The Battle of Conversion Rates — User Generated Content vs Stock Photos

Originally published on Bllush’s official blog: The Battle of Conversion Rates

The concept of user generated content, or UGC for short isn’t new in the world of eCommerce. It’s one of the biggest trends in recent years, but is it actually worth the hype? Should eCommerce businesses invest and support the trend? I’ll share some research I’ve done in recent months on the topic.

Today, we’ll talk about fashion & apparel images. Simply put, images of real people uploading pictures of themselves wearing products. It usually shows the product being worn in a real-life setting by people who look more like an ordinary person and less the professional model the brand chose to portray it’s “brand image”.

What are we testing today?

So I want to compare how well users engage with each type of image. Do people purchase more when shown a stock photo of a model or a real-life image from social media? Can using these types of images actually move the needle in terms of conversion rate?

Test #1 — Nike Sports Bra

Let’s take a look at one of the classic, bread-and-butter Nike products:

Product Page on Nike.com

This is the product page as it appears today on Nike.com (US website). The main image is clean, a model, and looks probably touched up. This is what’s being used to sell today, meaning it’s passed all Nike’s brand-image qualifications. Can we do better? We’ll test this image against a user generated image of the same product.

After some scouring on Instagram, I found this image that will be our use-case for comparison:

It’s now time to start the fun part. We’ll do a classic A/B test here, insert both images into a small campaign on Facebook and see which image results in more clicks to the destination URL (we’ll use CTR and conversion rate interchangeably)

Here we go! We let the Facebook campaign run for several hours, with each image getting a few tens of thousands of impressions, here are the results.

Interesting results. Both images were placed in the same campaign, with the same exact target audience and placements. Both images were seen by a large amount of users, to remove statistical “favoritism”. The UGC image, got a conversion rate of 0.90% against a 0.31% for our model. Three times better. Our Instagram image hit the branded Nike image out of the park.

Test #2 — Colored Zara Skirt

So maybe the previous success was a fluke. Let’s try another one, this time with a different type of product, a rainbow-colored skirt from Zara. I took the stock photo from the official website and the UGC image from Instagram.

We did the same kind of test here. Facebook campaign with both images on A/B testing and let it run for a few hours. Let’s see the results:

Again we hit a winner. The user generated image from Instagram got a conversion rate (CTR) of 0.62% compared to the 0.24% that the stock photo got. That’s 2.6x better. Not too shabby at all.

Test #3 — Nike Running Shoes

Maybe it’s just clothes that do better? Let’s try to throw in some shoes in the mix. Here is the next pair of contestants:

Finally. On our third test, we found a pair that the user generated image wasn’t better than the stock photo. We got a 0.38% conversion rate against a 0.42%, pretty similar results. Maybe using a different image we would’ve seen different results? Perhaps. Perhaps not.

Test #4 (the last one!) — Red High Heels

Let’s do one more test on shoes. But this time let’s try something different. Instead of showing different images, let’s mix it up. We created a landing page and designed it as an eCommerce product page. 50% of the users saw the version on the left, a product page with the single stock image. The other 50% saw the same page, but with three user generated images below it, showing the same product. Here are screenshots of the web pages:

Here we used Google AdWords to drive traffic to the landing page, using generic keywords like “red high heels”. Once the users reach the page, the A/B testing started. Here are the results.

Here, the CTR is measured by the amount of people who clicked the “Add to Cart” button we had on the page. With the three images, we had a click through rate of 5.31%. The standard image got 1.40%. That’s almost x4 more effective!

The Bottom Line

So there you have it: four different tests. Three successful. One not. This was a great personal lesson in A/B testing. The main takeaways were you can never know if something’s going to be a hit or miss in terms of conversion. Testing is the key. We could of just as easily found that it reduced conversion rate on some products. Doing as much tests as possible and keeping the winners is the key to succeed here.

Now you try, but keep in mind…

A few things to remember when trying your own tests.

  • Try different images. All the tests I did were of the same image in terms of stock photo and UGC photo. The ideal implementation of this would be to do an internal A/B test within the user generated image area. So within the 50% of users shown a UGC photo, they would see a random photo every time until the system gathered enough data to determine which images are the absolute superior.
  • Isolate tests. Pretty standard principle in A/B, but every test should compare one element of the page. Run a A/B test of just switching the image or adding a layer of images, don’t add this on to an existing test of anything else (like button color for example).
  • Test size. Depending on your traffic, you might need to wait a different amount of time until you can safely decide on a winner. There’s no exact figure, but it’s a good practice to keep testing until the results are stable and converge into a specific value. If the conversion rate is jumping around, wait more time. A good rule of thumb, you can stop testing when no single event / sale changes the needle too much (~3%)
  • Permission / copyright — Using random images on Instagram for commercial purposes is a big no-no and a copyright violation. Only use images that you have permission to use

Want to get these strong conversion metrics on your eCommerce website as well? Bllush’s technology specializes in getting long term results. Learn more about our products or better yet, reach out and chat with us.

What about you — what’s been your experiences with using user generated images within eCommerce websites?

Tomer Dean is a serial tech entrepreneur based in Tel-Aviv working on an SaaS startup Bllush which helps eCommerce websites smartly use user generated content.