3 A/B split tests to improve your LinkedIn ad performance

9 min read

3 A/B split tests to improve your LinkedIn ad performance

When the phrase “online advertising” is mentioned, paid social media posts aren’t usually the first thing to come to mind. While paid search is great for capturing users with buying intent (feeding the bottom of your sales funnel), social media advertising is more effective in reaching new audiences at the top of your sales funnel.

At Brand chemistry, we’ve seen LinkedIn Sponsored Post campaigns yield click-through rates as high as 2.2%, performing well above the industry benchmark of 0.3%. Although it’s quite common to achieve good results during your first few tries—there is such a thing as beginner’s luck or rather, random chance—maintaining this level of engagement poses a challenge for marketers.

And that’s where A/B split testing comes in.

What is A/B split testing?

A/B split testing is a tactic used in marketing to determine the performance of opposing elements in your marketing efforts. In the case of LinkedIn Sponsored Posts, split testing allows you to compare two versions of your ad or campaign, with only one differing element between them, to determine which variation of this element performs better. This variable could be a different image or body copy, or a different way of selecting your target audience for the campaigns.

By measuring the performance of your split test ad or campaign and consequently, the impact of the varied element, you can determine which version will return a higher ROI for future campaigns. Do keep in mind that the statistical difference must be significant enough to rule out the probability of random chance. We find that it usually takes more than three rounds of the same test to be conclusive.

I see! So… what can we test?

Before you start testing, we need to stress that you should only ever test ONE element at a time. This means that both your campaigns should be identical except for the one variable that you’re testing. If you’re altering two elements, it’s impossible to know which made a difference.

1. Copy style
(aka how to clickbait your audience without being obvious)

Ad copy is one of the first elements you should test. The aim is to find out what copy style triggers a desirable action from your audience. Are they tempted by questions? Are they moved by empathy? Are they provoked by a fear of missing out? Do they react to clickbait?

At Bc, ad copy is one of the first tests we run on our LinkedIn ad campaigns. Each ad campaign usually contains four types of copy style: question-led, benefit-led, empathy-led, and statistic-led.

1. Question-led copy starts with a question and ends with a call-to-action phrase. The question addresses your target audience’s pain point, and the call-to-action that follows prompts them to go a step further to obtain the solution for their pain point.

For example: Confused by the different marketing automation options out there? Download this guide to understand how to choose the best platform for your business in 10 easy steps.

2. Benefit-led copy starts by stating the value your target audience will get from the content piece, before ending with a call-to-action phrase. The value is the solution your potential customer is looking for, and the call-to-action that follows should come across as the next logical step they have to take to solve their pain point.

For example: Learn how to choose the best marketing automation platform for your business in 10 easy steps. Download the guide now.

3. Empathy-led copy starts by demonstrating to your target audience that you understand what they’re going through and ends with a call-to-action phrase.

For example: With so many marketing automation options out there, it can be a little overwhelming to figure out which you should invest in. But don’t despair! We’ve done the hard work for you with this guide. Download it now.

4. Statistic-led copy starts by stating a significant statistical figure that will trigger, move or persuade your target audience into action, which is then further prompted by the call-to-action phrase at the end.

For example: 85% of marketers who onboarded a new marketing automation platform AFTER they’ve done their research are more satisfied with their investment than those who didn’t. But who has time to demo the plethora of platforms available? We can help. Download this free guide to choose the best platform for your business in 10 easy steps.

At Bc, we have done split tests with a combination of some or all copy types and found that engagement varies between different styles for different audience selections.

The results of a LinkedIn ad campaign we ran for one of our campaigns at Bc recently showed that question-led copy performed better than the empathy-led copy in a campaign targeting marketing managers in Australia. The question style received twice the number of clicks (9 vs 4).

LinkedIn ad copy test: Question led copy wins

However, testing shows that preferences differ from audience to audience. Let’s go through a test we ran for one of our clients. Using the same two copy types as the above ads, this campaign was targeted at business owners and MDs of SMEs in Australia and New Zealand. As you can see from the screenshot below, the empathy-led copy performed better than the question-led copy.

LinkedIn ad copy test: Empathy led copy wins

The lesson? Different audience groups react differently to different copy styles, therefore it’s important to have a few ad variations in your campaigns to cater for varying tastes and to find out which variation works best.

Bc takeaway: Remember to run multiple tests with the same opposing element until you get sufficient and meaningful data to draw a conclusion.

2. Audience selection
(aka what’s the best way to reach your potential customer)

LinkedIn has a comprehensive range of targeting options for you to reach your target audience. You can filter them by job functions, job titles, groups, member skills, company size and industries, to name a few. If you’ve been using LinkedIn ads as part of your marketing strategy, you might have found that different combinations of filters result in different audience sizes for your campaign, even when you’re filtering to reach the same audience type.

For example, if you’re targeting marketing directors in Australia and using the geography and job title filters, you’ll get about 6,300 people in the pool. However, if you select this same audience using the geography, job function and job seniority filters, you’ll get about 28,000 people. We recommend using job titles with caution as some LinkedIn users will change their job title to get your attention and stand out. Writers or content producers could be flying under the radar as Word Wizards. Human Resources Directors could identify as Chief Happiness Officers. We suggest trying different variations to find your ad-target sweet spot.

Audience segmentation by job titlesAudience segmentation by job titles

Audience segmentation by job function and seniorityAudience segmentation by job function and seniority

At Bc, we have had excellent results with different combinations for various clients and industries. The screenshot below shows the result of one recent ad campaign we ran for a services client.

We ran two LinkedIn Sponsored Post campaigns with the goal to drive traffic to a BOFU offer landing page on their website. Both campaigns contained the same ad variations and were targeted at the same audience, but one of the audiences was selected using the geography and job title filters (JT), and the other was selected using the geography, job function and seniority filters (JF). 

As you can see below, the geography, job function and seniority filter combination performed better than the geography and job titles filter combination, with the former generating more than twice the number of clicks and click-through rate compared to the latter.

linkedin ads split test - audience segmentation results

Although it is easy to draw conclusions after a few tests, it’s important to stay agile and continue testing the conclusion between long periods of time (Bc recommends twice a year) to keep up with behavioural changes in your target audience pool.

Bc takeaway: Run campaigns with different audience segmentations respectively in different time frames to avoid audience fatigue and data corruption because it is likely that your audience pools will overlap.

3. Bid strategy
(aka CPC or CPM)

If you’ve meddled with social advertising before, you’d be aware of the two bidding strategies:

CPC: is cost-per-click, where you only get charged when someone clicks on your ad.

CPM: is cost-per-thousand impressions, where you only get charged every time your ad is shown 1000 times.

It is often misunderstood that the bid amount you enter determines the cost of your ads. This isn't entirely true because what you’re really doing is bidding against your competitors to reach the same target audience. The amount you end up paying per-click or per-thousand may or may not coincide with your initial bid amount.

At Bc, we have tested both bidding strategies for various clients and again, different bidding strategies have performed differently for different campaigns. 

The screenshot below shows the results of the bid strategy split test we ran for one of our clients in the executive coaching sector. Both campaigns have the same ad variations and target audience selected, but one of the campaigns was executed with the CPC bid strategy and the other with CPM. As you can see below, the CPC strategy worked better for this campaign, resulting in a higher number of clicks, engagement and consequently, a lower cost-per-click.

Image 6: CPC performed better than CPM

On the other side of the coin, we have also had successes with CPM bid strategy campaigns that brought in a comparable number of clicks and impressions, and also resulted in a low cost-per-click.

Image 6: CPM can perform well too

Bc takeaway: Bc recommends running ads using both bid strategies to achieve the best ROI for your industry and audience. 

Before you go...

Although it can be easy to draw conclusions and apply the same deduction to every ad campaign and audience group, DON’T. Little variations can have big impacts on your results. So keep testing and redefining your conclusions, and always remember to only have one differing element in your ad campaigns to achieve clear and conclusive results. Happy testing!

Social media advertising is part of a wider distribution strategy in an inbound marketing plan. Download our quick-start guide to learn more about inbound marketing and how to get your fantastic content in front of the right audience.

Originally published on 22 March 2017. Last updated on 1 July 2020.

Download Brand chemistry's quick-start guide to inbound marketing for technology companies

Brand chemistry creates brilliant content to get your brand in front of the right people. With lifecycle strategy, whip-smart word wrangling and striking visuals, our content marketing gets you results.

Get The Alchemist in your inbox weekly

Get your guide today!

Recent Posts

Latest Tweets