The Google Ads Learning Phase Explained: What Actually Triggers It, What Doesn't & Why Agencies Use It as an Excuse

The Google Ads learning phase is real, but it's also the most overused excuse in the industry, and it's time someone said it out loud. 

In this episode of the Google Ads Unleashed Podcast, host Jeremy demystifies the learning phase from the ground up, covering exactly what triggers it, what doesn't, how long it actually lasts, and the five destructive advertiser behaviours.

Learn why Google doesn't start from zero when you launch a campaign, why volatility is normal but failure isn't, and how to tell the difference between expected short-term variance and a genuine strategy problem your agency is hiding behind a two-week deflection.

Get your free 30 minute strategy session with Jeremy here: https://www.younganddigital.marketing/

Scale your store with 1:1 coaching: https://www.younganddigital.marketing/1-2-1-coaching

Transcript

Welcome back to Google Ads Unleash, guys. Hope everyone is doing fabulously this Monday. Is the learning phase actually b*******? I have actually asked myself this a couple of times. because it's one of the things first of all when you actually uh search for Google ads it's one of one of the first answers and one of the first things that come with answer the public UK um it's also one of the top searches that a lot of people search about all the time and I've realized I've never done an episode on it so I want to talk about what it is what triggers it what it isn't what it's a sign off because it's being talked about quite a bit and I just want to demystify the whole thing a little bit so let's dive right in. So what is the learning phase? So when you actually launch a campaign or make a significant change, smart bidding needs time to gather data and calibrate. We all know that, right? So for instance, uh um TCPA needs at least 30 conversions to and that's normal, right? An algorithm needs data and needs a lot of data to do its thing. Without data, it can't do its thing. And when you actually make changes or even launch a campaign, whatever, there's a period where Google is testing in which products, which keywords, which search terms, whatever it is, what auctions do I need to enter, how much to bid, whatever in order to achieve the target that I want to achieve um and to get user to convert and performance during the learning phase. Typically, this is across meta is the same as in Google the same sometimes volatile CPA spike, there's row swing, spend can eratic, and that is usually completely normal. So, what triggers it? Um it's uh launching a new campaign. So brand new campaigns of course launch it. Changing bidding strategies is a big one. Uh for instance, if you change from let's say max conversions to TCPA or from uh max conversion value to T roass and so on. Uh significant budget changes usually uh my sort of rule of thumb and that is why I've always saved to people don't increase your budgets more than 20 to 30%. So the commonly cited rule is sort of plus - 20%. So the maximum anything above that triggers the learning phase again changing conversion goals or changing conversion actions can trigger it. So if you add a new conversion action and start learning towards that while Google needs to learn the conversion action and and change its bit strategies right major edits to targeting. So for instance uh deleting complete countries or large scale changes for instance to ads do this as well. If you change your ads out completely or all of your assets or Also, what does it is if you pause campaigns for quite a long time and then restarted. So, essentially all of those are big structural changes, not like smaller changes to your ad account. And to be honest, when I'm going to list now, what doesn't trigger it, it kind of makes sense what I'm on about here. So, for instance, what doesn't trigger it are small bidding adjustments such as nudge and target rorowass a little bit like from 400 down to 380 or a little TCPA change that generally doesn't cause a reset. Adding negative keywords doesn't cause a reset. Um, pausing poor performing ads or keywords or products doesn't actually cause a reset. And also adding new ads uh to ad groups. All of those sort of small things are minor. So all of the small things that don't actually impact the algorithm to do a massive change is what um uh does not trigger the learning phase. All of the stuff that you do change and are massive, those are what re-trigger the learning phase. Um what is quite important is that to understand that there's no such thing as like campaigns learn everything now right in theory all campaigns are learning all the time right so um because that's what algorithms do they get new data points things change competitors may creep up and then the algorithm adjusts and it's always learning so the learning phase just marks the period of the highest instability at um the following quite a seismic change within settings or everything else. It usually lasts about seven days according to that's what Google but sometimes it depends uh on on a few other factors. For instance, how big the account is or on how often you made changes in between. But generally speaking, it actually depends on conversion volume. Right? If you have crazy budget and launch an ad and you know that you're going to get a huge conversion volume. Well, the learning phase is going to be exited pretty fast, right? Fail fast, learn fast is generally what Google lives here. So, if you get 50 conversions a week in a campaign, well, the campaign will exit the learning phase much faster than one just getting five, right? Much much much easier. Um, by the way, you may also want to go back to my episode, how many conversions do you really need? That is, uh, basically explaining pretty well how many conversions do you need to achieve periods of stability where you can actually draw upon the advantages of completed learning phases and using automated bid strategies. Right? So go back to that. Um that's actually something that's uh uh quite important to understand and will be able to help you choose uh and make the right decisions. Okay. So learning phase uh typically why is it a bad thing? Well um I'm going to get to that. It's actually not always bad. Uh but typically it's It's because it launches a period of small instability where results may not be as accurate as you want them to. Okay. Um, and what triggers it are sort of common behaviors and patterns of behaviors which I see time and time again, which is the pattern of beginners and the pattern of uh uh people who just don't know what they're doing. Okay. And the first thing is panic changes. Something that I see time and time again. is people are changing too many things because Google ads don't work or whatever and they change crazy number of variables all the bast of time and don't give the algorithm enough chance to learn and to actually do its job. So they launch something performance wobbles as expected they make a panic change to the bid strategy to the budget to the target whatever learn and resets there's more volatility more panic rinse and repeat right I see that time and time again and it gets in my nerves, stop doing it, make a change, wait till it's actually taken effect, see have stabilized performance and then um uh get on from there. And that is because those types of advertisers never see the stabilized performance because they never f****** allow it to happen. Right? So this is something that um is a common thing that I see. Um uh this goes hand inhand with this. I call it the daily fiddler or the the or the uh the ongoing theorists. Um they great activity with competence. I've done a uh podcast episode about this and in fact um one of my esteemed colleagues Bob Meyer from the PPC hub I've uh attended a really really good uh sort of uh um master class with him uh earlier in the year in Amsterdam where um he basically uh called it um I what did he call it? Stop stop being busy and actually manage your Google Ads account. What I mean by that lots of people equate doing stuff like oh here's another negative keyword or here this and here's a calibration. Um they equate that with activity and with progress and it's actually regress right. So what I would do go back to the episode patience why doing less is more sometimes episode. Um this is sort of the algorithmic explanation of why patience pays off in Google ads. Then the next thing is big budget jumps. Now I don't hate big budget jumps if they are justified and if you can take the risk but one rule of thumb is is incre incremental increases, right? 20 to 30% maximum is great. Spaced out over enough time, let's say every 4 days or weekly is amazing for the system to adjust bit by bit and to also get your budget tripled in weeks. Yeah. Um and that um with confidence rather than in days without confidence, right? Um then another sort of very common pattern that I see is bit strategy swapping I call it. So people swap from max conversions to TCPA. Oh, it doesn't work. Then they swap back to max conversions. Then they try target ROS. Then they do this. f****** pick one target that fits your uh strategy and go with it, right? Pick one tactic that fits your strategy and go with it and just ride out the storm, my friend. Um each switch is a full reset of the bit of the learning phase and accounts never accumulate enough data then to do its thing. They all sort of tie in together all the that I'm kind of seeing. But yeah, that's very very important. Then season uh the next point is seasonal panic and pausing. So I see very many advertisers just pausing their ad accounts because um well where I'm on holiday or s*** like that or they uh pause it during quiet periods and then think they can restart them later and just you know go back to normal. Wrong. That re-triggers the learning phase. And usually I've sometimes even seen accounts never return to the previous performance doing this. So Never ever do that. Use seasonality bit adjustments or um data exclusions and also uh try to work towards those events in advance rather than making sudden swings. Okay. And then lastly, restructures, right? People who launch new campaigns all the time or uh throw everything uh down and then try to rebuild every 3 weeks. That just does not work. So that restructures that um and it uh interrupts learning. Okay. Uh so what is sort of a few things that you should be doing? Uh batch your changes. One consolidated update with cause and effect gives you much better idea of what works and what is actually driving or or not working and also minimizes disruptions. Respect the 20% to 30% budget rule. Use seasonality bid adjustments. Set a no touch window after a significant changes. Um and do not judge your performance based snapshots. Judge it based on the period that is important to your business. I've done an episode on that as well. Go back on that. When should you actually uh interrupt learning? So the B my message here wasn't really actually to never touch anything. So of course if tracking's broken, you need um or if spend is running away with zero conversions after a meaningful sample, then you should probably do something about it. You have to um intervene. So the skill is to distinguish expected learning volatility from actual genuine malfunction here. And this also just comes with practice and experience, right? What is actually the problem or is this like um like a performance swing that is expected? Okay, but I also want to dive into something that is quite important here. Now, something that I really f****** hate to be honest is that people blow up the me the learning phase as something crazy that you need to really always take into account and in my opinion um the whole learning phase thing is a little bit of theater and the reason for that is I've seen plenty of campaigns perform brilliantly during the learning phase and plenty of other advertisers have launched campaigns that were profitable from day one and they've never looked back. So if the learning phase were generally a period of blind calibration let's say then this shouldn't really happen as often as it does. And the reason for that is Google doesn't start from your account from zero. Google draws on massive cross account auction data across billions of auctions every single day across years and years and years since Google Ads has been there and your website signals and your feed and your conversion history, etc., etc. So, it's actually not as volatile as people think. So, I believe in my opinion and probably if you're a business owner, you've heard an agency say, "Oh, yeah, the campaigns in the learning phase and still needs to learn. Well, it can happen, right? And to what I said just now, knowing what causes it and knowing if it's a genuine problem and knowing that that's the problem in the moment it presents to be a problem and to recognize that that's a skill, but to just say, "Oh yeah, it's still in learning is the most overused deflection in the industry in my opinion. It's used by agencies that are don't know what they're doing that by time uh by Google reps that try to defer um hard question s or by advertisers who just want to avoid confronting the possibility that maybe their product sucks or that their office sucks or that their strategy is simply wrong. The learning phase has become this f****** magic incantation that means stop asking questions for 2 weeks and just let the the Google do its thing. That is not true. You should really after very short amount of time know if something works or doesn't work and you have to act on that instinct. Okay. So um yeah, sometimes I've also seen uh campaigns in an ad account be in the learning phase but actually perform really stable. So I would probably use all of this more of an in indicator that um or that this is something to consider uh and that the learning phase is real as a mechanism. So algorithms generally do calibrate a new data but it's widely o overweighted as an explanation for performance. um it doesn't explain bad performance. It just explains short-term variance in performance. Okay. Volatility is completely normal. Failure isn't. And the industry's mistake is to use a variance concept as to excuse a direction problem or performance problem. Okay. So, um going forward, what is it? Well, I've very much sort of said that it's uh what it is. I've also mentioned how you can trigger it. not trigger it, what doesn't uh do it, and what does do it. I've also uh gave you sort of pretty uh clear and stark opinion on that. I don't think it's as big of a deal as people make, but it's very very useful to be aware of. And I think that is pretty much it. I think we've landed the plane. And if you want to have a just a chat about PPC, um just hit me up Jeremy Young Google Ads on LinkedIn. Jeremy at younganddigital.marketing. You can also book a one-to-one meeting with me via the website. If you want me to have a look at your Google Ads account and give you an honest opinion on what's going on or what you should do, message me as well. We also recently launched a new offer done with you service. That means we sort of run the ad account for you as if you are a client for a while and then give you a structured handover on how to run it yourself over a period of several months. If that's something that you're interested in, head to the There will be a new section for that going forward as well. Thank you very much for listening. This has been Jeremy Young, your personal Google Ads expert, and I wish you a happy and productive week ahead.