Tag Archives: Product management

What behavioral economics can teach about PM’ing: the affect hueristic

(NOTE: This is part II in a series of behavioral economics lessons for product management. You can find part I, on the remembering self, here.)

In part I, I covered the remembering self: the term describing humans consistently biased process for creating memories of events/products/experiences. It’s a critical lesson because the human biases are 1) understandable 2) predictable and 3) most importantly, hack-able if you understand how your users are likely to think.

Another key behavioral economics insight is the affect heuristic: the tendency of humans to rely on emotions to supercharge the decision making process by putting a subconscious stake in the ground at the beginning of the process. It’s particularly powerful because it kicks in *right before* weighing the pros and cons, and thus influences (read: biases) how pro and con lists are constructed.

For example, if a friend asks you what you think of Nike’s new running shoe, the affect heuristic kicks in immediately, queries your memory, collects any related emotions (positive or negative) you have toward Nike, running shoes, or this particular sub-brand. Then your conscious, rational thought process will kick in to tally pros and cons. But the damage is already done. If your initial emotional response is favorable toward an idea/concept/product/technology, you’re systematically biased to overrepresent benefits (pros) and underweight risks (cons). If your initial emotional response is negative, the opposite holds true: you’re biased to overrepresent risks while neglecting benefits. That’s the affect heuristic in action.

Paul Slovic, an expert in psychological heuristics, has studied this in many forms and in one study asked participants to note simply whether they liked or disliked a series of new technologies. Participants were then asked to enumerate the pros and cons for each technology they previously stated an opinion on. Slovic and his team found “implausibly high negative correlation.” If a participant liked a technology, the stated risks were implausibly low. If a participant disliked a technology, the stated benefits were shockingly low. Participants’ pro and con listings were remarkably biased by initial opinions.

Slovic’s team took the study a step further. After the initial the pro/con exercise, participants were asked to read additional information on the technologies. Some participants received a blurb *only* discussing benefits, others received blurbs *only* discussing risks. After reading, the participants were re-prompted to provide their opinions. Unsurprisingly, participants that read about additional benefits were now more positive about the technologies. However, the critical finding was that the participants who read about additional benefits now also believed those same technologies to possess fewer risks. A rational participant should have only found themselves counting more benefits, not adding more benefits and eliminating risks simultaneously!

The affect heuristic was kicking in. Participants were simplifying decisions by letting positive emotions (triggered by additional information) create a world that was more black and white than reality. A favorable technology has lots of benefits and few risks while an unfavorable one has few benefits and many risks.

Why should I care?

1) Immediate emotional reactions will bias any conscious, rational thought. No matter how rational anyone is, his/her judgment will be immediately biased by sub-conscious emotional cues. If you’ve successfully appealed to emotion, you’ve given yourself an unfair advantage.

2) Well defined benefits will proactively minimize risks. If you can clearly communicate the benefits of your product and/or service, you get a double win: the benefits will go-up and the perceived risks will diminish significantly.

Examples of the affect heuristic in the wild

Dropbox is a fantastic example of a product well designed to take advantage of the 1st principle. Think about Dropbox’s competitors Google and Microsoft. Any rational comparison of the features largely favors big G and MS. Both companies offer more free space: Google offers 5GB and MS offers 7GB, compared to Dropbox’s 2GB. Google and MS are stable, cash rich companies with nearly unlimited resources to support any initiative they want. Dropbox has ~300 employees and $250M in cash; Google makes $250M every 2 days.

But Dropbox is winning. And part of it is definitely due to a positive outcome of affect heuristics. They’ve hacked the first impression to successfully appeal to potential users’ emotions. Ask yourself what you think of Dropbox’s homepage (below). Most people say “clean” and “simple.”

Now ask yourself what you think of Microsoft SkyDrive’s homepage. I’ve heard “confusing”, “cluttered”, and “Why is that girl in a tree?”

Now ask yourself what type of cloud storage solution you want. A confusing one (eg “girl in a tree?”) vs. clean and simple. When it comes to pros and cons, 7GB of free storage is less compelling under the specter of a confusing, cluttered experience. It seems that simplicity has resonated with users as well, it’s not the be-all, end-all stat by any means but the Google traffic for the various services is telling… 

Airbnb is a great example of a company that will benefit significantly from nailing the 2nd principle: clearly defining benefits to boost appeal and reduce perceived risks. They stand to benefit dis-proportionally more than other consumer services – eg Uber, Fab, Amazon, Netflix where the risks of poor performance are relatively low (eg my Uber cancelled on me or my movie isn’t playing) – vs. my vacation rental was unsafe or my home was vandalized.

This is particularly important in the latter case for host sign-ups where people are putting their most important and personal asset (eg home/apt/condo) up for rent to relative unknowns. As the “meth incident” highlighted, the risks to putting up your asset for rental are real. But there is no doubt that Airbnb also has very real benefits they can tout to counter the risks: 1) the potential to gain a serious revenue stream from an otherwise un-monetized asset (see screen below) and 2) the more nuanced, but powerful, promise of community / friendship that many Airbnb hosts end up building with their guests. Given their recent stats of 200K+ rentals listed, it looks like they’re on their way toward overcoming the hurdles by communicating the benefits.

There are no doubt numerous other examples which I’ve missed (please feel free to call them out in the comments!) but the key point is that it’s critical to think about 1) what is the immediate, emotional reaction users will have to my product and 2) how can I clearly hammer home key benefits, which will not only raise user awareness of benefits but reduce their perception of risks.

What behavioral economics can teach you about product development: The remembering self

If I could sit down with anyone for one hour and then have them pick apart the motivations driving users in my product, it would be Daniel Kahneman, the behavioral economist behind this TED talk and Thinking Fast and Slow. Although he hasn’t set out specifically to discuss consumer tech, gaming or any reach of the interwebs, his insights are real, verifiable and directly applicable.

One key insight is that human memory is significantly biased. And it is consistently biased. So much so that it’s helpful to consider two selves: an experiencing self (the one that fully experiences every detail, nuance, perfect moment as it happens) and a remembering self (the one which packages up that experience, culls through it, retains a few key nuggets and forms a “memory” of it).

Unfortunately, much to the chagrin of traditional economists, the memory is not an exact reflection of the actual experience (eg, people don’t rationally weight all aspects of an experience). Instead, we predominantly build “memories” based on the most extreme peak or nadir and the final moment (eg, “all’s well that ends well”).

How does he know? Kahneman studied this phenomenoa, in multiple controlled experiments. In one instance he studied experienced and remembered pain levels during colonoscopy exams. Each patient participating was asked to rate his/her level of pain every 60 seconds from 0 to 10, (0=no pain, 10=intolerable pain). They tested 154 patients and the procedures varied in length from 4 minutes to 69 minutes. Here is a sample chart from two patients:

behavioral economics experiment

Any rational person can see that Patient B had a much worse go of it. If you totaled the integral amount of pain experienced by Patient B, it would significantly exceed that of Patient A. But, Kahneman’s team consistently found the opposite to be true when they compared patients’ overall ratings of their experiences.  The Patient As of the world invariably described their experience as more painful.

What gives? Two factors drove this:

The Peak-End rule: The top predictors of how one will remember an experience are: the most extreme moments and the moments near the end. In repeated experiments, the patients’ overall evaluation of pain most closely correlated with the average pain in the final minutes and the peak pain experienced.

Duration neglect: The duration of an experience doesn’t factor into one’s overall assessment. This is kind of mind blowing, especially given the knowledge that we are prone to conflating length with value (eg a 3-day vacation is always better than a 2-day one!).

What does this mean for tech products?

1) Do one thing remarkably well. If extreme experiences will drive user’s memory/opinion/position on your product, your best leverage point is to do one thing very, very well to create this positive association. This has been uttered many times in many different ways. Ben Horowitz and Mark Suster have both written about being 10x better when creating a new product, but until reading Kahneman’s experiments I hadn’t seen evidence that demonstrated why it matters so much.

There are lots of great examples of companies delivering incredible experiences in a single category. I’d certainly place Uber in that category for delivering no questions asked, quick, reliable transportation from A-to-B; Kindle (non-Fire incarnations) for delivering an awesome reading experience (despite getting nearly everything else wrong as I posted here); and Instagram for their sole focus on photos and creating the best photo taking & sharing experience.

2) Make the “sign off” experience killer. Again, if a memory is largely dictated by the final moment, you can hack the experience to your benefit by perfecting the finale. There is a lot of focus on “on boarding” and providing a great experience off the get-go, but fairly little on the final moments of a user session. In the service industry, there are lots of great examples of this: the after dinner mint, United’s “PS” service from SFO to NYC where first class passengers were offered warm cookies just before landing, Gary Danko’s sending dinner guests home with a pastry for the next morning are just two examples. These are successful initiatives that appeal to the remembering self.

Square is a great example of nailing this concept. The receipts they send (your final moment in a Square experience) are beautiful, simple and elegant. I’m shocked that Apple and Amazon (two companies that do a staggering amount of transactions) haven’t followed suit yet. Currently, their receipts are cluttered and hard to read at a glance. One last good example is Top Chef’s Last Chance Kitchen which is a 10-minute “encore” competition where the chef that just got kicked off competes one last time (these segments are offered free on Bravo TV). Every a time a Top Chef episode ends I go through a little emotional roller coaster: excited to see who gets the boot, dismayed the entertainment is over and finally, thrilled when I remember there is a Last Chance Kitchen I now get to watch.

3) There is no value in adding “empty calories” to session length. In other words, don’t try to eke out additional minutes to your engagement metrics if those activities aren’t adding any new value to the user. Noah Kagan and the AppSumo team have a good YouTube post where they discuss the decision to significantly reduce email frequency (NOTE: AppSumo is a daily deals type of service tailored for businesses). They decided they didn’t want “empty calorie” emails that didn’t “wow” their customers.

There are no doubt innumerate other great examples that I’ve missed, but the key point is that when you design your product, consider not only the experience you’re delivering, but also the experience your user will remember.Understanding how they systematically differ (peak end rule, duration neglect) will help you optimize your product experiences for the human memory biases.

How to build solid roadmaps: Part I

Recently, I’ve spent more time mentoring and managing junior PMs and the other day one posed a deceitfully simple question: how do you build a good roadmap?

At first blush, it seems like an easy question to answer. Consider what your current customer base wants, what unmet needs they still have, what your product vision is, how it aligns with company strategy, what new capabilities exist (or will exist) that you can take advantage of, what your engineering team wants to build, your sales team, your UX team, and so on.

You could answer those questions and build a roadmap. In fact, you could just stack rank a list of features you have on your to-do list, team idea wall, hack-a-thon whiteboard, or brainstorming area. That would make a roadmap. But it’s liable to be an ineffective one, and at worst, a train wreck. A roadmap should be a cohesive plan – grounded in data and beliefs – that will take the product forward in building the customer base and delivering more value.

You need a framework to get there.

One way to think about this is in terms of trends and beliefs. Beliefs come in two forms: what you fundamentally believe (read: I believe we need to deliver X) and what your customers believe (read: I need this product to do X). Trends also come in two forms. There are internal trends, which you can tease out from your usage data, statistics, and historical performance. And then there are external trends, like the fact that the usage of tablet computers is growing at X rate YoY (Brad Feld has a good post on identifying external trends here).

If you map this out, it looks like this:

The framework does *not* provide the answer. Rather, it serves as a forcing function to get yourself thinking about the right issues. For example, do you (your team, your company) want to put all the proverbial eggs in one quadrant (see Instagram example below)? Or are you under-representing a quadrant that you do believe to be important? If so, the framework acts as a catalyst for highlighting the trade-offs implicit in any roadmap (eg, Are we too focused on quadrant X to the detriment of quadrant Y?).

The purpose of the framework is *not* to advocate for a standardized pattern across the quadrants (eg, A balanced 4-quad approach is better/worse than an imbalanced 1-quad approach). Rather, the goal is to enforce clarity and alignment around the strategic bets you’re making.

The framework is also helpful for what it doesn’t specifically emphasize: what your manager wants, what your competition is doing, what other internal teams are doing, other startups, etc. Certainly all of these other sources *can* provide roadmap inspiration but they should only do so if they are congruent with your trends and beliefs. If your manager (board, investors, etc) are advocating a feature or project, but you don’t believe in it and your customers don’t need it, why build it?

Here are a few examples and where they fit in the framework (some are bigger than just a feature but help illustrate the concept):

Instagram: Founded as Burbn, an HTML 5 based Foursquare-esque local, social network, Instagram found success when it listened to internal data from customers which suggested all they wanted to do was share photos. (Customers beliefs + Internal trends)

Evernote for Food: Based off Evernote’s mission of being your brain’s external hard drive, Evernote for Food melded this core concept with the growing external trend of sharing food photos, moments, etc.  (Your beliefs + External trends)

The iPad: Jobs held a strong belief in the importance of tablet computing despite no positive external trends (Microsoft’s tablets had failed earlier) and no customers banging down the door. (Your beliefs + Internal trends*)

Twitter’s Hashtag: Started by Twitter power user Chris Messina, the hashtag took off and was eventually “productized” and built into Twitter (Messina’s post on the evolution of the Twitter hashtag). (Customer beliefs + External trends)

Part 2 now posted here! (Thanks again to the @Quibb community for great feedback on Part 1)

*NOTE: The iPad example likely belongs in the very, very upper left quadrant where a visionary like Steve strongly believed in the opportunity and that was the sole driving force (eg, He believed in it and served as his own internal data). The x-axis of trends can be thought of as a continuum from the extremely internal (only 1 person knows it) to the external (everyone knows it).