A/B Test Finds 55% Increase in Leads When Images Added to Tweets [New Data]

Does Twitter look a little different for you lately? Maybe a little more like Facebook or Instagram? That’s because on October 29, the social networking site announced that it will start automatically showing images in your stream (without you even clicking).

As a social media marketer, you can imagine how excited I was to test this out for @HubSpot. All I saw was a gold mine of opportunities for sharing our blog posts, ebooks, templates, webinars, and other expert resources in a visually appealing way.

While I was scrolling through Twitter after this new update was implemented, I noticed I was completely looking over the standard text tweets and began gravitating more towards ones with images. Let me show you what I mean.

There are the traditional text-based tweets like this ...

How to Create Five Fabulous Infographics in PowerPoint http://t.co/QNvq7LVEkp

— HubSpot (@HubSpot) November 3, 2013

... and now tweets with images that look like this.

How to Create Five Fabulous Infographics in PowerPoint http://t.co/t4T61Esbrj pic.twitter.com/MOK3TWV9tU

— HubSpot (@HubSpot) November 2, 2013

At first glance, this feels like a huge improvement, but I was curious to see if it actually affected engagement, conversions, and leads.

So, I decided to conduct a little experiment.

What is A/B testing?

Let met start by briefly explaining the concept of A/B testing, for those who may not be completely familiar with the commonly used marketing tactic.

Essentially, to conduct this type of test, you need two variants to compare: A and B. In this case, tweets with images and tweets without them are the ones we measured against one another in a controlled manner.

We ran the A/B test until it became statistically significant -- if you decide to run this test for yourself, be sure to check for statistical significance before drawing any conclusions.

Methodology for the Test

As for my specific methodology for this A/B test, I scheduled 36 tweets over a week's time through HubSpot’s Social Inbox. Of these 36 tweets, I used 18 different blog posts, ebooks, and templates, using the title of the resource as the copy with no modification.

For the other 18 tweets, I used the same exact blog posts, ebooks, and templates with the exact same copy as the original set of tweets. The only difference was that I attached an image in addition to the title and link in this latter group of tweets.

To ensure accuracy with the results, I posted tweets with the same copy on separate days of the week but around the same times of day. This means the traffic levels on our Twitter account would be about the same on the separate days the tweets were posted. Additionally, I implemented UTM tracking codes to the tweets (learn more about these here). Basically, these codes helped me distinguish which tweets were which in my campaign and easily track the results.

Higher Activity for Tweets With Images

The results of our campaign were crystal clear. You can obviously see the difference in engagement on the tweets embedded above -- but that’s just one example. On average, we found after our experiment that tweets with images vs. those with no images have a:

• 36% increase in clicks (Tweet this)
• 41% increase in retweets (Tweet this)
• 48% increase in favorites (Tweet this)
• 31% increase in visits (Tweet this)
• 33% increase in visitor-to-lead conversion rate (Tweet this)
• 55% increase in leads (Tweet this)

In order to separate the tweets into “image” or “no image” buckets, I’d tag each tweet with a unique campaign token within HubSpot. This helped me see engagement metrics such as clicks, retweets, and favorites. I could then export a CSV file of the numbers to compare, take averages, and determine the percent change.

To determine more ROI-driven results, such as visits, conversions, and lead generation, I was able to see specific lead numbers based on each tag.

Implement Images in Your Tweets

With a 55% increase in leads, it’s a no-brainer that you should test out images for yourself. Just be mindful that you need to optimize your images differently than you would on Facebook or Instagram.

Instead of a perfect square, you should be uploading an image with a 2:1 ratio (starting around 438x219 pixels for standard resolution). This was a problem that we ran into when testing images out for ourselves.
Instead of this ...

... you want it to look like this.

Also, you’ll want to keep an eye on your frequency with images. We usually send about 180 tweets in a week, but decided to cut down the frequency significantly for this test so we wouldn’t rub anyone the wrong way.

Impact on Your Campaign

Numerous users and brands are catching on to this hot new update, which means images are flying through Twitter feeds left and right. It's imperative, though, that you don’t overpost images on your brand's account. Post too many image-based tweets and you might risk losing some followers.

My advice? Start slow, collect some data about your audience, and then increase the frequency of your tweets with images over time if you find they're successful with your followers.

Have you tested out images on Twitter yet? What results have you found for your own audience?