A/B split testing ads explained | How to properly test with Google Ads and Facebook Ads | Nikola Lugonja | Skillshare

Playback Speed

  • 0.5x
  • 1x (Normal)
  • 1.25x
  • 1.5x
  • 2x

A/B split testing ads explained | How to properly test with Google Ads and Facebook Ads

teacher avatar Nikola Lugonja, HR and Marketing Instructor

Watch this class and thousands more

Get unlimited access to every class
Taught by industry leaders & working professionals
Topics include illustration, design, photography, and more

Watch this class and thousands more

Get unlimited access to every class
Taught by industry leaders & working professionals
Topics include illustration, design, photography, and more

Lessons in This Class

12 Lessons (1h 21m)
    • 1. What is split testing

    • 2. Examples of split testing ads

    • 3. Why is A B testing so powerful

    • 4. Important split testing rules (ukljuci link ka onom vebsajtu za metrics)

    • 5. Ideas for Facebook and Google A B testing

    • 6. A most important rule for ad testing (rule #2 from the previous video explained)

    • 7. Case study #1 (placement testing)

    • 8. Case study #2 (delivery optimization testing)

    • 9. Case study #3 (audience and ad creative testing - 2 variables)

    • 10. Case study #4 Google Ads (demographics and ad variation tool)

    • 11. Case study #5 Google Ads (test different parameters)

    • 12. Thank you

  • --
  • Beginner level
  • Intermediate level
  • Advanced level
  • All levels

Community Generated

The level is determined by a majority opinion of students who have reviewed this class. The teacher's recommendation is shown until at least 5 student responses are collected.





About This Class

Everything you need to know about split testing ads. We are going to take a look at examples, explain why split testing is so powerful and important, what are some important rules to have in mind... Moreover, we will take a look at 5 case studies based on which we will be practicing split testing set up. After this course, you will be able to test your ads with Facebook and Google's platform. 

A/B split testing ads (lectures):

  1. What is split testing

  2. Examples of split testing ads
  3. Why is A/B testing so powerful

  4. Important split testing rules
  5. Ideas for Facebook and Google A/B testing

  6. A most important rule for ad testing (rule #2 from the previous video explained)

  7. Case study #1 (placement testing)
  8. Case study #2 (delivery optimization testing)

  9. Case study #3 (audience and ad creative testing - 2 variables)

  10. Case study #4 Google Ads (demographics and ad variation tool)
  11. Case study #5 Google Ads (test different parameters)

Links from the lectures:




Meet Your Teacher

Teacher Profile Image

Nikola Lugonja

HR and Marketing Instructor


-Multi-year experience in both HR and digital marketing. I started my career in Marketing, but over time I dived deeper into the world of Human Resources. I find these two areas commonly overlapping (e.g. when it comes to employer branding), therefore I will also try to link them in some classes. 

-Here are 4 values that I always keep in mind when preparing and publishing classes:

Keep it short and sweet - eliminating waste i.e. everything that does not bring any value and ensuring the students get the most out of every single second Unscramble the content - making things simple to comprehend and outlining the most important takeaways Always explore - stepping into the unknown to extensively research new topics and broaden the knowledge spectrum Improve... See full profile

Class Ratings

Expectations Met?
  • 0%
  • Yes
  • 0%
  • Somewhat
  • 0%
  • Not really
  • 0%
Reviews Archive

In October 2018, we updated our review system to improve the way we collect feedback. Below are the reviews written before that update.

Why Join Skillshare?

Take award-winning Skillshare Original Classes

Each class has short lessons, hands-on projects

Your membership supports Skillshare teachers

Learn From Anywhere

Take classes on the go with the Skillshare app. Stream or download to watch on the plane, the subway, or wherever you learn best.


1. What is split testing: Hello, everyone. And in this first lecture, we're going to talk about what split testing is on. I prepared here short and I believe, easy definition. Easy to understand on what a B testing is. So as you can see here, a B testing, also called split testing, is a back deck by which you find out which at headlines body copy images called diffraction bottles or a combination of the above work best for your target audience. Moreover, you can experiment with other variables, such as audience placements, delivery of organizations, etcetera, in order to reach the highest efficiency off your rights So you can A B testing is really powerful goal and boat. Since this course we're talking about Facebook and Google, they managed to provide their users and advertisers with easy options to do a B testing so you don't have to duplicate manually before some basic basic testing's. And that is a really powerful feature that you're going to see for the wrong. So if you're really interested in this topic on, you want to see how those a B testings work on you can really find the most efficient way off publishing and promoting your business with with this tactic. So I suggest that you stick to the lectures on and we're going to see first of all, some examples. And then we're going to see some case studies based on which we're going to set up the different A B testings on different platforms. Thank you and see in the next lecture. 2. Examples of split testing ads: Welcome back, and here I want to show you five examples that I run to on the Jeff Bullets website that he used as examples for a B. That's things I suppose eso here on the first example you can see. I mean, it's really hard to see whether somebody is performing a B testing or they're just publishing ads at the different moments when you look out from this perspective. But let's say that this was just these examples were done. This a B testing's you can see on the left side that we have a one image, and on the right one, we have the other one. So the body attacks and description is the same. Cold directions. Bottoms are the same links here, provided with description and headline also. So the only thing that is different here is the image, and we're not, of course, sure what's behind this? Whether audience is the same, whether placements are, but I suppose the placement is the same. But whether the delivery optimization, whether the audience or some other variables, are the same, you cannot know that just from seeing the ad that that's only possible to know if you're the advertisers, advertisers, the person setting up the ad. But you can see the clear difference here on, uh, for the page here can. With this plan, testing can actually test which adis performing better, whether this picture on the left or this picture on the right. So where do which which add brings more engagement on which had will people clink mawr and make more conversions and so on? Here is the another example off Basically the same image on everything else is the same. But you can see here this is without a text on the image and this is with the tax and with the logo. So that is also something interesting. And keep in mind that Facebook has that 20% at next rule, which is not all mandatory to stick to, but you're at will we perform worse if if you break that rule. Moreover, some examples here, you can see that here we have a completely different next here. So here's said here on the left. It says, Get the small business funding you need and here it's a bit longer, but you can see here the images are completely different from my guests. here I can. I could say that this ad on the left was targeted. More towards men at add on the right was started. More towards women. That doesn't Miss Terry has to say it, but you can see here that it's a powerful tool to do just that. Moreover, another example where everything is basically the same besides this tax that it's added on the right and the last one is also similar. A Z, the 1st 1 that we saw with just different image on here. You see a person holding a phone here, you see just the phone so that fix here can test which I this performing better. So this was just the quick over. Here are some examples, some basic really basics examples where you could see that this pages and businesses were probably testing with different at designs on Facebook on, and we don't know whether data's that different audiences delivery optimization ins or maybe any other. Very so thank you for watching on C the next lecture 3. Why is A B testing so powerful: welcome back. And in this lecture I want to actually show you an example 12 which I will try to explain why the split last thing is actually imported. How you can profit from it. So here I I took an example from ad espresso. They were doing split at best think for their free e book. And as you can see here first on the left. So they had a short short back saying, Learned the secret to grow your business five times more effectively with Facebook ads, get to the free E book. So here you would have the option to downloaded. You would have option to go to that website, maybe to give some email or something. I'm not sure whether Strategy waas, but anyway, you would get a free e book. This was the add on the left. And Adam, the right war actually have the same objective. However, it was designed to be differently. You see that the image was different and the tax waas first in brackets it was written free e book and that it was learned how to successfully convert your Facebook ads traffic into leads and grow your business five times more effectively so they change attacks. But saying the meaning is pretty much to say on here to call to action but was the same links was were the same on the text had headlines as I can see are pretty much the same. So there was a difference in this text on the image so below you can see they published the results from these the stance they did and both ads they had 10,000 impressions on from and they spent $150 on each. So this is really important to have these two metrics equal, so we can see other man tricks from the results and compare which one every four batteries . So you can see here. This had on the left had 237 legs clicked. A rate was 2.37 on the on the right, it had 187 legs and clicked. Rate was 1.8 point 87% regarding the sale, so conversion rates On the left it was they had 28 taels, or 11.81% and the ride it, they had 16 sales. And don't get confused says here our sales since it is written free book, I suppose a conversion waas some kind of a sign up, maybe for you to input to given email and you would get a free book. So they consider that as a sales. Since I don't see here that they were selling anything, they just wrote sales. But they were referring toe make downloading the free evil Corgan or getting the emails or something like that. Anyway, they spent 100 $50 as I said, and cost for each sale was for this side of the left, $5.35 on on this one on the right. It was not born 37 is you can see here 70 deeper around 75% increased and more expensive for that was when they were performing this ad. So imagine if they were selling this book on the inside of giving it for free. Imagine if they were selling and they wanted to see what is more, more cheaper for them. What is actually cheaper for them. What is better for them to which adds to choose and by this statistic, by this metrics, we can see that Actually, this ad on the left is better for them because with the same budget and with the same number of impressions they had higher picture rate. They had higher conversion rate on those two things actually put the cost per sale to be lower than the add on the right. So, in the same budget, they managed to do perform or sales with the ad of the left, then with add on the right. So this is just a new example off what actually split at split testing is used for. And why is it so powerful? Because if you're not sure what works better for you, you can test for a short while with short with a small budget with only a part of the whole campaign. And when you see what it's actually better with your audience, where they like war, you can choose that option and just push it to the end. So that will be it. Thank you for watching and seeing the next to you 4. Important split testing rules (ukljuci link ka onom vebsajtu za metrics): Hi, everyone. In this video, we're taking a look at some in court. The A B test grows some rules and tapes that generally you would want to have in mind when you're performing split, besting and the 1st 1 would be that that's that you should test only one variable at the time. That would mean that although you can several ones and maybe you have several tests that you want to perform, you should definitely use that by step at best, only one at the time, except in some except in some rare cases that we're going to take a look. But you should test only one variable at the time, because that way you will be able to see the full full result. Who is that to see the main driver and the most important factor, which which makes an ad before better. So keep that in mind and best only one variable at the time. Second rule would be to use the right campaign structure, and this is really, really important. Maybe I would say this is the most important role on. We're going to explain this in detail in the following videos. Basically, this is really important because you don't want to your ads to get at optimized by by the platform. The 3rd 1 ensure that your A B test results are valid. That would mean that you should gather if you recall. We mentioned that the test should shouldn't be too long but shouldn't be too short. That means that past, as the word says, is a preparation. It's a way off just partially doing something which results you would use later on in the main campaign. So that's the situation right now. You don't want your test to be the whole campaign you want to test in order to make your campaign more efficient so that that should be only a part off your full campaign. And that's really important to know, because you don't want to get too small. Two small data, but you don't also want to get too much information. So important is Teoh know which which number, which the size of data is the relevant one on to collect data from which you can actually take the most important conclusions. So that's really that that's the That's the important thing. Maybe you can go. Uh, maybe it's better to have a big more data than to have less, because last day they can drive you with with worst conclusions. If you have more data that, then that would mean that you spend more money that you should have passed. But that's that's relevant and there are clear boundaries, so don't worry about that. But also important to know is that you should compare the results, and here I have a great website where you can actually, I will include the link in the description. But this is a great website where you can actually calculate the conversion rate. So here you can see let's say that you want to dust to add. So on the 1st 1 you you input the number of misters on the page that you had. Let's say that it was 500 on here also because we said it for the testing toe work. It should be equal, and the total number of conversions was here. Let's say 26 and here was 18 so you can see here that the conversion rate was 15. If I put like 74 here and there may be here, I can put I don't know 37 here. Okay, here it is, written with letters. I'm not sure why, but anyway, here it's 15% conversion rate is here is the 71 If I want to add another one, I can do that. So let's say that here I had 500 here I had, like, 97 so that would make it 90. And let's say that the 1st 1 had actually let's say that I had different number of ministers on the page. For some reason, eso I would also be able to compare in that sense so I can see what my conversion rate here was. And that's a great tool. You can see other things that are available here, so that would be the rule for number. That would be the rule number Tree at the rumor before would be to make a difference. Is currently visiting. This is a very, very important. And here I took an example for you to understand what I'm talking about. About two images about you can see that difference is very small. If you're still figure it out, the actually the only difference is actually here of the tree. There are different marks here. You have more of them here you have a few work, but it's really hard, Like considered. Actually imagine that split testing is like testing different ads, and the with that you want clients or users to give you feedback what they like more so if you provide them with the same ad basic. I mean with the as they have small differences and they're not able to see the difference Cleary that you can expect from them to provide you with the punctual feedback here on in the other example, You can see that the colors are different old, the other things a relatively similar, but you can see that here they can provide you with something by. They can maybe click more this ad than on this one, and then they can say that they like more this green color and now this blue one. So that could be some kind of a people here. It's really hard to see the difference. So keep those things in mind. Keep these rules in generally line, which, because they're really border on the tried to use them always. Thank you for watching and seeing the next lecture 5. Ideas for Facebook and Google A B testing: welcome back. And in this video we're taking a look at. So my ideas for Facebook, A B testing and for Google a B testing. First of all, we'll start with Facebook. And as you can see here, the first idea would be to split thus different campaign objectives, which is the broadest possible options so you can test different campaign objectives that are available on the platform. The next one would be to test different audience on Facebook to test the different placements. For example, instagram and the Facebook news feed or I don't know, audience, network and messenger audience, instagram and messenger and so on. Experiment with different bidding off methods and options. With this, the more best one split. That's your Facebook had images We saw those examples split that stuff all those versus custom illustrations. This is really, really deep on. You can also do that. I mean, these ideas can be endless. Split best images vs videos also very, really very important on the that way you can do You can actually do this with only images and videos, but you can do image. One image versus Caruso. You can do a video verses slide show. So a B test your Facebook headline ad copy on the landing pages. So I found this list on the Web. It's really useful. Those are some great ideas, but not the only ones. Keep in mind again those rules that we mentioned earlier and here I actually found also fate great. The image off the ads on the Web, where you can see that staying Facebook page was Osama was publishing different ads are not sure whether they were from the split testing, whether they were doing is protesting or not. But anyway, you can see here that they had a relatively same objective, I would say to They had call to action was signed up in every ad they had from the script. Different tax, different image. More or less different headline and description Link was, as I can see here, always the same. But you can see here what does when, when you do split testing and when meets mentioned that through that difference should be difference should be visible. This is what I mean. So he Ricans clearly see different ads, and if you're a customer, somebody else for your for your feet back you can provide them with the clear feedback and say, Hey, I like this more. I like this less. But if you're not able to see the difference than you have difficulties answering that questions and providing people so that will be it regarding the Facebook that's protesting , that we're headed to the Google ideas. And they keep in mind that Google split testing the two is different than Facebook or worse . We're going to see that later on when we take a look at setting up those protesting. But the Google is pretty different because their platform is different. So they're more concerned with some ad variations. Threat rather than clear, rather than having the clear button for split nesting in a space book does have. So split Testing, in my opinion, is generally easier to do on Facebook rather than on Google. And you're as I said, you're going to see that. But here are some ideas for your Google will be testing. So again, as on Facebook, you can split as different campaign objectives. You can choose one objective on, then choose two different campaigns to achieve that wanted to see whether you're performing better with one or the other one. You can test the different demographics of your audience. Different genders, different ages, different incomes in someone you can tell from placements for the ad. Maybe try some at you to maybe someone displaying that work and see what works best for you . Experiment with different voting methods, as as you can Facebook, but I would suggest that for the more advanced users as protest your images on this plane at work. Similarly, what you can do on Facebook with different images You can do that here on this by network, and see which banner and which which add is performing better as they head lies that they're from the ads copy and you're going to see this for these things. So on also landing pages for these things face, Google actually has a very good tool with which you can really fast change those options you can see. You will see that they have to, which where you can actually click just one bottom and invert that switcher headlines that that's it, so you don't have to add it manually, so that will be a Z, said a B. Testing is a much more used on Facebook rather than on Google. And for that reasons, you saw that during the course I was showing much more examples on Facebook that from rule . But you can do that. I mean, maybe even if they don't have some options building, you can do those things manually. It requires a bit after, but it's a bit more effort, but it's possible. So I suggest that you do that if you really want to make a ride efficient, Thank you for watching and seeing the next lecture. 6. A most important rule for ad testing (rule #2 from the previous video explained): welcome back and we're continuing with the course. This is a very, very important lecture. In order to understand split at testing if you recall in the previous video I went actually previous videos. When we were talking about those rules, there was the second World where I said that choosing the relevant campaign structure and the right campaign structure is really important. This is that thing, and this is very important to understand. So when you do split testing, it's really, really important to properly set your test because you might make a small mistake that seems not so relevant but can actually ruin your whole past. Let me show you what I mean. Here you can see the simple graph which says that in one campaign you have one at SET, which has three ads. This is the type off off a structure that is that is used on you two on the Facebook on Google. It is cold, different, but anyway, we always have three levers levels. If you recall, we start with a campaign than we determined the ad set and in the end, we had add at creative anyway. Important thing to know is that when you reform split testing. Okay, You need to put different at this in different asset. And that's a this craft, the 2nd 1 So as you can see here, each ad So in this example, we have three different as that we want to test here. You can see that each ad has its own at set. They're all under one campaign. But in this example, each at contains its own asset. This is different than this example where you have three ads within one at set. Why is this a very important to know no matter which variable you're using, it's important to use different ad sets because if you use this example that you're seeing right now, where one at that contains multiple ads, what's going to happen is that the platform is going toe out, optimize your at. And that means that if you put your budget and publisher had to work that class for will recognize at one moment which add is performing the best, and it will start pushing only that at it will start spending money on Lee on that ad because it's always as the best one as the most as the one that is performing the best. On the other hand, here you can see that each ad is within different Ad said, And that means that you can make a true test because money will be spent equally on each ad set. And that is really, really important in order to the sea or metrics and make relevant conclusions in the end. So keep this thing in mind. Make sure that I mean there's there always could be some exceptions, but for the most part it's important to keep different when you're testing different variables to keep different ads within different assets, because in that way you will avoid out optimization. So again, sorry about this. That's the next video. Don't use this this structure to keep multiple add within one that said, but keep each add in separate at set. That would be thank you for watching and seeing the next video where we're finally starting with setting up display. It adds a zoo. So we're going to take a look at some case studies, and from there we're going to set up a different test. Thank you for watching and seeing the next lecture 7. Case study #1 (placement testing): Hello, everyone. And now we're finally taking a look at how to set up split test Inc. And, uh, I was thinking about how to actually how to cover this topic, and my best guest was to actually provide you with some case studies as examples based on which we're going to set up and maybe testing. So the 1st 1 for Facebook would be like this. We want to run a split, the test for the same shoes at and the same audience on Facebook platform and Instagram platform. And in that way we want to see where it will perform better. Actually, with the lower cost per click in this case, basically, we're what what we're doing. It's using one variable, which is a placement, and, uh, we're going to set up. As I said, the shoe ad regular one. And the goal would be to gain traffic to our website. Let's say like that, So I will choose here. So I went to, uh as manager and creating that I got this window and I will choose, objected to be trafficked. Let's say I want to bring traffic to my website from where they can show I will name this campaign a be best shoes and this is the most important step. When you're using this option to click, create a B test, you can choose your variable here. In our case, it's going to be placements. They if you recall we mentioned it, we're going to test whether it will perform better on Facebook or Instagram. But if you don't choose it there, you can click. Continue and you will have to choose it. Now, in our case for this case that this placement and the now getting other other options I won't go so deep in these other options because I've covered them in the other part of the courses and lectures and so on so you could take a look there. But I will leave this to the website and what I basically have here displacement that is unbearable for at that A and that said, be so. As you can see here we have two s that not just one which is very important. What I want to do now is basically what I'm doing just to make to be clear of this, I'm using the same at so the same image, everything will be the same. I just want to test Facebook and Instagram to see where it will perform better. So for the first that's at a Let's say that I just want to use it for Facebook. So I will switch off all other options there. Even messenger. And I would just leave here Facebook for all devices. And for the 2nd 1 I will do the same thing. But for instance, so as you can see now, I excluded the other options and I won't left with the Instagram for the ad said to be and for the asset A. I'm only left with Facebook. You can see that here. Okay, so that's the variable that we're testing in this example, if I want to add another variable So let's say that I want to use the test. I want to do that for audience that work, for example, I can do that. You can do up to five, the ad says here. So keep that in mind for this example I'm not using at SETC and for that reason I will exclude it. So I have set B, and that's at a and the other things are basically the same since we mentioned that we're talking about shoes. Let's say that to be only want to target men on the location could be a US. For example, let's put the whole United States, although it is very broad. Moreover, I would exclude a Category two, let's say, for maybe two 35 and when it comes to interesting behavior are just shoes. So I'll see what pops up. So shoes. Actually, I'm not using. Sure, I think I will go with sneakers. I forgot. Yeah, I won't go actually with snipers. That's the image I I downloaded for this one. But anyway, let's say like this board on may be, I can take a look at suggestions like foot, where online shopping. Yeah, can actually add shoes more So it's the same interests. Okay, that will be it for it. You can play with this. You can bro's behaviors and graphics and so on. I don't want to go so deep right now into this because they go. This course and part is to generally show display testing and you can see the audience will be saying for the boat that sets. Moreover, delivery optimization is the same it will be for the link clicks and I will be charged based on the impressions. I can set my daily budget to be 50 euros, and it should run for five days. So, as you can see here I have it is written a chance of finding a winner. It's 85% and that is only the estimate. But what is saying that it's a good thing? And the thing there's Facebook is telling me is that they will have enough money, enough time to to test both as and to see which one is performing better. So that's really important thing. If you have a low number here, that's really bad, because maybe Facebook won't be able to find the winner for you after that period or with that budget here, I will spend enough like 250 euros for five days, and that should be enough for Facebook. Next on, I want to actually name I add should be shoes at before sneakers. But don't worry, and I will choose a single image that I just found. So it will be this one. And, uh, yeah, what I actually have here is the basic image I can put any attacks. So by now, look great on I can put She was all mine dot com. I can put the headline here to be my shoe. Yeah, that would be mostly it on, like, Jews shop now. So what I have here is and that and you can see here how it would look on this is for Facebook different, different options that are available. So, yeah, that that will be mostly it. So these ad adjusted optimized for the instagram, but with the same image text will also be published on Facebook and Instagram. And after some period of time, I will be able to see whether it's performing better on Facebook or Instagram. So thank you for watching. That would be our first case study. I choose to be simple and I hope that you understood it. Very clear what we did And what the final effect off this split best thing would be. Thank you. And see in the next lecture 8. Case study #2 (delivery optimization testing): hi, everyone, and we're continuing with the course where we're going to look now with second example case that and this one says that we want to test again. The same foot foot were at with all the same variables, but we want to see whether it will be cheaper for us toe, optimize it for impressions or link clicks, that is, should we be charged based on CPM cost per mile or cost per click? So the same thing is in the previous examples example. We're taking a look at the same ad, but instead off testing different platforms we did with the previous one, whether we wanted to be on Facebook or Instagram. Now we're going to put it on both platforms or no. Let's say that we run that. We ran that test. We saw that it was better to put it on Facebook, and now we want to run the second test to see whether it will be cheaper for us to optimize our delivery for impressions or link clicks. So in order to see how that works, let's go again. Back to you as manager click to create ad and choose the traffic. Choose to create a B test and let's say let's call it a B and, uh, foot where off the organization test. So the previous one was the placement test. Let's say that I want to choose now, the delivery optimization of which is now here and press continue. All other features were remained the same as we did in the previous one. But now my ad sets again. I'm saying that asset Way have different assets. Different ads, which is really important, will provide me with different delivery optimization. So that will be a variable here. What I want to change is here for the at set A. I want to get charged based on impressions and on that adds had be I want to get charged based on licked playing clicks that this cost per click basically what I did. So I will have an ad. It will run on Facebook and I want In the 1st 1st example, I want my abs to be charged based on 1000 pressures, and in the 2nd 1 I only want to be charged when somebody clicks on my on the link that I'm providing in the ad so I can add more ads. If I wish. As you recall, we set five maximum for now. And now the audience it will be saying for the boat that so we will put the US here. Sorry, we'll put the United States here. Okay. Facebook is not recognizing. Okay. And, um, the category waas from eight into 30 is very cool. And here we put put where? Sports? I think so. These these variables are will remain the same, Just takes sneakers along shoes on my shopping. That will be it. Placements. So, as you can see now, I only have one placement to choose because my variable is not a placement is it is optimization on, like, unedited since we said it in the previous test. Let's just imagine that I would say that Facebook warm and it was cheaper. And now we want to exclude other options on just trying it on Facebook. So instagram proved to be more expensive. But as you can see, for example, let's say that we did that test. We didn't test audience network and Messager. So that's why I said that these deaths could be run in like, indefinitely. You convert as many tests as you wish, and sometimes it's very hard. It can be costly, but that's why sometimes it's important to to narrow your your choices and options and to test only the most relevant things. Budget Again, we'll put it to be. Actually, I can put it now to be, let's say, higher to 70. And as you can see here, chance of finding a winner is 87%. And if I reduced the number of days, just let's say to as you can see here, the percentage Paltrow because Facebook is actually telling me, Hey, I don't have enough time to test both of your variables to tell you which one is is the winner. And if I let it But here one day and both on Lee, let's say 20. You see that the chance of finding a winner is chopping, so the best way is basically, as you can see here, you have to choose 30 days it up until 30 days. But you can choose, let's say, five days and you can put here to be 80 and it will show a pretty pretty good percentage to 91%. That's really awesome. Next one ISS to set up on that and again everything stays the same. Now it's only optimized for for Facebook, and the text is recalled by now. You look great and Web site is that job. So show with us. I don't recall the things that I put in the previous one, but don't worry about that. It sounds Robin anyway, The important thing. And now if you're done, you just go to confirm and that's it. So now what we actually did is basically chose to show this ad only on Facebook with the same audience. But we're testing whether it would be cheaper for us to optimize at delivery for impressions or link clicks. And that's the option that we're going to see after the after those five days after Caddis . So thank you for watching once again. I'm sorry that I didn't go into many details on the other factors perimeters because I covered that in the other courses in the other lectures. Take a take a look there. Anyway, the most important here is protesting and I hope that you realized and understood how it works. Thank you once again and see in the next lecture 9. Case study #3 (audience and ad creative testing - 2 variables): hi, everyone, and we're taking a look at the third case study, which is a bit advanced with delivery. Abel's. But I think that it will be helpful to you to understand the split testing and how it works . So we're thinking about our Facebook campaign for Valentine's Day and during the Christmas holiday. We want to test if it will be cheaper and more effective to use first, a option. A. At design, which will target to the male audience like, for example, promotion to buy a makeup set for their girlfriends, revives or be at design, which will be targeted to the female audience, for example, promotion to buy makeup for themselves. Maybe they're choosing their own gifts, or, if they're single, they want to buy something for themselves. So we want to run the split test with two variables. As you can see here with eight. Create at creative and general audience to see which will before better during the Christmas holiday and then use the Civil War strategy for the Valentine's Day. So since Christmas holiday involved as they can have similar impact when it comes to this present gift buying, we can actually test the things or Christmas during the Christmas and decide what are we going to do for development Wednesday, Since there is not so far, it's only one month, month and I have different. So what we're going the most important here thing to know is that we're looking at two variables that that is really, really important at creative and gender audience. In the previous two examples really took a look at one very able. And now I'm going to show you how to set up that cell. Come to the ads manager click to create on. Uh, it's important to know you will get this. You wait just a second, so you will be seeing this one, as it usually is. But in order to make the process faster, you should go to switch the quick creation here, bottom or wherever it disposition, depending on the on the update of the interface. And you will get this, you should name your campaign. In this case, it would be a the testing or ah for at creative and, uh, audience. That is gender in this case, and the campaign objective will be tried traffic. But we're going to do a split best. And here where it says, what do you want to test? Will choose more than one. We'll name this one. So men, men at creative and this will be women Woman at creative. So you can save this. And, uh, now, as you can see, we will be. Yeah, okay. It will see the whole campaign here. This is a bit faster way to make him mad. But it's very useful, believe me. And here, most of the campaign level, you did what you could. But on the at that level, for the first variation, that is the a variation here. So where they look at this, we're going to set up. So the daily budget first, this is for the boat boat at said levels here. So you can see here it will be evenly split on that set. A let's say, 50 euros here on at that be we will. It will also be 50 year olds. It will run for four days, which is okay. Daily budget is 100 euros and we're targeting the United States. So let's put it, the United States, everyone in this location on we only want to target men. Let's say from 21 to 45 for example. And that would be our first at set with Okay, I don't want to go now, Really? Deep into the just say gifts on Uh okay, sure. There. But she just just just I can Maybe as for some behaviors, like maybe I could based on Democratic, I could like target people who are in the relationship engaged or married. So those would be my okay, those would be my my picks. And I could just put here's, uh, makeup. Although it doesn't make sense for a man to to do. Yeah, I will put cosmetics that that's okay. So because we're showing add to them in order to promote our products so they could buy it for their love Lovely ones on placements. I would just leave it to automatic. The delivery will be linked, legs and so on on. That would be from the ad set level. And now I'm going to have to Teoh at creating level where I'm going to upload a image from the computer. So it's a bit different now, but Okay, I would just choose this photo and publish it here. So as you can see here. Now, this will be shown to the man who are ready to by makeup for their once on, uh, provide. Um, no. Give the best gift to your loved ones to your loved one. Okay, let's just say that, okay? And this could be targeted to the mail. So as I see here, I wrote you give the basket. So it's not replying to them to demand, but to their to their wives or girlfriends on. Okay, I don't want Oh, this time. Now here. That will be the example for the first ad set. And now we're targeting there. So we're going to this example to be option where we want to Ah, target the different as creative to the female audience. So, as you can see here, Budget and these things will say the same sold period is equal. Money is equal, but we're taking a look a different audience. Now we will stay to the United States. That's that's OK. But instead of a man who will target women and it's that off will target women from 21 to, let's say, in this case to 35. Okay. And when it comes to targeting. We can just put makeup. So we'll get quiz medics on the beauty's skin care, beautiful skin and so on. So we're targeting now the females female and the placements. I will leave it to automatic one. Since I want to compare, I don't want to change placement right now. I want to compare different audience on the different at creatives. So optimization will also stay the same. That's OK. And when we come to this one, so will have the option to select an image. And instead of using the same one, which does make sense, I would use this one, which should make sense. Gam. Sorry about that. It's the loading on. Uh, yeah, this is a different one. So now we're not talking about present on, uh, actually says it's not so Maybe I could put it like this. This we're not talking about present. Maybe it could be a different. Okay, this seems this is good. So I can right here something like Okay, the website will stay the same makeup. And maybe here I could write Surprise yourself for, uh, for this special day. Let's say let's say in this case it would be Christmas because we're testing during the Christmas period. It can also be for the Valentine's Day and, yeah, the the other things are most of the same self as you can see here. Just a quick conclusion what we did. We have to assets which to have creative when that sets ISS for men. And it has ad that shows the present the gift that man can give Teoh their women were girlfriends as a present for the future Valentine Days that, well, it is a and the other option we have on that said for the women audience, where actually we're showing up makeup that single women can buy for themselves as surprised as a gift uring the future Valentine's Day. So we use different creatives and we use different audience. So we used to variables on. That's after this five days or four days on the whole budget. We want to see what is actually performing better, and we're doing this. Remember during the Christmas holiday. So we assume that the same things would happen during the Valentine's Day. Approximately the factors would be the same. So by the Valentine's Day, February 14 we would know what will be better for us. So on that day before they're actually few days before that, they we could place arrives either to the male side or to the female side. So that will be Thank you for watching and seeing the next lecture. 10. Case study #4 Google Ads (demographics and ad variation tool): welcome back and we're taking a look at the case Study number four, where we're going to set up a split test for Google ads platform. And as you can see here, our task is we want does the same car ahead search campaign type with different audience to see whether it's performance better with younger or older audience who would engage more and who is cheaper to target. Also, we want to use add variation toe to switch the headlines in er care car at. So we'll start first off with creating the campaign, and you can come to do Google. Let's to campaign tab and click on new campaign and for this purpose that say that we want to drive people toward website where they can they can engage. And I can I for this, for the sake of this video shoes in the search campaign type here, and I would call it split adds campaign. I only want to publish it in this case for Search Network. I'm gonna run through this option because I don't want to spend here time. We covered this in the previous lectures and courses so you can check up there here I just want to make sure that I'm setting everything, probably so I can show you later on how how it works. When it comes to this protesting, because here is a bit different. It's not like it's that, Ah, it's not function function of allies. It is on Facebook, for you have just one bottom that everything else is the same here. You need to do a bit manually, some things, but it's not so hard. And, uh, here I would just put the audience. Let's say that it will be something about cars. Okay, that's let's put these things out to be a ghost. That's okay. Just budget. Let's put it to be 50 euros. I'm sorry. And the bidding conversions. That's okay. I would leave this body fold, and I won't touch anything here because currently, on the loose time there add Group One. Actually, I think I forgot. Yes, that's that's okay. One would be. So let's leave it to car Group one on. Let's just put some since it would be that's love your ho, I think. Let's just put it to be the slot cars electric course on. Okay, so let's see, What what we have Here is key words. Ideas. Okay, I prepared the image for the next case study with the test car. So that's why I'm actually he choosing this one. This one is an example on the Yeah, I would just put this quotation mark house for now. I don't want to go, so Okay, anyway, that would be the ad group on Finally have add to create. And now I want to put my final you are would be come here. I can put a headline to to be longest range electric cars. I can put it to be here, uh, save the apartment and, uh, I can, but here to be by now. So in short, this would be my Here, you can see how it would look like. So when somebody type some of those keywords that we saw, this would pop up as let's say that we are doing marketing strategy for Tesla cars and here you would see along this current range electric car said the environment by now on Yeah, most of that will be it. I would just save him. Continue. And now I Okay, sorry. I love this place. So best electric cars on the market. Check. Oh, so I will go back to the campaign. Don't worry if I if I forgot something, or I saw that there was a mistake with some extensions saying some other words. I don't worry about that. I need to change that in the campaign group. So that's that's not so relevant right now. And, uh, what I actually want to do here now is to go to add groups and in order to make a split at , I would just carpet this and in the same place basted. I want to check two balls new ad groups after fasting, because May off course, I want to make some changes, so I don't want it to run right away anyway, Now you can see that we that I will have to add groups number two, and what I can do now is, since this one is paused, I can click on her. Actually, I can make on this group and since there are under the same campaign, so remember that structure level? This is the name of the campaigns, but as campaign and within it we have two groups that are currently the same showing the same keywords and everything. But where I can do is come to, let's say, demographics. And, uh although we didn't check anything while we were making here, you can see that, For example, we said in the case study here that we won't test one that group to be for the younger audience and the other one to be for the old older audience because we're still not sure which which the graphics you want to talk. So legislative carload number one to be for the younger ones. So I will. I will excludes for the Advil number two and all the people who are under under 34. And for the cargo number one, which is younger audience, I would exclude the elderly one. Okay, so just to be clear right now, what we did is that for the car group number one. So first group, we put the age to be for mating to 24 from 25 to 34. And for the Carvel number two, that is Admiral number two. We put it to be the older population that is from 35 to 44 or 45 to 54 from 55 to 64. So the second group will be shown to the older population. The phrase group will be shown to the younger population. Everything else remains the same. Sochi murders and everything will just have to groups on the budget will be split among them. On we will see where we can make a better, better return. As you can see here I have audiences which is the same for both groups. So I cannot check that and same thing with the keywords. So I cannot check that actually here with the key words I can with the audience. I cannot and you can see here for the second group, I can exclude some keywords. I just leave more keywords for the first group. If I saw that these outer terms that younger population is typing Maurren. These are the key words that the overpopulation is typing more so you can you can play with that. There could be as many as you wish combinations. Anyway, The important thing here is to know that if you want to make different interests for the audience, you would have to make two campaign level, so you would have to go even higher. And that was the first part, If you recall. So here we're looking at the 4th 1 So we checked the younger and older audience and now we contest that to see who will engage more and who is cheaper to target. Next one, we want to use the same ever do to use the add variation toe to switch the headlights in our car at. So what we with what we're going to do is I prepared here that that Anyway, on the sidebar just usually would sit like this, scrolled out, go to drive and experiments. And when you get here, you will usually be on the campaign dress that instead of their go to add variations and click on this plus and you will get this, you should select the campaign, which is split, adds campaign done. And what you can do here is basically three options so you can find underplay certain keywords in different in the ad and make a creation that you can update text. So if you want to add something more to the next version two due to control the to test the variation and you can swap headline one and two. And that is exactly what I want to do. So I will press you continue on and variation. And what I can do is that that for or several days I don't want a test for one month, Okay? And I can create the variation here. And what I basically have here is, if I click on that, I will see my results here. So I will have a split at. You can see that here. So, um, just seconds. Okay. So what? I actually did this since I had to. Car groups. And I don't need this to do this variation to do for the boat groups, so I can I can do it for the both groups, but I can do it only for the one here. You can see the original ed. It was longest range electric car saved the environment by now. And instead of that, I want to switch the first and the second handle headline. And I want to say, save the environment, longest range electric cars by now. And I want to do that with both that group. So I want this modified add to be shown to the younger and older population. And maybe this seems like a no beak deal to you, like switching headlines. Who would care? But believe me that sometimes that could have a really, really powerful in fact, especially depending on the keywords that users are are typing. So that would be it. So just before I finish, I would like to provide you with fast summary what we did here. So we made a search campaign. That's nothing new. We usually that we did that before. In that search campaign, we made a duplicate add group. When that duplicate adds group, we changed the demographics. That is, for the first group. We left it to be for the younger audience for the second ad group. We liked it to be for the older audience. We came to this variation toe, and what we did here is actually made a variation for both. As for both ad groups, actually, where we switched the headlines off course, if I don't want to do that for the boat as I can modify that, but in this case I want to do it for the boat ads and I want to see. So besides whether my ads are better for the younger or older audience. I want also to see whether it is better to first say, save the environment along the strange electric cars or first say longer, strange electric cars and then save the environment. So that will be it. I can track my resource for the modified at here and resource for the original ads. Actually, two different demographics I can. I can follow that in the section off demographics and campaigns and ads groups in general. We do that. So thank you for watching and seeing the next lecture. 11. Case study #5 Google Ads (test different parameters): hi, everyone, and we're taking a look at our fifth case study, which says that we want it does the same Colorado which we saw in the previous example. But this time with this place campaign, that and we want to see how we can do a split test for different factors such as the audience demographics. And so so this will be a bit different because we're going to take a look at different campaign objective and different type of fat on. Then that way we will be able to see how you can create Split. That's if you decide to promote your ad differently than actually not to use search campaign, but the other one. So let's say that we're using brand awareness reach. We want to as we set promote. That's like our so in Tesla cars. And, uh, we want to promote them to read toe raise awareness. Although since it's a bit luxury print, I doubt that they would use banners on Google for their promotion. But anyway, let's just use its artistic, this video, uh, campaign cars, let's put sell it. Actually, all countries just put United States since we did that in the previous examples example on that would be regarded again. I would just go through these things because we cover them in the previous courses of lectures and I don't want to lose a lot of time here. 50 year olds would be okay. Delivery method. I would just leave it for the sport and other things. Now I don't want to go so deep regarding Teoh and date Just put it before for actually, for 11 that seems no, you can put it Plus, let's put it three weeks approximately. And the additional things I'm not so much ad group. So that would be like car at Group number one, OK, on again we would leave interest to be the ones that we already shows. So actually not used morals, just you. And let's see some I d years luxury vehicles. Yeah, we could say those are actually variable x review. Close in a way. So yeah, sure, Why not also? Yeah, let me just see whether there's something for the since I want to find thriller Electric Be , of course, whether there's a yeah, sure when I left cars hybrid cars also over this, actually Tesla's pure electric any weight. That will be it regarding the audience interest. So let's put it here to be actually literally this to be by default everything except actually this one. We don't want to lower the alleges. Uh, yeah. Okay, that will be it. And, uh, but this that group, I would just leave it. I get Let's put it to 10. So this is a viewable CPM bit for this saddle that I want to put, I suppose is the highest one. The Brits are here would just electorally Okay, changer. Exaggerated. Ok, OK, OK, I kept, but actually, I can put 50 here out on my Yeah, I can't really want that reach to be powerful. Anyway, that is the impression I want to upload display ad, and I'm going to choose a photo. So it's just small band red and website would be this one also. I'm sorry if I made any mistakes above. I'm really not focusing about right now. I'm trying to make this process fast. It's possible so we can focus on the split testing. Yeah, that would be this would be a banner size. Keep in mind when you upload the here, there must be some, uh, guidelines that you have to meet quality and size, so that will be it. I could make several leads here so they can choose to adopt toe just different different banners to different places and so on. But for the sake of this video, just using one. So what we did here is made one, that group inside one campaign and we're going to double that. Of course we want to do, sport. You see here we don't have a bottom actually to do this, at least not yet. I mean, Google is making some progress on this. Until recently, they didn't even have that variation to that we saw in the previous example. So maybe in future or maybe the new versions, Because I still haven't updated mind. There is something more that you can do with split testing, but for now, this seems as the as the best option. So yeah, that will be That will be it. Um And what I can do now is actually come to demographics. And if you recall for the previous examples, we can take a look at their here at different. Sorry, I forgot to change this one to be number two, not number one and then hashtag number two. Okay, now it will be easier to understand. Okay, We don't have any keywords. I hope that you know. Why is that? Because we used this plate display campaign type. When we didn't put any key words, it was necessary to do so. But here we have different audience. And as you can see here now, audience is not connected to the campaign. It's connected to the ad group so we can change that. So I can exclude some audiences here for the second group or Adam and so on so I can play with these things. Moreover, I can do the same thing for for demographics. You can see here you saw what we did in the previous examples. I didn't put any placements on topics, but I could also play with that even when it comes to their as extensions. Let's just see here I have for the two of groups have the same same banner. But I could change that. I could put here to be Tiatto this some other one. And in that way I could actually test which one is performing better. So That's the whole point of this off this thing. And, uh, yeah, that was actually what I wanted to show you. If you want to change any any, any setting inside of this campaign, I didn't show you that in the previous one, but you can do that here regarding the time budget location and so on. But here was the main focus. So just to get back here, we wanted to has the same carod which we did with this place campaign time, which we also did and see How can we do a split us for different factors such as audience demographics at extensions and so on? And we saw that here So quick summary of what we did We made campaign on display campaign type. We we choose our factors are are meters and we duplicated the first group. We renamed it the 2nd 1 And although we didn't change anything here because I'm trying to say, I'm trying to save you time but we can change. Those factors were we can make different tests on this. That group can run something decide room can run something else, and then we can compare which metrics are performing better. Which audiences performing better platforms as an extensions. Landing pages lame. So you can you can play with that. So that was it. I hope that you understand this. It's a bit addressed to and these methods and techniques in general, and I hope that they were helpful to you. Thank you for watching. If you have any questions, let me know. Thank you once again, and it's beautiful videos. 12. Thank you: Hi, everyone. Thank you for taking the scores. I really hope that you find it to be helpful. Please. If you have any suggestions, steps, advice on how I can improve my courses, let me know. And also, if you have any suggestions in which topics I should cover, please limit down below in the community tab or related within the review. Thank you once again and I hope to see in the following blesses