All About the learning process of Facebook and Google Ads7 min read
Did you ever encounter Google or Facebook saying it’s “Learning” whenever you have created/modified a campaign/Adset/Ad in your Google or Facebook Ads account? And probably this question might have erupted: “How to avoid this?”
Trust us, we have got you covered about this. After an encompassing research and pondering across support docs, we’ve reinforced a complete understanding of what learning period is, why does it come into the picture, how does it occur? and how can you deal with the negative entailments that are tagged along with it?
So, fasten your seatbelts, we are going to answer all your queries and save your time on your mission to navigate through the pages of information available.
The learning period – what the Definition signifies?
Definition by Google: “Whenever you make any sort of changes to your bid strategy, Google Ads takes time to gather performance data needed for optimizing your bids.”
Definition by Facebook: “Whenever you create a new set of Ads or even make a significant change to an existing one, our systems get hold of people to whom the Ad should be shown. This learning is not something associated with the functioning of our system, but we are showing the status to simply inform you when performance is still stabilizing.”
Briefly: The learning period is the time taken by the platform’s algorithm to ascertain from the recent, significant changes.
In the case of Google Ads: A learning period will be invoked only for an automated and smart bidding strategy. The bid strategies usually include Target CPA, Target ROAS, Maximize Conversions, and Enhanced CPC (eCPC). Usually, the learning period will be found in the status column at your campaign level.
In the case of Facebook Ads: Over here budget and settings are formulated at the ad set level and you will get to know your learning status in the delivery column of your ad set.
What types of changes bring out the learning period?
On Google Ads:
- Altering the settings of bid-strategy
- Enforcing a new smart bidding plan
- Encompassing large changes in the budget or bid
- Making changes in the composition of the campaign
- Modification to conversion actions: Creating a New Action or updating an existing action
Probably, modification to keywords, ad groups, or ads will not start the learning period. However introducing bulk alterations to several of these components, you may incur the learning period to your campaign.
- Huge changes in your budget
- Alterations to your audience targeting
- Enabling/Pausing ad set/campaign after 7 days of present status
- Changes like Optimization event, conversion window to your adset settings
- Changes made to Ads (Creative & Copy)
What are the significant changes observed during the learning period?
Both in the case of Facebook and Google Ads, you should probably expect delivery and efficacy to go down throughout the learning period. Yes, this states that campaign’s daily spend decrease while the CPA (cost per acquisition) increases and conversion rate decreases.
So, it’s not something ideal. However, you can’t just sit idly and evade optimizing your campaigns in the wake of the learning period. Instead, your primary focus should be on controlling the implications and allowing your campaign to re-learn and refine the changes you’ve made.
So, why does the learning period come into the picture?
Since we are aware, that the ad auction systems on Facebook and Google Ads are dangling upon the platforms’ algorithms which are reinforced from the machine learning technology. Not only Google requires your Quality Score and bid to determine when you turn up for the ad auction, but also Google needs to interpret what delivers the results (conversions) you’re expecting to optimize for. In short, the algorithm needs the time to adapt to the signal and get hold of what is proven to work and drive results.
The Algorithm is probably enduring the new information and learning how efficiently it can drive the results you’ve opted. During this phase, the algorithm learns from each and every delivery it makes. And according to the impressions built, the algorithm cultivates the significant data it needs to make decisions and grasps how it can deliver to the goals you’ve selected effectively.
What’s the duration of the Learning period?
In the case of Google: The learning period spans over a period of 7 days since the last substantial edit of that campaign.
In the case of Facebook: It will last until your ad set compasses 50 optimization events (ATCs, Checkouts, Purchases, leads) within a 7-day period since the last important edit.
Google focuses on a set timeframe for the learning process whereas Facebook’s algorithm necessitates a threshold of data (Count of Events) to re-learn.
The learning period’s after effect and How to Minimize it?
In the case of Google Ads
- Budget (The first and foremost rule): The 20% rule is what you should adhere to. Any budget changes larger than 20% of your present budget should be averted.
- Give Correct Signals: Let your chief aim is to drive conversions, but hold for a bit, if any sort of conversions is not set up by you, you probably can’t expect a savvy bidding strategy to commence any conversions. This is solely because the algorithm is clueless about what it’s supposed to find.
- Careful Conversions Actions: Utmost care should be taken in your changes to conversion actions. If you are planning to make several changes to your conversion actions, then a proper strategy and schedule are what you should keep in mind and follow. Through this method, you will obtain one learning period rather than several within a stipulated time.
- Strategizing your bidding: How about having the best bid strategy for you? As given above, there are only four bid schemes which are associated with this learning period. So, if you are planning to test out smart bidding, we would recommend building up an experiment first. Through this way, you can appraise the performance in a very controlled environment with a meager negative impact.
For Facebook Ads
- Budget (The basic premiere rule): Analogous to Google Ads, you should register the 20% rule when sailing through the budget changes on Facebook. Budget changes larger than 20% of your current budget should be averted. For Example: Say your budget to be $1000, a change like $1300 or $700 should not be made. With the passage of time these 20% changes can be implemented incrementally(Scale from 1000$ to 1100$ to 1200$ to 1300$ instead of direct jump) to handle delivery and hit the budget you are aiming for.
- Acting on Ads one at a time: Again, you should avoid bulk changes to your assets. For example: If you create a new Ad or pause out one ad, then most probably, you won’t find yourself a learning period. This is also true on Adset level, let’s say you forgot to exclude customers from adsets and now want to do them, do one adset at a time instead of invoking learning throughout account and making performance suffer.
- Selection of your optimization events: Facebook necessitates 50 “optimization events’ to break out from the learning stage. If reaching out 50 optimization events( ATCs, Purchases,etc) turns out to be a herculean task, then we would suggest to try out higher funnel optimization event.( View Content) For example: Getting a Transaction is your ad set goal, but somehow you are unable to reach out 50 Purchase Events within a 7-day period. Instead, you can go with landing page view as the optimization event. However, this will guide the algorithm to look out for people in your audience that are liable to click and successfully make it to your landing page. While this may not be the final result that you are watching out, it’s a preeminent signal and will give the algorithm the adequate data it requires to find your audience and reach optimal delivery so that 50 Purchases are recorded by Pixel and it has some learning and you can Target lower funnel events.
- Brush up your ad delivery settings and optimization: Cautiously dive through your optimization and the related conversion window. For example: If you set your conversion window to 1 day, you are commanding the algorithm to assemble 50 optimization events that take place within 1 day of the ad click. If you notice this retarding your learning progress, then you should probably go for a 7-day/ 28 day click window.
- Always brainstorm before you start implementing!
- Provide the algorithm with accurate and substantial data.
- Give the algorithms the precise signals.
- Be patient. Provide enough time for algorithms to learn, it will be fruitful in the long run.
“The Learning Period” shouldn’t Scare you anymore 🙂