a monthly or a large budget that you would like to run ads with, here is what I recommend. Ideally, your campaign budget should be at least 5X-7X your target conversion cost. If your target cost per purchase is $30, your campaign budget should start at $150. The reason being is that you want your budget to allow freedom for the optimization algorithm to find the best opportunities in the market. If your budget is limited to the same amount of your target bid, your daily budget might be spent early in the day and you might leave opportunities on the table where the algorithm predicted it would be able to get more conversions, but didn’t deliver because spend was limited.
Important note about bid caps (cost controls).
If you have read my post about bid caps, you know that I am a strong supporter. I recommend all advertisers who have 50 conversions in two weeks to only use bid caps. If you are new to Facebook and don’t have conversion data in your ads, you can start with automated bidding (no cost controls) and start with an initial bid of 3X your conversion. If your goal would be to drive conversion for $10, then your campaign budget would be $30.
Note that in some cases your daily budget might be exceeded. When you select a daily budget, you are selecting a daily average budget. If your goal is to spend $10 per day, you might spend a little more or less per day, but your average per seven days would be $10 per day. See the example below.
Ready To Scale
Once we reach our learning phase and performance turbulence is over, we can start to scale our budgets. There are many theories on specific ways to increase your campaign’s budget, however, none of them have supporting evidence.
When your learning phase is completed, you can expect performance to be steady when you do a large budget increase. I’m not talking about increasing a $200 daily budget to $10,000, but you can comfortably double or triple your budgets every few days. It’s simple. If you are concerned about daily results, increase slowly. If you evaluate your data based on 7+ days, feel free to increase more aggressively.
It’s important to understand that even with bid caps, Facebook will aim to deliver your ads and spend your full daily budget if it predicts it can get you conversions at your target cost. Don’t use large budgets as a reverse strategy or if you are not willing to spend that entire amount. The risk is all on you. Bid caps don’t guarantee results and are only used as a method to tell Facebook the maximum we can spend in an auction.
Repeat the scaling process as long as performance is on an uptrend in the last 7 days and if the campaign is profitable in the lifetime. If one of the two conditions don’t apply, don’t scale your campaigns.
If you do scale your budgets but delivery is limited, consider increasing your bids. If your bid is already 35% higher than your target cost and delivery is still limited, consider expanding your audience or running a different campaign.
Yes! There are so many myths about Facebook ads, but in this post, I just want to talk about the relevant ones to this topic. There are many “experts” that claim that they have figured out the most effective ways for scaling campaigns. Here are some of the most common ones.
- Slow scale method. In this method, they use gentle daily campaign budget increases such as a 10% daily budget increase to prevent the campaign from overspending or underperforming. In their case studies, these “experts” show the method they used and the results they achieved (usually to sell you a tool or a service), but they never show you test or control groups, meaning the case study is never statistically significant. If this was the only method they used at the time, how do you know that was the actual cause for the performance? Maybe it was a good week or a good product? Or maybe they are not sharing the results from the other campaigns that they used the same method in but delivered a poor result? The point is that every case study needs to have a test and control group and should show a statistically significant confidence level. I think that Facebook’s objective is for you to scale faster and spend more, the above theory contradicts it.
- Duplicating campaigns. Advertisers love this method. If you have a campaign running that with a $50 daily budget with amazing results, you wouldn’t want to risk losing that amazing performance, right? So instead of editing the campaign, you can simply duplicate it to spend double that amount. Right? Wrong.
Duplicating campaigns will create two learning systems and will work against the optimization algorithm. You would be better off having one campaign with 20 conversions (snowball effect) then 2 campaigns with 10 conversions each. The smaller amount of learning systems, the better and more steady performance will be.
To understand how to scale your campaigns, you should first understand the moving parts and structure of campaigns. The learning phase is the first step you want to reach before scaling your ads as it will provide more steady results that will allow you to work with large budgets. Evaluating data based on longer time frames is vital to scaling and will allow you to make more educated decisions and run more profitable campaigns. There are many different myths and thousands of different scaling methods. If you understand how the optimization algorithm works, you would understand which methods work along with the algorithm and which contradicts it.