Everything we do in Marketing is a test. We may be confident in what we promote because we have historical and current data that helps us understand what could be successful but positive ad performance is still never a guarantee.
When it comes to testing on LinkedIn Ads, just like any other digital ad platform, we can either throw spaghetti at the wall and hope something sticks or we can test systematically (hint: the latter is the better option).
So how should A/B tests look on LinkedIn Ads? You came to the right blog post. Let’s hit it!
What Does a Proper A/B Test Look Like on LinkedIn Ads?
When we talk about A/B testing on LinkedIn Ads, we’re talking about testing between two different variables in your ads.
This could be a test between two variations of intros, headlines, images, etc, and leaving all else in the ad the same. This results in a true test because you can directly attribute an ad’s success to one of your differing variables.
For example, say you’re conducting a test between two ads. Both ads have different intros, headlines, and images. If one outperforms the other, it’s hard to conclude which variable (intro, headline, or image) led to success.
On the other hand, if you conduct a test between two ads but the only difference between them is the intro text, all else remaining the same, then if one outperforms the other, you can have confidence that the intro text is what led to the success of your ad.
Note that you don’t necessarily have this luxury when A/B testing different offers. When testing offers, your imagery, intro text, and headlines are likely going to be different, but finding which offer is going to outperform the other is still a very worthwhile test.
Once you’ve found an offer that works and generates the desired results, you can optimize for better performance by A/B testing other variables. If you need help deciding which offers to start with, here’s our guide on which offers tend to work well on LinkedIn Ads.
So with all this being said, does it mean you can only be limited to running two ads at a time in each campaign? Not necessarily.
You can increase the Frequency (the average number of times your ad is delivered to a single member of your target audience) of your ads to up to five times in a 48 hour period if you choose to run at least five different ads in a single LinkedIn Ads campaign.
However, Frequency caps at five in 48 hours, so running any more than five ads in a given campaign won’t increase your Frequency beyond that.
That said, when it comes to A/B testing, you may not want to run that many unique ads anyway. What we’ve found works well is running up to four ads in a given campaign but still only testing two ad variations.
What you’re doing here is essentially testing between two ad variations but duplicating each ad one time, resulting in four ads total in a campaign. This allows you to still A/B test between one variable while increasing your ad Frequency at the same time.
Measuring Results and Concluding Your Test
When it comes to measuring the results of your A/B tests, there are several metrics you can compare. The most relevant metrics are likely going to be your click-through-rate (CTR) and conversion rate (CvR).
Many of the variables you test at the ad level are likely to affect your CTR. Variables like your intro text, headline, and imagery, for example, may influence whether or not someone clicks on your ad.
Variables like the offer you’re promoting or your landing page experience are likely to affect your CvR.
So when analyzing ad performance, depending on the variables you’re testing, these are the metrics you can measure for success. To know where your ad performance stands, check out this post for a list of LinkedIn Ads benchmarks.
You might also be wondering, how long do I conduct my test or how do I know when to conclude? To answer this question, it’s important to understand statistical significance.
In layman’s terms and as it relates to LinkedIn Ads, statistical significance is a mathematical approach to determining whether or not you can say with confidence that one ad is outperforming the other.
To illustrate this, say the ads your A/B testing have about 10,000 impressions each. If one ad has 500 clicks and the other ad has 5,000, you can probably say confidently that the ad with 5,000 clicks is the clear winner.
However, if you only have 100 impressions per ad and one ad has 10 clicks and the other has 50, you may see a clear winner here, but you’re also analyzing a substantially small data set. The next 100 impressions could drastically change the results of your test, in this case.
So when analyzing the statistical significance between two ads, it’s important to make sure your data set is large enough to confidently conclude a winner.
Going back to our question, “how do you know when to conclude your test?”, the short answer is, until you have enough data to where you can confidently identify statistical significance between your two ad variations.
Statistical significance is not an easy topic to explain or understand. Thankfully, this podcast episode on LinkedIn Ads testing methodology explains it really well and sheds even more light on how to know when to conclude your LinkedIn Ads tests. Check it out!
How to Decide What to Test
Your offer is the most important aspect of your LinkedIn Ads. If your offer is good, you’ll generate leads at low costs. If not, lead volume will be low and costs will be high.
So, if you have multiple offers, start by testing between those first. As mentioned earlier, once you find an offer that works well, then start testing other ad variables in order to optimize performance.
If you’ve decided on an offer, the next important aspect of your LinkedIn Ads is the intro text. People tend to log in to LinkedIn with a specific purpose in mind, not for leisure. That means we as advertisers only have a limited amount of time to capture their attention.
Your messaging needs to be both captivating and compelling in order to see LinkedIn Ads success, so test intros to find the one that’s going to bring you the best results.
Following these two variables, you really have the freedom to test whatever you’re interested in testing after that. Testing offers and intros first is the most efficient way that we’ve found to identify early on what our audience is most interested in and then optimize from there.
Test, Test, Test!
Rather than going the “spray and pray” approach, systematic testing on LinkedIn Ads is a sure-fire way to find ads and offers that work. Rarely do we ever get ads right the first time. It takes patience and constant testing until we find the right combination of variables that result in LinkedIn Ad success.
What tests have you tried? What have you found to be most successful? What hasn’t been successful? We want to hear from you, so feel free to leave a comment below!
Also, if you can’t already tell, we really dig this stuff. 😉 In the 11 years we’ve run LinkedIn Ads, we’ve spent $150M+ on the platform, are official LinkedIn Marketing Partners, and have managed some of the largest LinkedIn Ad accounts in the world.
B2Linked increases your lead quality while lowering costs at the same time. Say goodbye to wasted ad spend! If you want to ramp up your LinkedIn Ad efforts, apply to work with our team of experts.
Thanks for reading and happy advertising!
Written by Eric Jones
Excellent blog as ever! Sooooo good!
If I create two sets of identical ads to do a a/b test, doesn’t the LinkedIn Ads platform know that 2 ads in the 4 are completely identified?
Thanks, Mark! 🙂
And great question! LinkedIn does not recognize if two or more of the same ad are in a single campaign. You’re totally in the clear to test this way!
One thing that lots of advertisers find annoying about linkedin ads is that a true a/b test isn’t possible since the audience is not split into random groups and only shown one of the versions of the ad, they are seeing all versions and we have to assume that the one they clicked on is the ‘winning’ version. In actual fact the influence of seeing the ad a few times although in a differing variation might be the real reason the person clicked the ad that time.
Hey Ben,
You’re absolutely right! This A/B testing solution isn’t completely accurate. In fact, LinkedIn announced a few years ago that a more accurate A/B testing feature would be coming to the platform, but it may still be a while before we receive it.
Even though this isn’t a perfect solution, because LinkedIn tends to favor ads that receive a higher percentage of engagement, if you find that one ad is generating significantly more impressions, clicks, and a higher CTR than another ad, you may be able to still confidently conclude that one is outperforming the other.
Hope this helps!
Hey Erik! Thanks for this great article. You said, your testing four ads in a campaign. Do you combine this method with splitting the targeting groups in different campaigns, like recommended in a podcast episode from you?
Hi Lisa,
You bet! Thanks for reading 🙂 And yes! We like to A/B test the same ads across all campaign/audience segments. This allows us to both test ad variations and audience segments at the same time (since all audiences are receiving the same exact test). So, if you’re targeting HR professionals, for example, your campaign and ad setup might look like this:
HR-related Job Title Targeting
Ad A
Ad B
Ad A
Ad B
HR-related Skills Targeting
Ad A
Ad B
Ad A
Ad B
HR Job Function Targeting
Ad A
Ad B
Ad A
Ad B
HR-related Groups Targeting
Ad A
Ad B
Ad A
Ad B
Hope this helps!
Thank you 🙂 but is this a/b testing “clean enough” since there will be overlaps? So for example people you target with the job title HR Manager will also be reached if you target people with the job function HR. And what would be the minimum of budget per month for each campaign to make sure the results are significant?
Hi Lisa,
You bet! 🙂 And very good questions.
You’re right that this A/B testing solution isn’t going to be perfect. There will definitely be overlaps. That said, LinkedIn tends to favor ads that receive a higher percentage of engagement. So if you find that one ad is generating a significantly higher CTR or conversion rate than another ad, you can still confidently conclude that one is outperforming the other.
As far as how much budget to allocate to each campaign, I think it’s also worth considering that time plays a factor in determining whether results are statistically significant. You can have smaller budgets allocated to each campaign, but you’ll need to run ads for a longer period of time before concluding which ad(s) are performing best.
In general, we find that we get statistical significance to the CTR after spending $1k in total budget and to the conversion rate after spending $5k in total budget. Though, this is a general rule. You could totally get it sooner or later than this.
Hope this helps!