We have set the CompanionLabs app to give the best performance for most users right out of the box, but we've also built in ways to make sure you can tune it, so that's it's just perfect for you. As you're getting started with CompanionLabs, here are a few things you may want to check to make sure you're getting the most out of the app.
Be sure to review your Target Performance Threshold, Grading Confidence and Recency Weighting. These are your 3 key inputs to the CompanionLabs algorithm.
Target Performance Threshold
This one is kind-of a big deal. We like science, so when we set your grades, we use that data science to establish a threshold for the low end of B grades. Ad sets above this threshold are eligible for Predictive Budget Recommendations. But sometimes you want to be more aggressive (or conservative) with your grading and recommendations.
Some scenarios where this may be handy:
You're rocking this... even your "average" performance is good performance that you want to scale.
You're getting started, and even though you have A and B ad sets, they are not performing at a level where you'd want to spend more on them.
You're coming out of a time that is not reflective of the expected performance going forward (like when holiday season ends).
Through some change in strategy or tactics, you've just established a new normal for the ad account and your past 30-day performance isn't applicable going forward.
If your target threshold is currently a ROAS of 4.2, for example, but you'd be perfectly happy scaling with a ROAS of 3.3, click on the hyperlinked value in the B row on your Grade Guide to override the Target Performance Threshold. This will kick off a re-grading on your ad sets. You should expect to see more A's and B's and more budget recommendations.
It works the same way in reverse, if you want to pull in the reins a bit.
On the Settings page, you can choose from 3 different levels of Grading Confidence to line up with how you advertise.
Relaxed: This is good for you if you have a smaller spend ($5000 per month or less), fewer ad sets, fewer results (like conversions), or are just getting started. Grades will come faster, as will Predictive Budget Recommendations. Your grades may be more volatile than on other settings, but you can always change it back later if you want to. (Please allow up to 4 hours to re-grade when this setting is changed.)
Standard: This is good for most folks and the default setting you will see. No need to change this if you are happy with the speed of your grading and budget recommendations. Standard is best for those with a monthly spend of $10,000-$50,000 per month.
Strict: The is the most reliable confidence level. Use this if you have a large monthly spend ($100,000+) and you want to favor precision over speed. You need a lot more data (spend, conversions, ad sets, etc.) to get grades and recommendations at this confidence level, but the grading is less volatile.
Do you look more at the last 2 weeks or the last 3 days when it comes to judging an ad set? That's not a rhetorical question... we've added a setting to CompanionLabs so that your preference is now an input to the algorithm. Choose the time period to which you want to provide added weight and CompanionLabs will apply that weighting appropriately.
There is no right or wrong answer to this; it's based on your preference and the nature of the campaigns in your ad account. The default value is 14 days, but set this where you are most comfortable. If you want your grades (and recommendations) to be most impacted by the most recent performance, 3 days is right for you. If you take a longer view of your ad set performance, 14 or 30 is a better choice.
Grading on CPA versus ROAS
By default, the CompanionLabs app grades purchase ad sets based on Return On Ad Spend (ROAS) as opposed to Cost Per Acquisition (CPA). For most ecommerce models, this is a good way to go. It's a better metric for cases where pricing and number of products can vary widely (i.e. one cart/order may look very different from another). ROAS is also beneficial if you're not capturing complete conversion data (like total cart value) from your Facebook pixel.
If you choose to do so, you can use CPA instead of ROAS for grading ad sets with a purchase goal.
CPA may be a better metric if you have few products and/or customers can purchase only one at a time (i.e. most carts/order look very similar). Please note that changing this setting will also force the app to re-grade ad sets, so expect that to take a few hours.
Hold budget recommendations I apply until...
Time is a funny thing. CompanionLabs gets you Predictive Budget Recommendations as early and as often as the data science will allow; sometimes as frequently as every 4 hours. However, if you have a large budget increase recommended, based on ad set performance (Congratulations!), you may not want to apply that near the end of the day, as Facebook will sometimes hurry to spend that money as the day runs out. This is where this setting is important.
Setting this to "on" will take your applied budget recommendations and wait until the time you specify to execute them. We recommend you set this for midnight in your timezone so that the new budget level starts with the new day. If you rely on Predictive Budget Recommendations, take this as a pro tip.
If you have this setting "on" and, for some reason, you still want to apply a budget recommendation immediately, you can still choose to do that on the Recommendation screen by choosing "Apply Now."
When this setting is "off" applied recommendations take effect immediately.
Reset Guided Tours
If you're like me, you probably pitch the instruction manual on a new "toy" before the box. If you just wanted to get down to business, and more of it, you may have dismissed the guided tour of the Dashboard. We get it, but there can be some helpful tips in there as well. If you want to give it another look, simply click the "Rest Tours" button.
Grade Groups improve Grading and Predictive Budget Recommendations on ad sets with similar target geographies, products, mobile OS and more. This one is big too. Big enough to have its own article and video. Check them out now.
Not a CompanionLabs user? Activate free trials and enter payment info at https://app.companionlabs.com/billing