BLOG

on Conversion Rate Optimization

Stock Photo or Real Image? A/B Testing Finds Out Which is Better

Stock Image or Real Image – what should you use? The debate has been raging for a while now. That’s unfortunate, because there is no one answer that will work for all businesses alike. Why speculate at all, when we can throw the contenders into an A/B test and sit back while statistics find us a winner? Think of it as WWE, except A/B tests are real, and they get you better business. Let’s get right to it then, shall we?

Background

160 Driving Academy is an Illinois based firm that offers truck-driving classes and even guarantees a job upon graduation. Visitors to the site primarily use the contact form on the homepage, or the prominently displayed phone number, to contact the academy. Looking to improve the conversion rate on the truck-driving classes page,  the academy reached out to SpectrumInc, a lead-generation software and internet marketing company. The rest (as they have not yet begun to say, but soon will) is a future of great conversions!

The Hypothesis

The academy had been using a stock image of a man driving a truck on its homepage. When SpectrumInc came on board, they decided to test the page with the photograph of a real student instead. The hypothesis was that the image of an actual student would outperform the stock image the academy had been using. On being asked about the background of this test, Brian McKenzie from SpectrumInc explains,

… in this case we had a branded photo of an actual 160 Driving Academy student standing in front of a truck available, but we originally opted not to use it for the page out of concern that the student’s ‘University of Florida’ sweatshirt would send the wrong message to consumers trying to obtain an Illinois, Missouri, or Iowa license. (These states are about 2,000 kilometers from the University of Florida).

Better sense prevailed, and they decided to test it anyway.

What Goals Were Tracked?

The primary conversion goal: Number of visits to the ‘Thank You’ page. These are the pages that visitors are taken to after they fill out a conversion form, like the ‘contact us’ form on the main page.

The secondary conversion goal: Number of visits to the ‘Registration’ page. The academy carries a CTA button on its page that says “Register for Classes”. A conversion would be recorded every time a visitor clicked on the button and visited the “Registration” page.

The Test: Stock Image or Real Image

Comparison Image

The Result

An incredible 161% lift in conversions, at 98% confidence level. Or, the possibility for such a massive change in conversions occurring simply due to random chance (and not because the variation actually is better at converting visitors) is just 2%.

Secondary Goal: Registrations, too, saw a 38.4% spike on the variation compared to the control, at 98% confidence level.

Why did the Variation win?

As with any retrospective analysis, the key lies in exploring the data and connecting it to the knowledge that is already out there. First, let’s understand why images are such a big deal, and what part they play in user experience.

Short (and borrowed) answer: An image is worth a thousand words.

What does it say?

Concepts learned in the form of images are more easily and frequently recalled than other ideas learned through text. In fact, Wikipedia explains that this effect is much more pronounced in older people than the younger ones. So if your business targets the age group of 25+, images are a great way to pass on brand-related information for better recall.

Billion Dollar Graphics explain, and I quote, “human brain deciphers image elements simultaneously, while language is decoded in a linear, sequential manner taking more time to process.” This is further illustrated in the following image.

Illustration - Graphic vs Text

 

Do you see how much easier it is to understand that the reference is to a square from the image than from its textual description? In fact, if you are in the mood for some serious reading, I strongly recommend this incredibly insightful post on the power of visual communication.

This frequently quoted eye tracking study from NN also confirms that we spend more time dwelling on images on a webpage rather than on the text itself. When they tested an “About Us” page that contained thumbnail portraits of each of the members of the team, this is what was found:

Here, the user spent 10% more time viewing the portrait photos than reading the biographies, even though the bios consumed 316% more space. It’s obvious from the gaze plot that the user was in a hurry and just wanted to get a quick overview of the FreshBooks team, and looking at photos is indeed faster than reading full paragraphs.

Eye-Tracking Study on Images By NN

Evidently, people focus more on images on a page than on the text itself. And they retain it longer. The case for images cannot be overemphasized.

Now that you and I agree upon the need for using images, let’s dive right into analyzing the case. We start with:

The Control, with the Stock Photograph

Why did it convert so poorly?

  • We Love Ignoring Images That Look Stock

Stock images were a rage back in the late 90s, when taking a good picture was best left to professionals with complex, expensive cameras. Naturally, online businesses that were just starting out had to resort to the relatively inexpensive and definitely good-looking stock photos.

Here’s the issue: we have been exposed to banner advertisements for so long that our eyes have gotten trained to ignore any web element that evokes the feel of an advertisement. The adage “familiarity breeds contempt” holds true and banner blindness has been confirmed to be a real phenomenon in numerous studies. More stock images, anyone?

  • Stock Images Are Not Unique

I popped the stock image from the client’s old homepage into TinEye, a reverse image search engine, and this is what it threw up.

TinEye Reverse Search Result

That’s 30 other instances on the webpage where the same stock photo was found.

Just to hammer home the point, I let Google Image Search do its thing. And here’s what Google found for me.

Reverse Image Search - Google Images

That’s 175 results. So much for uniqueness and product differentiation.

So there are more of that image, how’s that a big deal, you might ask.

Where do you suppose the stock image of a man driving a truck would figure on the web?

Same Stock Photo Used By Competitiors

That’s right, on other business websites that are related to trucks; websites your potential customer might have visited already. Google took just 0.45 seconds to find 175 places on the web where the image appeared. Human users would take longer, but they’ll get there eventually. And when a potential customer sees a familiar image on your site, how would they judge your business and its credibility?

Go on, ask me, how would anyone recollect seeing the same image somewhere in a corner of the web?

Because we are super smart and can identify images seen for as little as 13 milliseconds. To put that in perspective, the time we take to blink once is about 300-400 milliseconds.

Enough of beating the life out of stock images. Actually, using stock images, in and of itself, is not the real problem. There are ways to use good, relevant stock images without running into the problem of duplicates; like having a Rights Managed Licence. Instead, the real problem is:

  • Using Irrelevant Stock Images

Okay, stop being yourself for a moment. Slip into the user’s shoes, and I promise we shall see better.

You are looking to get a truck licence. Google suggests you check out 160drivingacademy.com

So you do what you always do. You click and reach the site.

Now, remember, you form the first impression of a website in 50 milliseconds. And you’d typically leave a website in 10-20 seconds unless, you find a reason to linger on. What you are looking for is relevance, sort of a validation that you are in the right place.

Let’s get back to you. You scan the page. And you see our man in the truck. But, what do you understand?

  • How established is the place?
  • Does the academy look credible?
  • Wait, why is there a severely cropped image of a man sitting inside a truck?
  • Is he the coach?

Oh wait, no! I’ve seen this image before!

The Verdict?

I can’t trust these guys. Where’s the back button!

And, curtains down!

Now, let’s take a look at the variation and try to understand why it converted visitors so well.

The Variation, with the Real Photo

Would you stay in the user’s shoes for a while longer, while I take you on a visit to the variation?

  • It’s All About Relevance

You know the drill. Google tells you. You listen. You are on the academy’s page; but it has the real image now.

“How does the place look?”

“Don’t really know. But that’s a big truck. Branded and all. Place must be established.”

Student's Image Used in Variation

“Is it credible?”

“Can’t be sure, but it looks real! That guy in the picture looks happy, he must be a student. I might even get to learn on one of those trucks in the picture!”

“Alright, no harm anyway, where do I contact them?”

Enough said.

  • We Love Images That Look Real!

This study shows that users focus their attention on images that look genuine with real people and objects. Consequently, we ignore images that seem to only have decorative (read stock-ish) purposes.

Real images evoke trust. On a business site, users are not looking for emotional gratification. They are looking for hints, information about what they’d get if they decide to buy your product/service. A website that uses real images screams at its users,

This is exactly what you will get if you choose us! It’s great, and we know it!

Get the trust, make the sale.

Over the years, we’ve been so indiscriminately exposed to every kind of scam, sham and spam, that we don’t trust easily. Least of all, on the internet. A website that reveals its offerings, plain and clear, tells us there won’t be any nasty surprises. Hence, we trust.

  • Clever Branding and the Hidden Call To Action

What? Where?

Without the variation image, there was exactly one part of the site that called out “160  Driving Academy”. With the variation, there are three such places.

We’ve already seen how our eyes are drawn to images much quicker than it is to text. The variation image draws attention to itself, and in the few seconds that a visitors’ eyes stay on it, the mind picks up two strong branding signals. The brand name itself, and the color associated with it — yellow — generously splashed across the truck in the image. A deceptively simple way to make sure that even users who bounce off the first time remember the brand. I think I wouldn’t be wrong in assuming that a considerable number of the conversions resulted from users who revisited the page.

No, that’s not all.

A call to action. That little big thing.

“Call Today”

Hidden Call to Action in Image

What better place to have it than in the image itself! That too, right next to the contact form. It gives the user direction on what’s to be done if they are interested in taking things ahead, and it creates urgency using the term “Today!”.

So there, little relevant things really matter.

Room for Further Testing

If you check the academy’s current page, you’ll see that the “Florida Gators” print has been edited out of the student’s sweatshirt. If you remember, Brian had pointed out how the reference to ‘Florida’ might confuse prospects who are primarily from Illinois. Removing the “confusing” text from the image should improve conversions even better. Brian also pointed out that the average age of a student at the academy is close to 40, while the student in the image is closer to 25. From this context, Brian shares his vision for further testing,

..trying to narrow down whether pictures of actual customers, pictures of actual employees, or pictures of actual products/equipment/objects convert best. Then you can do more incremental tests, like whether a 40-year-old student would convert better than a 25-year-old or whether the student should be holding up his license or just standing in front of the truck.

Are Your Images Relevant?

What do you think? Is relevance the most vital criterion in selecting an image?

If you feel so, I would like you to head back to your website and reconsider the relevance of the image(s) used. Are they relevant? Would you like some help figuring out if it’s relevant or not?

And if you feel relevance is not the primary consideration, I would love to know your take on it.

Tell us right here, or, if you are a person of few words (couldn’t help it) let us know on Twitter @VWO or, get to me straight @SharanTheSuresh.

Before I leave, here are two more brilliant ‘Stock Image vs Real Image’ case studies from our archive.

45% Increase in Conversions Using Real Image

How about an 89% increase in conversion?

And as always, we’re listening.

Marketing | SaaS | Product. Loves words, people, images, and experiences.

Comments (10)

Leave a Comment
  1. I just wanted to comment about how you explained the confidence level. It is a little off, the 98% confidence level is saying that there is a 98% chance that the change seen is not a false positive. The confidence level doesn’t give any insight into the lift maintaining itself, that is way we have confidence intervals.

    Other than that great read man! I am a huge advocate for genuine photography – down with stock photos!

    1. Hey Justin,

      Thanks for that quick heads-up! The initial text was indeed confusing, I got that changed.

      Great to know that you liked it otherwise. And yes, long live genuine images!

  2. The images aren’t a fair comparison because a direct gaze from a young, friendly face will out-perform a non-engaging, stressed-out, older looking guy in a sea of dangerous looking distractions. A classic case of the effect of eye gaze in connecting with the audience is the Kids and Cereal Boxes phenomenon. Here’s just one of many links to that interesting story… http://foodpsychology.cornell.edu/op/cerealeyes

    That said, YES!!! Real people are the way to go! It’s one of the reasons why brands like Virgin Airlines and Tom’s Shoes work…they represent a brand that is driven by a person that IS the brand. And why crowd sourcing and social media work…we want to connect with those we trust. Stock images will never be obsolete, but they are def old school.

  3. Hi Ann,

    That’s absolutely true. In fact, we’d previously covered it in quite a few cases (including https://vwo.com/blog/dutch-major-achieved-7-8-increase-conversion-removing-line/). The reason why it works is, of course, because we are ‘tempted’ to follow the eye gaze of human characters in an image. And it probably has its roots in evolution. Imagine a time before language was developed – the only way we could warn each other of a danger or point out an opportunity was by directing our gaze or our arms toward it.

    Thinking on those lines, I think if the student in the image had his gaze pointed towards the “contact form”, conversion rates could further improve. This excellent post on the topic (https://vwo.com/blog/guiding-visitors-eye-path-to-increase-conversions/) should be of interest. It’s also a technique that great street photographers use to direct audience’s attention from one element in the frame to another.

    That said, would an irrelevant stock image with a warm friendly face effected the same conversion rate? I wouldn’t be sure. Then again, we can always test it 🙂

    Thanks for the excellent input, Ann. It’s much appreciated.

  4. I agree when we’re talking about the eye gaze – I ran quite a few large scale tests on people imagery for Belron (owners of carglass, safelite, autoglass, 39 countries, blah…).

    Eye gaze does indeed grab attention and that means that when people are *fucking ignoring you* in the photo – it can even be a big downside. I tested several angles of eye gaze – towards the CTA, then closer to the viewer, directly in their gaze and the other way.

    It was in a crazy period where I was testing some very random stuff but it was interesting. The best was just over the shoulder (left or right) or direct eye gaze. It doesn’t always have to be perfectly spot on locked down eye contact.

    OK – now for the next understanding. It isn’t about the image you are trying to convey. The reason the student works better is because indeed he is looking at you but also because he’s smiling and he’s more ‘authentic’.

    Authenticity can be staged but it’s harder than just grabbing a camera. I tested stock photography against low cost photos taken by people at the various offices and the latter performed better.

    One thing about smiles and stock photos in particular. People are subconsciously adept at spotting fake smiles – those where the cheeks and eye crinkles in particular signify the person is really ‘into it’. So a lot of stock photos have kinda fakey smiles or inauthentic body language that makes it feel creepy.

    I’ve got a whole guide stashed away to people photography and tests I’ve run if you’re interested. Body language makes a HUGE difference – folded arms, arms behind, strange hand poses – all these don’t work at all.

    Some recommended stuff – What Every Body is Saying – Joe Navarro + The Collins Guide to Body Language. Really useful books for your own photos!

    Also this book – Photo UX – by @chudders : http://www.photoux.co.uk/

    If you ever want to write a follow up article, grab both of us. He’s got some very useful info and I’ve got a stack of insights from people image testing from millions of sessions.

    The best thing I can say about the image is that it should be used to persuade, support or explain. If the image does none of these, it is simply not relevant. However, the strongest thing of all (IMHO) is the reaction in the mind of the viewer, when presented with the image.

    It’s not about what you think, or your boss thinks, or the designer thinks. What does that picture make them feel? What little video sets off going in there? What plays in that little cascade of feelings or ideas? Aiming for that reaction is where you get the conversion.

    For example, at Autoglass, we had some images of really beefy ripped guys vs. an older female who looked really sympathetic. She ripped the other guys in testing, because people have smashed windscreens and need help, not ripped and hunky guys. They have a distress purchase to make and reflecting that emotional state in photography worked very well indeed.

    In classic art direction terms, she was ‘ugly’ and older and oh, I heard everything bad about that lady. But she slayed the others in a test and proved a point about the viewer reaction not being ours.

    If I was a student, the student in the photo looks ‘real’, he is smiling, he kinda looks normal, like me – and maybe not 100% sure but going for it anyway! Gimme some of that – if he’s up for it, damn, so am i!

    Lastly, for some fun, try this BBC fake smile test (the smile and eye crinkle in photos are very important) : http://www.bbc.co.uk/science/humanbody/mind/surveys/smiles/

    1. Wow! that’s quite insightful Craig. I can’t wait to get your inputs for a content piece. And about us being very good at spotting fake smiles, I in fact remember a study that the Discovery guys aired a year or so back on the exact same topic. It’s incredibly ‘satisfying’ how evolutionary science has such a big role to play in a niche industry like A/B testing and optimization 🙂

Leave Comment

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes : <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Contact Us / Login

Product
Resources Home