You’ve read about color psychology, system one and two, emotional persuasion, etc. I know you have because it’s everywhere. It’s on KISSmetrics, Forbes, Entrepreneur, Inc., Help Scout, HubSpot… you name it. Hell, we’ve covered some of these topics ourselves.
Why? Well, because many psychological triggers do, in fact work.
But there’s another side to using psychology online that almost no one is talking about: backfiring.
Psychology isn’t a magic formula that can be applied to optimization seamlessly in all scenarios, despite what many self-identified experts are preaching today. [Tweet It!]
Before you apply another psychological trigger to your site, do yourself a favor and understand the risks that come along with the rewards. Better yet, understand how to reduce those risks.
What Is Backfiring and Why Does It Matter?
A psychological backfire is when psychology is poorly applied and the exact opposite of the intended behavior is triggered.
So, for example, you believe that social proof will psychologically trigger conversions. When you add the social proof to your page, conversions decrease because you’ve chosen the wrong people to feature or because your audience prefers to go “against the grain”. That’s psychological backfiring.
Think of backfires the way you think of A/B test validity threats.
Recently, Dr. Brian Cugelman of AlterSpark and Dr. Agnis Stibe of MIT Media Lab published a scientific paper on psychological backfiring. When asked why he decided to take a more in-depth look at backfiring, Dr. Cugelman had this to say…
Dr. Stibe added that because many people are afraid to talk about backfiring, visibility is important if we want to preserve the scientific integrity of research…
But that’s idealistic and there’s another side to applying psychology to optimization that few people know about, let alone are willing to discuss. That’s where backfiring comes into play.
Here’s Dr. Cugelman on the trend of applying psychology to optimization…
Examining the different types of backfiring brings a healthy dose of reality to those using psychology to convert. It’s not a magical formula that can do no wrong. In fact, a lot can (and does) go wrong.
Being unaware of (or unwilling to accept) that is dangerous. You see, psychology can be used in optimization… it works, but rarely as drastically as promised in overhyped articles and not always in your favor.
Right now, you only have half the story on using psychology to optimize your site.
The Intention-Outcome Matrix
Every psychological implementation will fall within one of four categories on the intention-outcome matrix. Two are positive (target behavior and unexpected benefits) while two are negative (dark patterns and backfiring)…
In optimization, we focus almost exclusively on the two positive outcomes, which is a problem. It’s important to be aware of all of the possible outcomes, which will occur whether we have the ability to recognize them or not.
Here’s what you need to know about each category…
1. Target Behavior
This is the positive outcome you were hoping for. For example, young adults reduce nicotine dependency.
2. Unexpected Benefits
Any positive outcome that was not intentionally sought. A “happy coincidence” or “positive side-effect”. For example, regular nicotine screenings encourage young adults to go to the gym more often (i.e. lead a healthier lifestyle).
Includes a variety of negative outcomes, including: the reverse of the target behavior and side effects of the target behavior. For example, regular nicotine screenings lead to young adults competing to see who can smoke more cigarettes.
4. Dark Patterns
Any outcome that benefits the experimenter at the expense of the audience. This is unethical manipulation, which is discussed at length in our article, Online Manipulation: All The Ways You’re Currently Being Deceived.
The Likelihood-Severity Matrix
Of course, backfiring is more complex than “negative and unintentional”. In their paper, Dr. Cugelman and Dr. Stibe propose that all backfires fall within the four quadrants of the likelihood severity matrix…
Within the quadrants, their research found that there are six types of backfires: credibility damage, inexperience, fineprint fallacy, personality responses, poor judgment, and social psychology. Within each type of backfire, there are a number of actual backfires.
Note that this list is incomplete. This is just the beginning of research into psychological backfiring. As psychology continues to be used more widely in marketing and other professions, new backfires will come to light. For now, it’s important that you’re familiar with the original 12…
1. Credibility Damage
- Self-Discrediting: Low likelihood, major severity. This occurs when your message is inauthentic, difficult to believe or based on discredited information.
- Message Hijacking: Low likelihood, major severity. When your audience (or the general public) takes your original message and gives it another meaning.
Example: Heroin Chic
In the 90s, heroin chic became popular…
If you’re not familiar, it’s essentially characterized by pale skin, dark circles underneath the eyes and angular bone structure. Fashion models in the 90s began to adopt this look, a clear reaction to traditional models like Cindy Crawford.
The trend began because heroin prices dropped, its purity increased dramatically and insufflation (vs. a needle) became more popular.
All of this led to heroin chic. Teenagers hung posters of heroin chic models on their walls… they wanted to be just like them. What should have been a startling warning against the effects of heroin became sexy, it became idealized.
You might remember this one. In 2012, McDonald’s thought it would be a good idea to get people to share their heart-warming stories about McDonald’s on Twitter. “We bought Timmy a Happy Meal after he won his first big game,” style stories. Big mistake.
Here’s what people actually ended up tweeting…
Their message was completely hijacked. Instead of the heart-warming stories they were hoping for, they got horror stories and mockery.
- Superficializing: Low likelihood, minor severity. The application of a theory without an understanding of the underlying strategy / logic. In other words, applying superficial tactics without reasoning.
Example: Social Sharing
The dramatic response to Twitter’s decision to stop supporting social share count demonstrates how wide-spread the use of social sharing buttons is. It’s become a best practice to show social sharing buttons with share counts.
Here’s an example from a mommy blogger’s site…
Can you deduce how this might be a backfire? 0 Facebook shares, 0 tweets, 0 pins, 1 G+1, 0 email shares, 0 comments… This is essentially anti-social proof. It’s the blind application of a best practice without an understanding of the underlying strategic / logic behind the practice.
3. Fineprint Fallacy
- Overemphasizing: Exclusively highlighting the benefits while hiding negatives / side effects in the proverbial (or literal) fine print.
Example: Rebill Offers
Anyone who has dealt with rebill offers is familiar with the overemphasizing backfire. Miracle muscle building formulas, acai berry super fruits, etc. Here’s an example…
The entire site is dedicated to highlighting and glorifying the product. Note the phrase “See If You Qualify For Our Trial Program”, which is used to motivate those who are unsure. You don’t get the free trial because you want it, you need to actually qualify for it (which isn’t true).
In the fine print, however, you’ll find that once they submit their information and receive their trial product, they have X days to cancel before they end up with a hefty recurring bill. Will many people cancel? No because the need to do so is hidden.
4. Personality Responses
- Defiance Arousing: When a message does not align with the audience’s self-identity, they are inclined to defy it (i.e. do the opposite behavior).
- Self-Licensing: When someone is succeeding or behaving correctly in one area of life, they sometimes believe they can neglect or misbehave in another.
Here’s our beginner’s guide page…
If we sent this page to optimizers who’ve been practicing conversion rate optimization for years, it would not align with their self-identity. As a result, they probably wouldn’t read through it or download it for later.
5. Poor Judgement
- Mistailoring: When a message is not tailored to the audience correctly and creates a negative reaction among part (or all) of the audience.
- Mistargeting: When a message intended for one segment of the audience is misinterpreted by another.
- Misdiagnosing: When you do not correctly analyze your audience’s behavior and cognitive processes.
- Misanticipating: When controlled changes lead to unexpected changes in beliefs and behaviors.
Example: The Economist
When The Economist realized women make up online a small portion of their audience, they decided to run this…
Yikes… not sure who approved that one. Obviously, they meant that The Economist is for everyone, regardless of gender. But since their audience is mostly male, it certainly didn’t come across that way.
6. Social Psychology
- Anti-Modeling: When a negative behavior is shown to demonize and condemn it, but the negative behavior increases as a result of the increased awareness.
- Reverse Norming: When a negative behavior is shown, it can become a social norm, which increases the likelihood that it will be repeated.
Example: Growth Hacking on CXL
Here’s an example of anti-modeling from a previous article of mine…
I clearly show how “growth hacking” should not be looked at by condemning the fluffy “growth hack” articles. If this were to backfire, readers of CXL would write these types of articles to boost their traffic. (Hopefully that didn’t happen.)
Where Conversion Rate Optimization Falls Short
When asked about what optimizers can takeaway from the backfiring research, Dr. Cugelman had this to say…
As optimizers, we’re more concerned with superficial psychology. We want to implement whatever psychology improves conversions (positive behavior). We don’t consider the long-term or underlying psychological modifications (negative behavior) whatsoever.
Is this a bad thing? Not really… assuming you’re not out to psychologically damage your audience, of course.
What it does mean, however, is that optimizers remove backfires without even recognizing them as backfires. They’re just “something that decreased conversions for an unknown reason”.
Imagine if you knew what that reason was. You could explain it better to your colleagues / boss / clients, and you could train yourself to identify that “something” before launching the treatment or when conducting heuristic analysis.
Being aware of the different backfires (especially poor judgment and social psychology, which are highly likely and very severe) makes you a smarter optimizer, a more efficient optimizer, an optimizer with fewer losses (and thus, more wins).
How to Reduce the Risk of Backfiring
It’s impossible to completely remove the risk of backfiring because you can’t possibly predict how everyone within your target audience will interpret every element of your site. You have to learn to live with it.
What you can do, however, is identify and reduce the risk of backfiring by following these four simple steps.
1. Be aware of what can go wrong.
For best results, take a risk management approach when creating sites and running tests. With your specific test in mind, consider what could backfire. Audit your site or test for each of the 12 backfires. Start with the high likelihood quadrants. Make a list of threats and keep them in mind as you design your tests.
2. Test your psychology.
Don’t implement psychological tactics blindly. You’re an optimizer… test it! It’s not just about testing to find what works to increase conversions. It’s about removing what’s not working, too. That’s not always immediately obvious.
If you think X will cause a behavior change of Y, test it on a portion of your audience before pushing it live to everyone. After determining whether it improved conversions or not, look for evidence of backfires using qualitative research (e.g. user testing, click maps, scroll maps, surveys).
3. Consider the long-term impact.
If you do successfully modify behavior of your audience, how will that impact similar groups?
Consider the people just outside your target audience. If you change the behavior of your target audience, that might influence the way adjacent audiences behave.
An example I like to use for this is CrossFit. So many people love CrossFit that it puts off others who are “sick of hearing about it”. As a result, people who are interested in going to a gym might turn to GoodLife instead of CrossFit, which they have now developed a negative opinion of without even trying it.
When you implement psychology, you must ask yourself: How will this impact my ability to scale?
You have to look at backfires holistically.
4. Understand the complexity of backfires.
Psychology can work with one part of your audience and backfire with another part. For example, some people will love Peep’s blunt, “I call your bullshit” style. Others will be turned off by it and seek out a “softer” alternative.
There isn’t always a “right vs. wrong” when it comes to backfiring. Just because it backfires with some people does not mean you should remove or change it. Only remove or change it when you find that it backfires holistically.
Also, consider that you sometimes lose control of the backfire. Consider the #McDStories example. Once the message is hijacked, you lose control of it. You can’t remove or change it.
This is is just the beginning of optimizers using psychology to increase conversions. As more “experts” preach the magical results, more backfires will plague unsuspecting optimizers. [Tweet It!]
Here’s how you should handle backfires going forward…
- Remember the 12 backfires and where they fall within the likelihood-severity matrix.
- Audit your site for psychological backfires.
- Before launching a test, audit it for the risk of psychological backfires. Do what you can to reduce those risks.
- Launch an A/B test, limiting the risk of widespread backfires.
- Determine whether the treatment increased conversions or not.
- Using qualitative research, look for evidence of backfires.
- Record all of the backfires and refer to them before running future tests.
- If you still have control of the backfire(s) and it’s causing an overall negative outcome, remove it.
For more scientific / academic papers converted into actionable optimization insights, check out the CXL Institute.
Oh, and if you happen to find any new types of backfires or new examples of the ones identified above, please share them below.