Post 1: The Disposition Effect

Right Brain

A colleague recently emailed me a challenge: Can you destroy your best ideas that aren't working?

In the email, my colleague was pondering the words of the Berkshire Hathaway legend Charlie Munger, the business partner of the even more legendary Warren Buffet. Munger had told the Wall Street Journal that part of what made him more successful than other investors was his ability to intentionally destroy his "best-loved ideas."

Challenge accepted.

I stood up from my office chair, looking out a glass window and exhaling the question: Could I or couldn't I destroy my best-loved ideas? In less than a minute, I had my answer, and it was resounding. Yes, I thought. Making tough choices came natural to me. Been doing it all my life, showcasing the mental grit to work through hard decisions. Of course I could. Actually, I'll do it right now.

I dropped back into my seat.

Next thing I know, as an eCommerce guy and digital marketer, I'm opening a browser on my computer and clicking new tabs in a wide monitor. There, in front of me, the marketing dashboards begin to populate. Campaigns from Google and Facebook fill my screen with grids of numbers--metrics on how much we spent on advertising and the revenue those ads produced. Which best idea can I destroy today?

Some metrics look terrible.

Emboldened by this Munger's insight, I'm ready to take action. My really bad best-ideas are in my sights and some are totally killing my overall digital marketing "Return on Ad Spend (ROAS)."

As I'm about to stop the campaigns, another thought really hits me: Oh man, I set up the campaigns. If these campaigns are bad, then the inputs must really suck, and if the inputs suck, then I must be a very awful marketer. The perceived failure stings. Ugh. To avoid that mental image of myself, I quickly shift my thinking, looking for something more positive. Scanning the number grids, I look for signs I can rescue a campaign and save a best-idea with a tweak to the text copy, bidding strategy or landing pages. But my focus on the rescue doesn't last long. I'm hung up on this feeling that I suck at marketing. It hurts.

I go get coffee :(

That, my friends, is a true but very dramatized story of The Disposition Effect (DE).

At that moment, I avoided destroying a best-love idea because of the perceived emotional pain I would feel by having to admit I failed at something I was responsible for.

Stock traders everywhere should be able to empathize here, as the Disposition Effect has been a well-documented psychological behavior in the financial industry. Specifically, in the investment field, research has shown that investors are more likely to sell off stock with gains rather than stocks with losses, just to avoid emotional pain.

(BTW: This is in stark contrast to a “rational agent,” which might be a lead candidate for the next post, perhaps...).

With that said, how does Disposition Effect relate to ROAS? Great question.

But first, a quick word on Kahneman

I first heard of the Disposition Effect from Daniel Kahneman, the Nobel Prize-winning economist. He explains the Disposition Effect and other key psychological concepts in his seminal book on decision-making, "Thinking, Fast and Slow." The guy is amazingly articulate of complex ideas.

It's a must-read. It will change your life. Nuff said.

So, why can Disposition Effect kill digital marketing ROAS?

Before getting into the breakdown and logic of why DE can be so damaging to ROAS, let's go with the summary up front.

The Disposition Effect occurs in digital marketing when we keep spending money on under-performing ad campaigns (lower ROAS) instead of admitting a campaign we commissioned (our best idea) has failed, shutting it down and re-directing funds to better performing campaigns.

Now, let's unpack the why. If you are unfamiliar with any of the concepts, just click on the links for deeper explanation.

1. At a high level, modern digital marketing portfolios often rely on a strategy of Test-and-Learn (T&L) or digital experimentation.
2. Digital experimentation is employed so marketers can measure advertising tests, such as running a bunch Facebook video ads, and see what works and what doesn't empirically.
3. Knowing what works and what doesn't empirically allows marketers to digitally optimize campaigns--to customize campaigns for certain relevant audiences at what lowest cost.
4. With optimized campaigns and more predictable ad performance, marketers can make rational decisions based on calculated probability and expected value.
5. Calculated probability (and the law of large numbers) help reduce opportunity cost (yes!), or the lost benefits of associated with other options. See a visual example below, comparing the stock price growth of Amazon and GE over time, you can see how much value an investor could have lost if they held onto GE as a best-idea, instead of selling it off and buying AMZ shares.

1. Finally, in a portfolio, less opportunity cost means stronger ROAS.

Since the Disposition Effect gets in the way of digital marketers making mathematically rational decisions and figuring out how to optimize ads, Disposition Effect increases opportunity costs of letting sub-optimal campaigns run too long in portfolios and drive down ROAS. That really sucks.

And now more than ever, with advertising budgets being the lifeblood of digital marketers and ad costs rising, every dollar counts.

In case a marketing example is helpful...otherwise skip down to Left Brain

The scenario: A digital marketer starts up with a Facebook Ads account.

She builds out the account with audiences, creative assets and ads, claims and landing pages and is ready to go with her test-and-learn method.

She runs a campaign with a mix of 8 ads, testing 2 creatives, 2 product prices and 2 audiences. (Why 8? Three variables with 2 settings each = 8 ads.)

She eagerly tracks a host of performance metrics but her unit of analysis is ROAS with an objective to find the highest one.

She runs the campaigns--volumes of impressions and clicks--and results populate in her Facebook dashboards.

She finds statistically significant differences among the ROASs of the 8 ads. The winners are in!

(For those of you who apply stat sig already, have a look at the multi-armed bandit approach. It can be a nice lift:)

The marketer leaves the 2 top performers to run to optimal success and pauses the bottom 6. Now she has more budget for other advertising budgets or new experiments on Facebook.

But a more lasting effect is that a marketer now knows the measured difference between two variables, such as Creative ad X drives 15% better ROAS than Creative ad Y. She can add this lesson to her tool kit to drive better user experiences in all sales channels.

That's a big win.

Left Brain

How do I avoid letting the Disposition Effect kill my ROAS?

It's the goal of this blog to help inform colleagues of interesting concepts in psychology and economics and take deliberate actions to optimize their businesses.

So, here are three general actions and three specific ones to help you mitigate the threat of the Disposition Effect. Go get 'em.

General

• Take a tactical pause. We all use mental shortcuts to live our lives more efficiently (you don't consciously tell yourself how to tie you sneakers), but shortcuts can contain biases like the Disposition Effect. These biases can damage business decision-making, so schedule time each week to consciously review your recent business decisions and campaign performance. Can you destroy a best-idea? Friday mornings can be as a good time as any.
• Take the "me" out of it. When discussing best-ideas, try to focus on the substance of an idea and strength of arguments, as opposed to overly crediting a single person for a best-idea. It makes the idea generation and execution cycle much less personal and destroying best-ideas less painful. Let the merits and test-ability of an idea shine.
• Adopt a test-and-learn mindset. Keep using logical debate as an effective tool to refine ideas and hypotheses. Applying that type of mental rigor can sharpen business ideas. But consider a testing method to (1) find empirical differences of ideas, (2) leverage perspectives from multiple team members and (3) avoid dead-locked debates. As an example, teams don't all have to agree only one image, marketing copy or audience in Facebook Ads. They can test suggestions from multiple team members, as seen in the above example.

Specific

• Define a target metric. Specificity in a goal and how its to be measured makes determining success less subjective. The cold hard data can decide winners and losers. If you run a test on a website checkout layout with the unit of analysis of conversion rate and find that a new layout delivers a stat sig lift over an existing layout, teams can spend energy on unpacking the why and ideating on new test options.
• Draw an issue tree. Writing a tree allows marketers to think through a sequence of ideas to drive business outcomes. It can open thinking to the vast (and testable) options that are available to driving business and ad campaigns. With more options and a testing mindset, less pressure can be applied to individual best-ideas and people. Honestly, when facing a expansive issue tree, a marketer will be more inclined to prune best-ideas to ensure that the best of the best-ideas will get to market.

• Run tests to statistical significance. "Stat sig" is a way to assess how reliable a metric is against random chance. For instance, if the results of a website homepage a/b test of a button color are 90% that means you can be 90% confident that results were not the result of random chance. The point is, a marketer doesn't have to put his name on the line when subjectively endorsing a result and this helps mitigate Disposition Effect. Let the stats do the talking. Lastly, go for higher confidence levels, say 95% or 99%, if you want more reliable results.

That's all for today! Go forth and find a winner!

P.S. Thanks to all the smart people at #McKinsey, #Optimizely, #Investopedia, #Kaushik, #behavioraleconomics, #Google, #Facebook for helping explain these ideas to people like me.

Key concepts w/ links!

1. The Disposition Effect
2. Opportunity cost
3. Return on Ad Spend
4. Statistical Significance
5. a/b testing

About Right Brain/Left Brain

The human brain has two hemispheres, according to neurology. The right side controls creativity. The left side controls logic. RB/LB translates interesting concepts in psychology and economics into deliberate actions to optimize your business.