NEW Google Ads Experiments Setup

Key Takeaways

google ads campaign experiments have been around for a decade if not longer and have been a really valuable tool for me over the course of my career if you're not familiar with them campaign experiments allow you to take your current campaign also called the control duplicate it and make adjustments to that duplicated campaign called the experiment so that you can test different settings different tactics all sorts of things against the control you then split the traffic you are going to get with a certain percentage between the control and the experiment so you minimize any sort of interference among the campaigns this is a very scientific approach and can be very valuable to test out different theories and see what works best in your account recently google changed the setup flow for how you create campaign experiments so what we're going to do today is run through that new flow to get you back into testing things as quickly as possible i'd like to start off this video by showing you what the original process for campaign experiments in google ads was that way anybody who's run them in the past can at least have a quick refresher on what we used to do and then make it a little bit easier to understand why the process moving forward is so new if you've never set up a campaign experiment before you can skip this first portion and go right to the part where we do the first experiment setting up a campaign experiment in google ads used to be a two-step process you first created a draft campaign made all of your changes to that draft and then applied it to an experiment for this video i'm going to use this search halloween campaign and our placeholder account as the example so we used to go into the campaign itself we would then navigate to the drafts section and then we would create a new draft based on this campaign we'll just give it a name really quick and then click save once the page reloaded you noticed that there is a new campaign listed over here on the bottom left and it is just the draft version of the campaign that we created earlier what we would do now is since we are selected into this campaign we would make all of the changes to keywords ad copy the campaign settings the ad schedule whatever we wanted to do in this campaign experiment all of those changes would be applied to this draft campaign once we made all the changes we would head into the experiments tab and here there would have been a button that said create new experiment you would then choose this draft apply it to the original campaign and run an a b test basically comparing this search halloween campaign to the new draft halloween campaign that we had set up and we just made adjustments to as you can see by this view that is no longer the flow there is no section on this experiments tab that allows you to create a new experiment instead we need to navigate to the all campaigns view now before i do that i do want to answer the question that i had originally when i looked at this which was then what are draft campaigns for anymore what's the point even though draft campaigns can't be used in experiments they do help you stage changes in google ads that you want to roll out to your campaign before actually making those changes in a draft campaign you can make all the adjustments you want and then the only option you have is this apply button up here which means that any of the changes in this campaign will be applied to the original campaign that you made the draft off of think of this effectively as google ads editor just in the platform itself so if you don't have access to editor or you really hate working on that platform and you would rather make changes in google ads and then roll them out all at once draft campaigns are how you would do that so now we're going to start talking about the new experiments flow in google ads as you can see here i'm on the experiments tab and it directs me to the all campaigns view so i'm going to go ahead and just click on this blue link here and now you can see that it's directed me back to the all campaigns version within the interface if i scroll down in the second menu bar here you can see there are a handful of different options within the experiments drop down all experiments which is where i am now add variations custom experiments and video experiments we'll get to each of those different options here in just a minute but effectively this is going to be the main hub for your experiments in google ads i have two examples that i'm going to run through in this video the first is going to be creating an experiment with new ad copy and the second is going to be setting up an experiment to test two bidding strategies against each other in this account i'm going to set up an ad test so i'm going to go up here and click the blue plus button to start a new experiment now we're directed to choose what type of experiment we want to use we can optimize text ads a video experiment or a custom experiment you'll probably notice that those corresponded to the options that were available in the experiments navigation as well so for the example in this account i'm going to do add tests so i'm going to click optimize text ads i'm going to click continue the first thing we need to do before creating any new ad variants is to determine what campaigns we want to run this ad variant test in you can opt in for all campaigns or select campaigns for now i'm going to choose just that search halloween campaign that we had in place to make sure i don't get any errors but you can choose any number of campaigns that you want as long as they are of the same format and the experiment that you're going to run will be applicable across them next we get to choose the target ad type for this add variation test this is going to be something that will be useful for the next couple of months and then it likely won't be useful anymore for the text ad variant because right now you can see it defaults to responsive search ads but if i click on text ads we get the error here that's starting on june 30th 2022 you'll no longer be able to create or edit expanded text ads this is something that we've known about for a while the only thing i'm not quite clear on at this point is whether or not you'll still be able to create ad variation experiments based on those text ads that's something that we'll have to find out on july 1st 2022 so for this video i'm going to use just the responsive search ads since that'll have more longevity and hopefully help all of you out for many years to come before we set up the test itself there's one last way that we can narrow down which responsive search ads we're going to create new variants for in the halloween campaign and that's using this drop down for filter ads once i do that i'm able to select from a drop down which component of the ad i want to filter for so maybe i'm trying to only test against certain ads that have a specific headline if i wanted to do that i could click on this and then i would enter the contains or equals starts with whatever makes the most sense language into this field and say that i want to narrow down to only ads that have paid media pros as the headlines so now any of the ad variations that i make will only test against ads in this campaign that have paid media pros as one of the headlines in the responsive search ads you can set up any number of filters to make this as specific or as broad as you want to start creating the add variant we now click continue the first option when we create a new variation is to determine what type of changes we want to make do we want to find and replace text within an ad do we want to update the urls effectively making an a b test between one landing page and another or do we want to update the text completely i'll show you an example of what find and replace and update text would look like for find and replace let's say that i want to change the headline that includes subscribe on youtube and change it to visit the youtube channel so i would type in the text that currently is in the ad variant in the find text section then you can opt into matching the title case or whole words only whichever makes the most sense for you and then you add in the language that you want to replace it with so now any variant of the ads we have in this campaign that say subscribe to the youtube channel in one of the headlines will be replaced with visit the youtube channel so that's one type of experiment that you can run with ad variations the other is the update text field so for this type of test you determine which portion of the text you want to update whether it's headlines or descriptions you then decide whether you want to add remove or pin the language that you're updating into certain places on the responsive search ads or into the rotation so for this example let's say i came up with a new headline variant that i wanted to test as a part of my responsive search ad and i wanted to run that updated headline variant against the control here i would just write in whatever my new headline is and by leaving the action as add that means that it's going to add just this one headline to the responsive search ad in my experiment campaign if you wanted to remove a headline from the existing ad variant you just need to change the action to remove and then you would need to add in the language that would match to the headline that's in the existing variant to tell google ads to remove that headline in this experiment campaign the last option as you can guess is to pin certain components of the ads in any specific area so here we're choosing to pin a specific headline and you would need to add in the value here so that you could match to that headline and then pin it in either no position position one two or three effectively allowing you to set up an a b experiment between your existing responsive search ad without any components pinned and a new control variant where a number of components of it are pinned into a certain location so if you're unsure whether or not pinning is improving your performance you can utilize an ad variation campaign experiment to set up an a b test between a pinned and unpinned version and get scientific data telling you whether or not it's working or not once you're finished setting up your new ad variation for your experiment the last thing you would do would be to set variation details so i'm going to click this open really quickly you would give your ad variant a name determine the start and end date and determine the experiment split i'm going to talk through each of these more in depth when i do the next experiment which is going to be comparing two bid strategies against each other so let's hop into the other account and i'll show you how to set that up since i want to set up a test between two bidding strategies in this account i'll need to choose a custom experiment i then get to determine whether or not i want it to be on a display or search campaign in this instance i'm going to use search and click continue you start off by giving your experiment a name i personally usually duplicate the campaign name and then add additional text to the end that at a high level describes what this test is going to be so in this instance the campaign name is search dash services and i'm setting up a max clicks bidding strategy against our manual click strategy so i've just applied max clicks to the end of it so i'll be able to easily tell which campaign is which once i'm looking back at it in the google ads interface you can also add a description down below if that's helpful for you next we get to choose the base campaign or the campaign that we are going to run the experiment against so just click the pencil here i then get to choose which campaign i'm going to opt into so like i said it's going to be the services campaign i'm going to scroll down a little bit and now even though it had me apply a new name up at the top i have to adjust the naming convention a little bit because there's a new field down here that says suffix for the treatment name and it's got the language that i used up above to name the experiment itself but then when that applies to a campaign it duplicates the first portion of that so for down here i just need the new suffix that i added to the campaign name so i'm just going to delete the search dash services when i did that the name refresh down below to the exact name that i wanted to show in the interface so now i'm going to click save and continue and now we're back to a similar flow of campaign experiments the way that things used to be you'll remember from the previous portion if you watched all the way through it that this looks very similar to a draft over here doesn't it sure does but it's not called a draft anymore it's just the staging area for your campaign experiment and it will live exclusively in the experiments portion of the account and you can tell that by looking up here at the breadcrumb it's in all campaigns experiments search services max clicks but just like we did with the drafts version of the setup the next thing we get to do is make the adjustment to the campaign that we want to make at this point you're able to do whatever you want you could upload an entirely new set of keywords you could upload all new ads you could change the location targeting ad scheduling everything that you can control from any of these aspects over here will be part of this new experiment campaign and it'll run against your control once you get everything set up as i've mentioned a couple times for this specific account the only thing i want to do is change the bidding strategy to max clicks and i'm going to leave everything else the same so to do that i'm going to head into settings i'm going to click on the campaign itself and now i can head down into the bidding section and make the changes that i need to to opt into max clicks if you're interested in a rundown of all of the available bidding strategies on google ads you can check out the video that's at the top of the screen right now but for the sake of this experiment i'm finished doing everything that i wanted to do once you're ready with your campaign you just need to head up to schedule and now we'll be prompted into the settings that are part of the campaign experiments so as i mentioned earlier with the ad variation test i'll run through some of these other areas a little bit more specifically like the budget split and the experiment dates but all of them are going to be pretty much identical to the working process that would be for the ad variant type of experiment compared to this custom experiment one that i'm setting up now but with custom experiments it's not entirely sure what your goals are because you can change so many different items so the first thing you need to do with a custom experiment is determine what your goals are and you can do that by selecting certain metrics so let's choose from this drop down and these are the only metrics that are available clicks impressions cost cost per conversion conversion value per cost conversions and conversion value hopefully one of these is going to be an outcome that you would like to see in your account personally for this experiment i'm really trying to see if we can get the cost per click down as compared to a manual bidding strategy you'll notice here cost per click is not one of the metrics i can choose but clicks is going to be about as close as i can get so i choose clicks and then i get to choose which direction i want that to go since i want my cost per click to decrease logically i would like the number of clicks to increase so i'm just going to click increase and since that's really my only goal i do think it's important that you set another variable that google will see that tells it that you really don't want to see any significant change so in this second metric i'm going to click down here and another important metric for me is cost because i really don't want my cost to increase while i'm seeing more clicks come through so here i want to see no significant change for cost while my clicks are increasing depending on what your goals are with this experiment you might have two metrics that you want to increase a couple that you want to decrease or maybe you're just making a lot of changes but you really don't want anything to change that's a viable outcome that you want to see in these experiments but you do need to give google some direction as to what you're trying to achieve in the next section we get to determine budget split the main takeaway here is that you need to allocate a percentage of your budget to this new experiment compared to the control or as google calls them the trial and base campaigns to make sure that each of them is getting a good amount of volume through so that you can actually make some decisions the tests pretty much always default to a 50 50 split and that's almost always what i use for my own campaign experiments but if for some reason you're willing to let a campaign run for a really long time and you want to minimize risk you can adjust this so that your trial campaign only gets maybe 25 percent instead so once i've changed that you see that the trial campaign gets 25 the base campaign gets 75 depending on what your timeline is and how risk averse you are make sure that you set this up to make it make sense for you in this advanced options field down here you get to decide which campaign split options you're going to use do you want to use search or cookie based breakdowns of your budget so if i hover over each of these it'll give me the insight as to what they mean with a search base experiment split you assign users randomly to either your experiment or original campaign every time a search occurs so that means that if i search for your keywords twice i could theoretically end up in both the trial and the base campaigns on the other hand a cookie-based test means that each person will only see one version of the campaign regardless of how many times they search so i can search for your keywords many different times and i'll always be opted into the trial campaign if that's what i saw first or the base campaign if that's what i saw first basically whichever campaign you are introduced to the first time around is the one that google will constantly send you back to with those same settings ad variants whatever you're trying to test personally i nearly always use cookie based because i think that makes the most sense but if you have a reason to run a search-based campaign you can change that option in these advanced options down here we're going to scroll down a little bit just to get to the rest of it next are the campaign experiment dates you get to choose your start date which will always default to tomorrow whichever day you're on it'll default to tomorrow but you can change it to activate the day that you're setting up the campaign experiment if you want to you can then determine an end date and you have three options for that you can either say none which means that this campaign will run indefinitely or until you manually turn it off you can tell it to run for a certain duration and apply a certain number of days that you want this campaign to run maybe you only want it to run for 7 or 14 days you can choose that option or you can determine a specific end date that you want something to end on if you want that option you just choose select a date you get to choose from the calendar breakdown like we saw from the start date a minute ago and your campaign experiment will turn off on that end date lastly we are in the enable sync section which i think is very useful because it does cut down on some of the admin work from the original way that campaign experiments worked if you leave enabled sync on which is the default setting changes made to your base campaign will automatically sync to the trial campaign so you don't have to make changes manually in this specific campaign i'm trying to test a different bidding strategy if during the course of this experiment i also need to update my ad copy tests or i decide that i want to opt out of the search partner network leaving the enabled sync on means that whenever i add new ads to the base campaign or change those network settings they will automatically be applied to the trial campaign as well so i don't have to do it that was not an option with the previous version of campaign experiments the previous version had this off setting which meant that changes to your base campaign will not sync to the trial campaign so in the past or if you turn the sync option off each time you make a change like adding new ad variants to your base campaign you will also need to manually apply them to your trial campaign as well depending on what your workflow is in the platform i would say that most people are probably best off leaving the sync option turned on but if you have a number of different people making changes on the campaigns or if you plan on making adjustments that might actually impact the specific variable that you're trying to test first maybe don't do that because you're going to invalidate your test but it might make sense to turn sync off but in the vast majority of cases i think it's going to make the most sense to leave it on once you're finished with that you just come down and click create experiment now to show you what an experiment for a bidding strategy would look like i'm going to hop into one last account and show you one that's already been active for a little while once your campaign experiment has been running for a while it'll look something like this you'll have two line items within the campaign portion of the interface and one will be the base campaign or in this instance the general campaign and then the experiment or trial campaign will have another line item that has your naming convention associated with it so here you can see that i was also running a max clicks bidding strategy so it's got max clicks applied to the name each one of them will have its own set of stats over the period of time that it's been running so you can see at a very high level very quickly which one is performing best over any given date range in the account if i then head into the experiments portion of the interface you'll then be able to see all of the campaign experiments that i currently have active in this account and you can start to make new tests and get new things set up if you want to but once you're ready to make changes or end a specific campaign experiment you would come in here click on the experiment name and now you can see at a very high level what the impact has been based on the metrics that i told google were important enough for me so here it says a 19 increase is statistically significant but then it also shows that the cost per conversion which i wanted to remain relatively stable is also up 74 it says not enough data because this account is relatively low volume on conversions so that's not a complete surprise to me but it at least shows at a pretty high level what the two key metrics were that i told google to keep an eye out for depending on what the findings are you can either apply the experiment or effectively roll out whatever settings and applications you had set up in the trial campaign to your base campaign and that will overwrite that original campaign and it'll follow through with the trial settings or you can simply end the experiment and make whatever adjustments that you want to the base campaign but not automatically from the trial campaign settings although the new campaign experiment flow in google ads is a bit different than what it used to be i actually think it's a little bit more streamlined and makes more sense it's also very nice that they have new experiment types for just ad or video experiments i didn't show you the video experiments test because it's pretty similar to the ad experiment so if you want to get more information on that you can go back to that add test section but as you can see by this experiment summary tab using campaign experiments in google ads can be a really easy way to test different hypotheses against each other in a very scientific way and then be able to get really high level looks at the metrics that are important to you and how they're being impacted by those changes if we missed any portion of the experiment set up that you'd like some answers to or if you have any other questions about anything in this process feel free to leave a question in the comments below thanks for watching our video if you thought it was useful give us a thumbs up below we release a new video at least once a week so if you want to get notified of when a new one comes out be sure to subscribe to the paid media pros channel