from the

LINKEDIN MARKETING VAULT

 O PTIMI ZE. I TE RATE . SEE R O I.
So, you want your content to resonate with your audience...

A/B testing is like putting together a puzzle: Oftentimes, tests reveal that the smallest
you need to test and iterate until you optimize tweaks can make the greatest performance ‘Failing forward’ = getting ahead by
every component of your campaign, including impact. We find that our best campaign not being afraid to fail. Fail in an effort
captions, images, CTAs, etc. Another popular strategies are gut-led but data informed. to improve. Fail for the sake of being
analogy is the ‘Goldilocks Effect.’ Is your However your gut might not always be right
proactive rather than reactive.
campaign too hot, too cold, or just right? You and while you may be confident in your
won’t know until you taste test each bowl hypothesis, testing may prove you wrong. It’s
of porridge. And this is the process for each happened more than once to us!
and every campaign we run at LinkedIn Read on to uncover testing principles & tips,
Marketing Solutions. In a world where brands like Amazon and a super quick crash course in using LinkedIn
Netflix are leading by providing excellent Direct Sponsored Content, tests we’ve run here
customer experiences, B2B marketers can’t at LinkedIn Marketing Solutions and more.
afford NOT to A/B test. Consumer expectations Hopefully this will spark some ideas of tests
for personalization and excellent content you’d like to run with your own team of B2B
According to Aimia Institute, marketers, or should I be so bold as to call us
millennials are 44% more experiences have changed. If you build a
testing culture and aren’t afraid of ‘failing modern data scientists?
likely to permanently
forward’, you’ll always be in a great position
disengage with brands after
to create great consumer experiences. Alex Rynne
receiving high volumes Content Marketing Manager
Furthermore, the more you A/B test, the more
of mass generic email relevant your content will become to your
LinkedIn Marketing Solutions

communications. prospects and customers.

2 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va u l t

 WH AT ’ S I N T H I S
G U I D E , YOU AS K?

First Thing’s First:

3 Testing Principles

Our 3 Finest
Testing Tips

LinkedIn Direct
Sponsored Content 101

5 Types of Tests &

What We’ve Learned

Testing Ideas We’re

Keen to Try Next
 Can You Spot
the 5 Differences?

1: Top left graph fi ll o n n o tepa d, 2 : To p l eft ke y boa rd fun c t i o n ke y s , 3 : To p r i g h t m o us e , 4 : Pe n c i l i n p e n c i l j ar, 5 : Mi d dl e r ig ht co ffee c u p ha n d l e

F IRST TH I N G’ S FI RST: 3 T EST IN G P R IN C IP L ES

Before You Start Testing In a survey,

These three principles apply to all A/B testing on LinkedIn. They will help make sure you
generate meaningful data for optimization. 41% of overall
respondents
1. Plan to Run Tests for at Least Three Weeks If you want to test several variables, you can
For your test results to be truly meaningful— run multiple tests simultaneously – one that say they would
that is, good indicators of how to improve tests the headline, one that tests the visual, consider ending
performance in the future—it’s important to get one that tests targeting. But in each case make
a large sample set. Three weeks is the minimum sure you have one “control group” and one a brand
amount of time we recommend. With that much with just a single variable.
relationship
data in hand, you won’t be misled by outliers.
For example, if your three-day campaign 3. Keep Bid Rates the Same When because of
included a national holiday, that atypical data
would heavily skew your results. A longer run
Testing Targeting
When trying to determine which audience
irrelevant
time smooths out those bumps. will be the most responsive, you can A/B test promotions, and
targeting different groups. Just make sure to
2. Only Test One Variable at a Time keep all other aspects of the campaign the an additional
It was good advice in your high school same, save for the change in targeting. That 22% say they
chemistry class, and it’s good advice in includes the bid rate—there may be variations
marketing. Your test campaigns should be in bid prices due to audience and region, but would definitely
identical save one single aspect you’re testing. try to find a middle ground for your bid.
defect from
For example, if you changed the headline and
tried a different visual, it would be impossible to the brand.”
tell which change made a difference.

5 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

Source: https://www.sparkreaction.com/blog/
if-your-content-marketing-strategy-is-not-rele-
vant-it-does-not-matter
O U R 3 F I N E ST TE STI N G T IPS

Tips For Testing

The following best practices emerged over time as our team learned what was most effective for our own marketing on LinkedIn.

1. Run Multiple Creatives 2. Set Campaigns to “Rotate Variations 3. Create Unique Tracking Parameters
When running a LinkedIn Sponsored Content Evenly” for Testing Make sure there is a way to easily attribute
campaign, you can gain more exposure and You can choose from one of two options when traffic from each of your campaign variations.
more ability to optimize by testing multiple running multiple campaigns: Either have Using unique trackable URLS to your landing
variations on your creative at once. Our team LinkedIn’s software optimize automatically, page enables you to determine which variation
runs at least four for any major campaign. After making your top performers appear more is more effective at driving conversions.
some lead time, you will be able to optimize often, or have LinkedIn serve each variation
based on performance. Shift spending to the evenly without weighting for performance.
best performers, turn off the lowest performers While the automated option can serve as an
and replace them with new challengers. excellent time saver, to get more meaningful
data, select the second option.

Rotate ad variations

Optimize click-through rate - better performing variations appear more often

Rotate variations evenly

Cancel Save

6 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

LINK EDI N D I RE CT S PON S O R ED CO N T EN T 1 0 1

First, what’s the difference between LinkedIn Sponsored Content

and Direct Sponsored Content?

LinkedIn Sponsored Content allows you to

publish relevant content and reach a targeted
audience of professionals beyond just your
Personalize your
LinkedIn Company Page followers.
message to target
Direct Sponsored Content is a feature which each audience
allows you to share content directly in the
feed, giving you the ability to personalize and
test content without having to originate posts
on your LinkedIn Company Page. Make your
content more relevant by sending personalized Test variations
messages to specific audiences. Then test and of your content
retest a variety of content in real-time to
optimize performance.

Got it? Good. Moving on to the good stuff, the

tests and the data…
Control what gets
published on your
Company Page

7 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

5 TYPES OF T E STS AN D W HAT W E’ VE L EA R N ED
1. Image tests
2. Test variations in copy
3. Character count test
4. Test creative placement to
draw a reader’s eye
5. The kitchen sink test

1. Image Tests
I’m sure you’ve heard us say it before and we’ll
say it again—visual is the new headline. When
we’ve added more visual interest to our content,
we’ve gotten more attention.

Every LinkedIn Sponsored Content post is Rich media images have 38% higher
CTR than linkshare updates with a
accompanied by a photo (1200x627 pixels will thumbnail image preview.
render best on the LinkedIn platform). We
also try to sprinkle some eye-catching visuals
throughout, featuring short stats and quotes
lifted from larger pieces of content.

When we tested rich media versus a non-rich Image Testing Tip

media post, we found time and time again that
To ensure the text on your image isn’t
rich media performs better. In fact, CTR for rich
cut off, make sure it’s within the text
media can be up to 4x the CTR of updates with
safe area, which is 1000x586.
text alone.

8 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

5 TYPES OF T E STS AN D W HAT W E’ VE L EA R N ED
Our team typically creates 3-4 different
Sponsored Content images to test against each
other within each campaign.

For example, we pulled these 3

stats out of The Sophisticated
Guide to Content Marketing.

9 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

5 TYPES OF T E STS AN D W HAT W E’ VE L EA R N ED
Image Test #1
An object versus a human.
This test featuring an image of a person
versus an object (keyboard and smartphone)
proved that the photo of a person generated
significantly better results.

The image of the human saw +160% CTR

and +289% CVR.

Image Test #2
Background color.
What is it about woodgrain? Does it feel more
authentic? Does it just provide a little visual
interest? Either way, the post with the darker
background saw 32% more clicks than the plain
white background on the right.

This image saw a .332% CTA + 32% Clicks.

10 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

5 TYPES OF T E STS AN D W HAT W E’ VE L EA R N ED

2. Test Variations in Copy

Copy Variation Test #1
CTA Test
What a difference a single word can make! In the
following example we tested the CTA “download
the guide” vs. “download the eBook.” The
version on the left with “guide” saw a 95%
increase in CTR and 50% more impressions.

The post with the word ‘guide’ rather than ‘ebook’

saw a +95% increase in CTA and +50% impressions.

Copy Variation Test #2

Incorporating a Stat
When we tested an update with a statistic and
without, we saw a 37% higher CTR and 162%
more impressions for the former.

The image with the stat saw +37% CTR and

162% more impressions.

11 > 5 Tes t i ng Ideas Fro m th e L I n kedI n M a r keti n g Vaul t

5 TYPES OF T E STS AN D W HAT W E’ VE L EA R N ED

3. Character Count Test

It might seem strange that character count would
matter when the text is as short as this. Surely an
audience can read 140 characters as easily as 70? But
we saw an 18% boost in engagement just by getting
to the point a bit more quickly.

Insider Tip
When we tested the impact of shorter character length, we saw more Keep it short and sweet. Captions should be
than an 18% boost in engagement rate. 150 characters or fewer.

12 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

5 TYPES OF T E STS AN D W HAT W E’ VE L EA R N ED

4. Test Creative Placement to

Draw a Reader’s Eye
Did you spot the difference between the two
pictures below? In the first image, the model is
looking right at the CTA button, subtly drawing
readers’ attention to it. That version saw an 89%
increase in clicks over the first image.

This image saw a CTR of 0.339%. (In other words, an 89% increase
in click-throughs.)

13 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

5 TYPES OF T E STS AN D W HAT W E’ VE L EA R N ED

5. The Kitchen Sink Test

In this example we wanted to test image
headline, subheadline and the call to action
verbiage. We found the one with “download the
research” received more clicks and impressions
than the one that read “download the eBook.”

Insider Tip
Ad Impressions: 8,727, Ad Clicks: 33, CTR: 0.38%
We typically recommend keeping text on
our Sponsored Content images to a
minimum, but that is exactly why we continue
to test and iterate until we know exactly what
our audiences in different regions prefer.

14 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

 This is just a small
sample of the tests we
have run on our LinkedIn
Marketing Solutions
team. And we continue to
experiment every day. The
moral of the story here is to
keep testing and iterating
until you find what
copy and imagery
resonates most with
your audience.
 TESTI N G I D E AS WE ’ RE K EEN TO T RY N EXT

Test & Iterate = A Happy and

Engaged Audience
Branded vs. unbranded
My hypothesis is your logo holds a We’re not saying, “Make these specific changes
certain amount of credibility and and get results,” of course. The delightful
therefore the branded version may challenge in marketing is that every audience
see better engagement. is different. What works for one group will leave
another group cold.

Take these tips back to your own marketing

Including a CTA button lab and use them as inspiration for your own
experiments. Commit to ongoing testing and
vs. no CTA button improvement because each piece of data you
My hypothesis is that the CTA button get from these tests will help you understand
would entice more folks to click. your audience just a little better, and in turn
improve your results.

Which resonates best?

eBook, guide, workbook,
crash course?
My hypothesis is that eBook is quickly
becoming a dated and overused term.

16 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

R ESOURCE S TO S E T YOU UP FO R S UCC ESS

Secret Sauce: How LinkedIn uses LinkedIn for Marketing.

We’re offering a behind the scenes tour of our process—how we use each tool in the
LinkedIn Marketing Solutions toolbox.

Sophisticated Marketer’s Guide to Content Marketing.

Everything you need to know about implementing an effective content
marketing strategy on LinkedIn.

Get Started with Advertising on LinkedIn.

Create an ad in minutes. All you need is a LinkedIn account.

GET STARTED

17 > 5 Tes t i ng Ideas Fro m th e L i n kedI n M a r keti n g Va ul t

Today, LinkedIn members number more than 450 million
professionals. That’s more than two-thirds of the 600 million
professionals on the planet, representing the largest group
anywhere of influential, affluent, and educated people.

For more information, visit marketing.linkedin.com