Friction: what is it good for? Quite a lot it turns out…

15 Min Reading Time

Part II of my series on Friction

In Part I of this series, I explored the concept of Cognitive Friction and highlighted the difference between “good” Friction that serves a useful purpose and “bad” Friction that does not.  I noted that, all too often

Friction is deployed unknowingly, unthinkingly or without the designer fully understanding the nuances of the impact it will have on the target population.

In this, Part II of the series, I want to tackle that challenge.  Firstly, by equipping my readers with an understanding of what “good “uses of Friction are so that they can assess whether the ways in which they are currently deploying it are appropriate and effective. Secondly, to help them identify situations that are relatively frictionless, where perhaps the introduction of Friction might serve a useful purpose.  Finally, I hope it will also reveal where they are unknowingly deploying Friction and could review its use.

As with most behavioural interventions, the determination of whether a particular use of Friction is “good”, or indeed “fair”, is a matter of perception. Context also matters; the use of a specific form of Friction may be entirely acceptable in one situation, but unacceptable in another.

In this blog, I will cover three broad categories of Friction:

  1. Preventative designed to stop the target audience (“users”) from doing something undesirable;
  2. Qualitative designed to improve the quality of the user’s decision-making; and
  3. Engaging designed to improve the user’s engagement.

These are not “neat” categories; some uses of Friction will focus specifically on one of these objectives, others will deliver two or even three. While I am attracted to the concept that Friction, like Smartphone Apps, should focus on doing one thing well, I also think there is a strong argument that meeting more than one is also a good thing. After all, if we are going to slow something down or make it harder, then it is easier to justify doing that when we’re meeting several objectives rather than just one.

I should add that the categorisations I have used for Friction intentionally focus on the desired outcome, rather than the mechanics used to get there. Readers familiar with Daniel Kahneman’s Thinking Fast & Slow, might want to think about the interplay of the various examples with System 1 and System 2 thinking. I won’t explicitly cover it here, because I want to look at it later in this series.

The examples I give will lean heavily, but not focus exclusively, on the digital environment (“Digital”). Not because the techniques aren’t relevant in the physical world (they are!) but rather because Digital makes it far easier to design and implement Friction. It also allows for a degree of flexibility and personalisation that the physical world renders difficult, impossible or prohibitively expensive. That said, many examples of Digital Friction lever the same thinking as things previously deployed in the physical world. To prove that point, we’ll begin our review of the different kinds of “good” Friction with a modern-day version of a form of it that we can trace back to the 7th Century BC.

1. Preventative Friction

As the name implies, Preventative Friction is designed to prevent an undesirable outcome.

1.1  Prevent Unauthorised Access.

The idea of barring certain people from having access to a physical location is not new.  Hence the invention of the lock and the key; the earliest known versions of which were traced back to Nineveh in Ancient Assyria.

Although this function is increasingly being performed by keypads, fobs and swipe cards, the principle remains the same. In Digital, we use logins, passwords and biometric security devices to restrict access to legitimate users. We’re even using Friction to keep out machines in the form of bots, through the use of CAPTCHA boxes that require us to prove that we’re actually human:

You’ll definitely have come across CAPTCHA, but do you know what it stands for? The answer is Completely Automated Public Turing Test To Tell Computers and Humans Apart. So now you know. Sometimes they use pictures rather than words. To understand the fascinating reason why I recommend this article.

1.2. Prevent Errors

Our second form of Friction will also be well known to you. Try to delete a file, and you’re almost bound to come across a message like this:

This form of Friction is designed to prevent us from making mistakes.  It is so easy to implement that system designers now routinely use it to guide us through decisions where the worst case outcome is that we are inconvenienced or confused.  After all, a deleted file can usually be recovered if the deletion was in error.

Preventative Friction is also deployed in more critical situations where the consequences of the decision are irrevocable, and/or where there is a chance that the user might not fully comprehend the magnitude of the impact of their actions.  Like this:

Image result for warning sign this cannot be undone

To achieve their objective, the two previous examples rely on the user being able and willing to reject a particular outcome. But even when those presumptions might not apply, preventative Friction can still play a useful role.

Here’s a compelling case study involving “over-the-counter” (OTC) drugs; those that are available for purchase without a medical prescription. Because they are easier to purchase than prescription drugs, there is a higher risk that people using them will overdose; either accidentally or on purpose.


To reduce this risk, the UK imposed packaging requirements on manufacturers. As this NY Times article explains:

In September 1998, Britain changed the packaging for paracetamol, the active ingredient in Tylenol, to require blister packs for packages of 16 pills when sold over the counter in places like convenience stores, and for packages of 32 pills in pharmacies. The result: a study by Oxford University researchers showed that over the subsequent 11 or so years, suicide deaths from Tylenol overdoses declined by 43 percent, and a similar decline was found in accidental deaths from medication poisonings. In addition, there was a 61 percent reduction in liver transplants attributed to Tylenol toxicities

This is a really smart intervention. On the one hand, the packaging serves to deter people from overdosing, either deliberately or accidentally, in a way that selling it in a bottle would not. But the Friction it creates also serves a useful purpose for the average user; the blister pack gives a clear visual indication of how many pills the individual has taken, making it easier for them to medicate in the right dosage. An excellent example of Friction serving more than one purpose.

1.3 Improve Data Quality

Friction can also be used for data quality assurance. Fill a form in on a website, and you’re likely to find the data you enter being subjected to Form Validation checks, which ensure the accuracy of the information that is entered. For example, to prevent a user from registering with an email address that has already been used, or to ensure the address is in the right format. Like this:

Image result for your passwords don't match

In this example, Form Validation operates in two distinct ways. Firstly, it introduces Friction on a reactive basis, when it detects an error in what a user has typed. Secondly, it does so on a pro-active basis by requiring the user to re-enter their chosen password twice. The latter is necessary because the system has no way of knowing whether the password the user has typed in is what they intended; hence it requires it to be entered a second time.

Typo prevention isn’t the only use-case. Enter an event on a date that has already passed and many Calendar programs will ask you to confirm that you really did mean to create an entry in the past. The presumption on the part of the designer, being that most people won’t want to do this; unless, of course, the user is Stephen Hawking, attempting to discover whether time-travel is possible.

1.4 Deterrent Friction

To experience the next type of Friction, I recommend trying to get into a nightclub without being on the guest list. You’re highly likely to find yourself standing in a line for an arbitrary amount of time before you’re subjected to assessment by the door staff who will determine whether or not they’re happy to let you in, based on a set of criteria that appear to be modelled on the whims of Roman Emperors.

Screenshot 2019-01-27 at 17.03.25.png

Arbitrary or not, it’s an excellent example of Friction which is mostly there to deter or prevent undesirable users. In situations where something is offered, either by design or unavoidably, to a larger population than is the intended target, Friction permits screening to occur. Nightclubs need to publicise their existence widely to attract people who will want to go, but they don’t want everyone who knows about them to come in. Of course, it isn’t the only reason they make people wait in line; we’ll come on to those later on.

The use of Friction as a deterrent is explored by Cass Sunstein in his research on “sludges” in the context of benefit payments:

If the process of getting benefits is too easy, people who don’t actually need the benefits may be incentivised to apply for them as well. As a result, those who need them most may not get the benefits because the money has already been given to someone else. Adding in a bit of sludge, like annoying paperwork, could stop those who don’t actually need benefits from trying to get them.

But does this count as sludge? While this practice doesn’t look after the self-interest of every single individual, it does look after the self-interest of certain individuals: those who are most in need.

The obvious counter-argument to this form of Friction is that it sounds good in principle, but there are likely to be unintended consequences. If you make claiming benefits too hard, you will deter those that don’t need them, but you’ll also put obstacles in the path of genuinely deserving people who might be least able to afford the time and cognitive load required to navigate the sludge. However, outside the context of social security, there are other valid reasons for deploying this kind of Friction.

Take Snapchat. For reasons I’ll come on to, it is highly likely that many of my readers might not know what it is. Those that do can feel free to skip on to below the Snapchat logo. For the uninitiated, Snapchat is a multimedia messaging app that allows users to send “Snaps” (pictures and short videos) which are automatically deleted after a pre-set amount of time; immediately once they have been viewed by all participants or after 24hours if the sender chooses this option.

The reason I suspect that some readers won’t be familiar with Snapchat is that was intentionally designed to target younger users, and implicitly exclude older ones. Both in the user proposition, but also in the execution of the idea. Most obviously in the UX (User Experience) which can be incredibly frustrating for anyone who is, to make a sweeping generalisation, aged 35 and above. Snapchat’s CEO, Evan Spiegel, has claimed that this is on purpose, insisting at a conference that:

This is by design. We have made it very hard for parents to embarrass their children.

In other words, they can’t stop parents from downloading it, but they want to deter them from using the service. Don’t believe me? If you’ve not tried it, give it a go. While some of you will find it easy, it’s likely to put many of you off. Admittedly, as this article highlights, SnapChat has recently made some changes designed to address this issue, but the fundamental Frictions that were there in the inception of the App remain and are very powerful.

Community websites face the same problem. They want to be accessible to everyone on the internet, to reach their target audience, but they want to control who participates in their community. Product Hunt, a tech review website, makes aspiring contributors jump through a number of hoops.  They explain their reasons on the right:

Image result for product hunt friction By effectively increasing the barrier to entry through Friction, they are able to weed out users who don’t show sufficient commitment to the cause. Contrast that approach with frictionless signups to social networks like Twitter where the upside of no user screening is that everyone can join and share their wisdom. The downside is that everyone can join. Including ill-informed Dunning-Krugerites who can spout nonsense to a global audience.

 2. Quality inducing Friction

2.1 Improve the Quality of Decision-Making

Having filtered out undesirable users, let’s now look at how we can use Friction to help improve the quality of the decision-making of the users we do want. The best example of this is the “cooling off” periods which regulators require financial services firms to offer their customers for certain products.

Photo by Tomek Baginski on Unsplash

Studies have shown that these can be very effective. A recent one entitled, “Cooling Off and Backing Out” examined customer rescission (i.e. the reasons why consumers rescinded) during cooling-off periods and identified that they tended to do so for very compelling reasons, including:

  • “a mismatch between product features and personal circumstances” – in other words, the product didn’t meet their needs;
  • post-purchase concerns about product value” – the product was poor value for money;
  • reassessment of financial capability“- they couldn’t actually afford it;
  • reflections on sales presentations” – they realised they’d been fooled by a slick sales pitch; and
  • cautionary influences of reference groups” – they consulted with friends or family who persuaded them it wasn’t a good idea.

What this highlights is that if you give people more time to think about something, they’re very likely to make a better decision. Or at the very least, avoid making a really poor one. Because it isn’t possible to give them thinking time before they buy a product, regulators use Friction to create it artificially after the event.

Friction isn’t just useful for influencing decision-making around products that might inherently pose a challenge for consumers. It can also help when the rationale for making a purchase is more impulsive; particularly when the drivers for that impulse are strong emotions.


Research published in the US last year, highlighted that forcing people to wait before they can acquire a firearm, reduced gun homicides by 17 per cent with no significant increase in other types of homicides. In other words, not being able to buy a gun straightaway, doesn’t mean people just “displace” and commit murder by other means. The reason for this, is the power of emotions in our decision-making, as this Behavioural Science magazine article explains:

 intense emotions like anger and sadness—“visceral factors,” in academic language—can cause people to take actions they later regret, such as resorting to gun violence.

Yet…these emotions are often transitory. Given sufficient time to cool off, the types of intense negative emotions that lead to violent tendencies can pass.

This suggests that inserting even a short delay in the gun-buying process has the potential to reduce gun violence, without restricting anyone’s right to own a gun.

The concept of a cooling off period is not new.  The Preußische Beschwerdeordnung was a Prussian Army law that required anyone who wanted to file a complaint to their superiors to wait for a certain amount of time (the so-called “Prussian night”) after the complained-about incident occurred. A cynic might conclude that this was less about ensuring the quality of complaints was high and more about the introduction of Sludge to keep the number down. Perhaps it was both.

2.2 Improve Recall

Struggling to remember things you’ve read? Try Sans Forgetica, a downloadable font created by Melbourne-based RMIT University’s behavioural business lab and design school, that is scientifically designed to help students remember their study notes.  It looks like this:

The font deploys something called “desirable difficulty”. What we know as Friction. As the designer explained to the Guardian newspaper:

When we want to learn something and remember it, it’s good to have a little bit of an obstruction added to that learning process because if something is too easy it doesn’t create a memory trace.

If it’s too difficult, it doesn’t leave a memory trace either. So you need to look for that sweet spot.

The idea of finding balance in creating Friction is also relevant to our next category.

3.Engagement driving Friction

3.1 Goldilocks Tasks

In his book Drive: The Surprising Truth About What Motivate Us, Daniel Pink refers to what he refers to as  “Goldilocks tasks”:

challenges that are not too hot and not too cold, neither overly difficult nor overly simple.

He explains why these are important:

One source of frustration in the workplace is the frequent mismatch between what people must do and what people can do. When what they must do exceeds their capabilities, the result is anxiety. When what they must do falls short of their capabilities, the result is boredom. But when the match is just right, the results can be glorious. This is the essence of flow.

A need for balanced Friction is also a key component in Gamification; the application of techniques commonly found within game environments which are widely used to drive user engagement.  It works because we’re naturally drawn towards playing games; whether on a computer screen, the page of a magazine, a board, or on a sports field.

Friction is a pre-requisite for games. Without challenges for the players there would be no reason to play, but make them too difficult, and players will give up.  Getting this balance right is the essence of a good game.  Take the board game Othello:


(c) Mattel


Part of the reason it’s been so successful, with reported sales of over 40 million units, is that beginners are attracted by the apparent simplicity, yet there are enough complexities in the game to retain the interest of advanced players. All nicely encapsulated by the game’s tagline: “A minute to learn…a lifetime to master!”. 

Placing value on things we spend time on, is a key driver behind our next use-case for Friction: driving user engagement.  We’ll begin with a store whose product offering is about as Friction-filled as it gets.

3.2 Ikea Effect

If you’ve ever had the pleasure of going shopping in Ikea, you’ll know what I mean. You join 100s of other shoppers in entering a lengthy process that involves shuffling along a path littered with attractively priced temptations. It’s no wonder people reward themselves with Swedish meatballs in the cafe at the end of it all!

But the Friction doesn’t end there. Because when we get our flat pack furniture home, we have to begin the process of assembly.  Ikea tells us that selling furniture this way allows them to sell it more cheaply; they can transport more cost-effectively, and the labour effort to build the furniture is transferred to the customer.  In a, perhaps surprising discovery, it turns out that there’s an additional benefit.  Meet the Ikea Effect:



Researchers discovered that because we place disproportionately greater value on things we’ve expended effort in creating ourselves, the Friction that the need to assemble creates actually serves to make us more positively disposed towards our furniture. A double win for Ikea and a prime illustration of Friction counter-intuitively building user engagement.

3.3 The Locksmith’s Paradox

It isn’t just our own time that we think of in these terms. We apply similar principles when others do things on our behalf, and we tend to value other people’s work by the effort we perceive that they’ve put into it. Usually, we’ll measure this in time, but it might also be based on arbitrary metrics like how much they sweat if physical labour is involved. As a result, if someone makes something look easy, then we won’t value it as much as something that seems hard.

Rather than judging it by the utility we gain from the labour (i.e. the output), we judge it by the perceived level of input. What this means is that we can find ourselves inadvertently rewarding incompetence more highly than competence. This is known as The Locksmith’s Paradox, for reasons that Dan Ariely explains in this video.

Image result for locksmith

Many companies exploit this to make their product seem more valuable. Starbucks requires its baristas to steam each customer’s milk individually; it takes longer but feels more personal, and so we value it more. Travel comparison site Kayak deliberately creates a pause before it delivers results to its users. As this HBR article explains, we actually appreciate something more, if it takes longer and they’re told why.

Conversely, my local Thai restaurant tells callers placing takeaway orders that their food will be ready in 10 minutes. Rather than being grateful for being fed so quickly, this makes me nervous; I expect restaurant food to be “cooked” not “heated up”. As a result, they’ve lost my business. What they should probably have done is create Friction by telling me it would take 30 minutes and then just delaying their preparation process for 20 minutes.

elli-o-65548-unsplash (1)

3.4. Exclusivity

The Ikea Effect and Locksmith’s Paradox also illustrate another use case for Friction: creating a perception of exclusivity. One of the other reasons the nightclub I referred to earlier, wants to have a long line outside is to create the illusion of being a highly desirable exclusive venue.  The queue of people signals that it is popular and that people are prepared to stand in line to get in, and the slow speed of entry and selection criteria gives a sense of exclusivity.  All of which will make the experience feel much better for you if you’d just been able to stroll in.

Retailers use Friction to create exclusivity around ‘limited edition’ products that are hard to get hold of while airline frequent flyer schemes only offer Olympic medal coloured statuses to passengers who’ve spent long periods in the air with them.  Both require commitment, which in turn strengthens the user’s perception of exclusivity.

gold-colored Olympics medallion

3.5 Creativity

Plato once said that “necessity is the mother of invention”.  In doing so, he helpfully highlighted one of the more surprising benefits of Friction, which is that it can promote creativity.

An interesting example of this is the term “workaround”, which has both positive and negative connations.  As some research into Nurses’ workarounds in acute healthcare settings highlights, workarounds can:

solve problems; sidestep ‘problematic’ rules; bypass workflow blocks created by safety mechanisms; address poor workflow design and organisational and system issues; save time; backup software data applications; compensate for inadequate technology; patch software glitches; or offer solutions to a range of problems including shortcomings in staffing, equipment and supplies

Each of these types of workaround, whether “good” or “bad”, are driven by a desire to :

circumvent or temporarily ‘fix’ an evident or perceived workflow block

That evident or perceived workflow block is what we refer to as Friction.

We can also see how Friction encourages creativity in the kitchen.  Take ready-meals, which are, in effect, an almost frictionless form of “cooking”.  What they offer in convenience, is offset by the minimal scope for personalisation.  In direct contrast, cooking from a recipe is a process filled with Friction, but which allows the chef latitude to experiment and tailor the dish to their own desires. In both cases, the amount of Friction in the process is proportionate to the amount of creativity the option permits.

Creativity can also be driven by appropriation; using something for a purpose for which it was not designed.  The word is most often used in negative contexts like “cultural appropriation” and “misappropriation”, but that doesn’t mean that all appropriation is automatically wrong. A chair isn’t intended to be used as a ladder, but people often successfully use them in that way, when there is no ladder to hand. As with workarounds, appropriation is encouraged by the existence of Friction; whether the outcome is a good or bad thing is a matter of context and perception.

This doesn’t just apply to physical items. In research entitled “Seamful Design in a Seamful Society, Karin Anderson explores how appropriation also applies to thought:

An example of appropriation is the use of mobile phones. One of the most common seams for a mobile phone is irregular network coverage and local variations in signal strength.

These seams are commonly accepted as reasons for not answering or dismissing a phone call, thus giving the user the opportunity to exploit knowledge about them. A user can adapt his or her behavior by pretending to have bad network coverage (for example in tunnels) or poor reception if he or she does not wish to talk to someone.

Clearly, the use of Friction to drive creativity is more applicable to some situations than others.  We probably wouldn’t want people in charge of a nuclear reactor or air traffic control, to have the latitude to try new things “on the fly”.  Much will also depend on the skillset and experience of the users; allowing a Michelin starred chef to experiment is a good thing, whereas McDonald’s needs to operate on a more controlled basis.

Over to you

So there we have my list of reasons why Friction can be a good thing. I stress “can” because I recognise that the appropriateness of a particular deployment of it, is context dependent.  As with other forms of Behavioural Intervention, the designer’s perspective on what is appropriate might differ from how the user feels about it.

If you think I’ve missed a form of Friction, let me know.

In the next in this series, I’ll look at how we can deploy BeSci to help us when Friction is undesirable, yet unavoidable.


Familiar with Friction? You should be. Why Risk & Compliance Officers need it in their toolkit

Estimated Reading Time: 10 Minutes

Ask any Behavioural Scientist how you can get people to do what you want them to, and you’re very likely to get the same answer. Here it is courtesy of Richard Thaler, the Nobel Prize Winning “father” of Nudging:

“The mantra I always give, is that you have to make it easy”

Given I’m a proponent of bringing Behavioural Science (BeSci) to Compliance, it might, therefore, seem odd for me to focus on precisely the opposite; making things harder. The answer, as is often the case with BeSci, might seem somewhat counter-intuitive.

Photo by Nick Abrams on Unsplash

Slowing things down or making them harder by introducing “Friction” can actually be a good thing. Using it as a restrictive tool to make something we don’t want people to do, less straightforward, is one obvious way.  But it can be equally useful, in situations when the activity in question is something we do want them to do. By understanding the full potential of Friction and deploying it intelligently, Risk & Compliance professionals can arm themselves with an incredibly powerful risk management tool.

However, all too often within organisations, Friction is deployed unknowingly, unthinkingly or without the designer fully understanding the nuances of the impact it will have on the target population.

Increasingly, the proliferation of Friction is blamed on control functions. Whether or not they are actually responsible. Sometimes, it is the processes built or curated by these functions that are guilty of over-deployment of Friction. However, on other occasions, they, along with the bête noire that is regulators, are convenient scapegoats for inefficiencies caused by other parts of their organisation. By highlighting the issue of Friction, I want to help my profession make better use of it. That way we can avoid further tarnishing of our brand and deliver better outcomes, more cost-effectively.

I also want to help my Risk & Compliance peers be more aware of how Friction works in reality, rather than in theory.  This will also allow them to identify poor uses of it by other parts of their organisation. 

Because a mis-deployment of Friction can hugely increase both the Operational and Human Risk profiles of the organisation.  Put the wrong kind of Friction in people’s path, and they’ll find ways around it that are not in the organisation’s interest.  While the right kind of Friction can enhance risk management and inspire innovation. 

In this first blog on Friction, I’ll explore how it arises and highlight examples ranging from “virtually frictionless”, to “good” and “bad” Friction. In later blogs, I’ll explore the impact Friction has on the target audience and then identify ways in which we can deploy it as a powerful risk management tool. To begin with, let’s look at what the word means.

What is Friction?

In the physical world, we define Friction as follows:

(c) Melloo

As the Friction I’ll be referring to is cognitive rather than physical, we need a slightly different definition. While the idea of “slowing a moving object” down is still relevant, cognitive friction isn’t as quantifiable as physical friction. We can measure the Friction on a banana skin, but the Friction inherent in a cognitive process is dependent on the perception and experience of the individual involved.

As a loose metaphor, think of it this way. If I’ve never done something before, then it’s likely I’ll encounter more Friction in the process, than someone who is an expert at it. You only have to watch television cookery or home improvement shows and then try to replicate what is made to look simple on screen, to understand this point. 

For that reason, I prefer to think about Friction along slightly different lines to the physical definition above. For these purposes, what I mean by the word is closer to the way it’s described in a book called, appropriately enough, Friction


(c) Friction: Passion Brands in the Age of Disruption by J. Rosenblum & J.Berg 

That doesn’t mean we can’t bring any science into this. Part of what we’re dealing with here is explained by Hick’s Law which states that the more choices you present people with, the longer it will take them to reach a decision. This helps explain why technology companies are often fixated on removing as much Friction as they can. In the words of Facebook’s Mark Zuckerberg:

Our goal is to make it so there’s as little friction as possible to having a social experience

Their key focus is on User eXperience (UX). That’s because poor UX results in Friction which deters users. For my more scientifically minded readers, here’s one UX designer’s take on Hick’s Law:

(c) Kristina Szerovay 

We’ll begin our exploration of Friction by looking at how one tech company removes as much of it as it possibly can.

Close to Frictionless

That company is: 

Amazon Logo

Not only exemplary in reducing Friction, they’ve also been helpful enough to talk about the techniques they use, so we have some useful insight. As highlighted in this fascinating article, founder Jeff Bezos gave his team a specific vision for what he wanted:

“We need something to make the ordering system frictionless. We need to make it so the customer can order products with the least amount of effort. They should be able to click on one thing, and it’s done.”

What came out of this was something called “1-Click Checkout”. 19 pages of patent and hundreds of lines of code later and Amazon was able to deliver the feature that Bezos had asked for. Of course what went on behind the scenes was invisible to customers; all they knew was that if they clicked the button, they could buy things with just a single anyone who has used Amazon can testify, this makes shopping dangerously easy. Providing, of course, you’ve previously entered your payment details and delivery address. In reality, then, they haven’t eliminated Friction entirely but have reduced it to the bare minimum necessary.Amazon’s other products serve a similar purpose.

The Amazon Echo might be sold as a voice-controlled smart speaker, but the real end game for the Alexa virtual assistant is to provide a voice interface for…you guessed it: more shopping.  Voice is even more frictionless than a keyboard and mouse or smartphone screen.

One day, the Internet of Things will mean we’ll have washing machines that can order their own liquid supplies. Until then, Amazon is bridging the gap with has “Dash” buttons. These allow us to order things, from Amazon of course, “in situ”. After all, we’re most likely to remember we need more washing powder when we’re in front of the washing machine. 

(c) Amazon

Amazon clearly takes Friction seriously; it takes time and effort to conceptualise and then implement interfaces of this kind. In an interview, Amazon Director Kintan Brahmbhatt explained the taxonomies (my word not his) that they use to categorise the different causes of Friction that they have identified:

Source: First Round Review


What I like about their approach is that it is user-centric. Friction either exists to help the user; what Brahmbhatt calls “by design”. Or it doesn’t, in which case it is something that they seek to eliminate.  Their approach is based on understanding how users think:

This is what we’re trained to do as human beings — take the easiest route. Reducing friction is all about creating a path of least resistance. 

To help find it, Amazon has identified three factors that induce a perception of friction on the part of the user.  

The first is anxiety, caused by having to make decisions and from a cognitive bias called Loss Aversion. We all understand the former; taking decisions requires effort. Loss Aversion is the idea that we intuitively feel losses disproportionately more than gains, so if in making a choice we risk losing something, it weighs more heavily on us.  Of course, most choices do precisely that: in a choice of A or B, we lose one of the options.

When Amazon first introduced digital music, consumers who had historically bought CDs and vinyl might reasonably resist the new format. After all, you’ve invested in your existing collection, and digital might not be as good as what you’re used to hearing. So to remove anxiety on the part of customers, Amazon launched a feature called AutoRip where they automatically gave people digital versions of any music they bought in physical format.

(c) Amazon

What’s more, they did the same for all the music people had already bought on Amazon.  This meant customers inherited a pre-existing library (at least for what they’d purchased on Amazon) and no longer had to choose between formats.  Above all, they wouldn’t feel a sense of loss by missing out, and as a result, Amazon helped make the shift to digital as frictionless as possible for the customer.

I discovered recently that they do something equally clever with books. My local bookshop didn’t have something I wanted, so I ordered a physical book from Amazon. Because a Kindle version of that same book is available, Amazon also let me read the first few chapters for free straight away.  A cunning way of eliminating the Friction of having to wait for the book to arrive, which also introduces buyers of physical books to the benefits of e-books. Of course, this costs them nothing, because all they’re doing is giving me the free sample that I could get for myself from their site. But it feels like they’ve made my life better and added value to the purchase I made.

The second factor is reducing the cognitive load on users.  In other words: the amount of effort something requires. Amazon only asks you to make decisions if really necessary, and each decision is carefully framed to be as straightforward to understand as possible. Take a guess at how many products they have on sale. It’s probably a lot more than you think. This chart shows the size of the product range a year ago:


Under normal circumstances, offering people that many products could be somewhat overwhelming, thanks to something called The Paradox of Choice. We think we like optionality, but in reality, we don’t.  This was wonderfully illustrated by some research that experimented with offering consumers a more or less extensive selection of jams to choose from.  Surprisingly, selling a smaller range, meant people bought more. That’s because we find making choices with lots of options difficult and we’d prefer to walk away. 

Amazon counteracts this with menus that are extremely easy to navigate:

(c) Amazon

The final factor that induces Friction is minimising the need for context switching.  This is arguably a subset of reducing cognitive load. Put simply it means “letting people do everything in one place” without requiring them to go elsewhere. Storing the customer’s credit card details means they don’t need to go to their purse or wallet to retrieve it. All of Amazon’s services are available direct from their website. Order a Kindle book, and you can read it right there without needing to load an App or change device. 

Compliance as Friction

Next time you’re on the Amazon website, think about how easy it is to find what you want and order it. Now imagine we could achieve similar aims in meeting our Compliance objectives. It sounds crazy and yes there are differences between what Amazon is trying to deliver and what the Compliance function in regulated industries is there to do. But that doesn’t mean we can’t learn from them and that we shouldn’t try to adopt some of their user-centric philosophy. Particularly in how they analyse Friction.

Notice how the criteria they outlined was from the perspective of the customer, not Amazon. What they look to address are things that make it harder for the customer and, critically, things that make the customer feel like things are hard. Customer perception might not match reality, but that’s irrelevant. It’s like justifying an offensive comment by saying  “I didn’t mean to offend you” and assuming this automatically means something isn’t offensive to someone else. The intent is irrelevant when it comes to how people perceive things. When it comes to Friction, perception matters a lot.

I think we can learn a lot from Amazon because we’re both in the business of trying to influence human decision-making. Amazon wants to make it easy for people to buy things. We want to make it easy for people to be compliant. The processes we deploy to manage risk can cause similar psychological responses in our target audience.  Yet we don’t often think about things like anxiety or cognitive overload when designing Compliance processes.  This is a missed opportunity.Sometimes we’ll want to minimise Friction, just like they do. But on other occasions, we’ll want to actively deploy it to achieve our goal. 

With that in mind, it helps if we understand what “Good” and “Bad” Friction look like.  We’ll start with the Bad.

“Bad” Friction

You’ll intuitively know what “Bad” Friction is.  So I won’t spend as much time explaining it as I did with the Amazon case study. The most obvious example is what Cass Sunstein calls Sludges; so called because they’re the opposite of Nudges. 

“The term should be taken to refer to a kind of friction, large or small, that people face when they want to go in one or another direction. People might want to cancel a subscription to a magazine in which they no longer have the slightest interest, but to do that, they might have to wade through a great deal of sludge”

Sadly Sludges are commonplace, and we’ve all been on the receiving end of them. For more on Sludge, I recommend this helpful summary by Wilte Zijlstra, an enlightened BeSci practitioning Dutch regulator. 

Though I was pleased to see a recent announcement from Mastercard with the headline “Free Trials Without The Hassle”.  In it, they stated that they will now prevent merchants from billing users at the end of a free trial, without getting their approval.  Sadly and inexplicably, its only for physical purchases, but its a start.

The types of situation that Mastercard is, belatedly, looking to prevent are obviously intentional; the person who has designed the Sludge, purposefully wants to make it hard for the end user. However, its also possible to achieve suboptimal ends, with the best intention.

Take password security. For understandable reasons, websites require us to enter passwords with a certain level of complexity to make them less hackable.But you don’t have to go too far to find a website that takes this idea a little too far and imposes password requirements that are feverishly difficult to comply with.

Like this example from the Attorney General of Texas Child Support website. It’s worth reading them in detail to understand quite how ridiculous they are; it’s not even clear what are requirements and what are recommendations.  What is clear, is that picking a password is going to be hard.  And good luck if you forget it and need a new one!

Source: AG of Texas, Child Support website. Image capture by ITSP Magazine

The end result here is user frustration and, ironically, a higher likelihood of security being compromised as people write down their fiendishly hard to remember passwords. And that’s before we consider what impact the sludge might have on the process of claiming child support, which appears to be the purpose of the website. It’s almost as if they don’t want to pay any out!   

As I said at the outset, “Bad” Friction is something we’re all familiar with.  But that doesn’t mean that all Friction is bad.  Let’s now take a look at an example of what I think is exemplary Friction. 

“Good” Friction

It comes, perhaps surprisingly, from a tech company.  Even more surprisingly, it’s from a company that hasn’t exactly always had a good reputation. Though as we’ll see, the Friction was introduced in response to a customer backlash, so when I give credit, it’s not necessarily for coming up with the idea.  The execution, however, works well. 

The company in question is Uber, and the Friction is introduced whenever “Surge Pricing” kicks in. Depending on your perspective, Surge Pricing is either a way of ensuring that there is always a supply of drivers available to meet customer demand (the Uber position) or a piece of usurious price gouging (the argument used by quite a few customers).

It works as follows: if the demand for cars exceeds supply, Uber puts prices up to try to recalibrate supply and demand.  Or to increase their profit margins. Take your pick. Following complaints by customers, Uber introduced additional confirmation steps to ensure that users couldn’t plead ignorance of what multiple of the standard price they would be charged for their journey.

Order a car when surge pricing is in force and two things happen. Firstly you’re presented with the multiple of normal fare price you’ll be paying which you have to click to agree to. Then you have to re-type that multiple yourself.

Source: Forbes

As a result, riders can be in no doubt, before they get into the car. They might not like it, but they will be aware of it. It’s two small steps that slightly delay a generally friction-less process and in doing so discharge a moral responsibility.  For that reason, I see this is “good” Friction. 

Uber isn’t the only tech company that sees benefits in introducing more Friction. In an article entitled The Internet Needs More Friction, Justin Kosslyn argues that the obsession with frictionless services has resulted in some bad outcomes:

The internet is facing real challenges on many fronts. If we truly want to solve them, engineers, designers, and product architects could all benefit from the thoughtful application of friction. The philosophy of the Internet has assumed that friction is always part of the problem, but often friction can be central to the solution.

What’s remarkable about the article is that Kosslyn leads product management at Jigsaw, a unit within Alphabet, who are the parent company of Google. Echoing this, a recent New York Times article headlined “Is Tech Too Easy to Use?” asked a simple question:

Could some of our biggest technological challenges be solved by making things slightly less simple?

In the spirit of that question, in the rest of this series of blogs, I want to explore how we might use an improved understanding of Friction to manage Human Risk.  

Until next time…

In my next blog, I’ll look at obvious and less obvious use cases for Friction and explore the impact these have on the target audience.  In the meantime, keep an eye out for examples of Friction in your world and share them with me. I guarantee that once you start looking for them, you’ll see lots of it about.


What a spying scandal in football can teach us about Compliance

Estimated Reading Time: 8 minutes

Usually when there are newspaper headlines involving football that refer to cheating, then it is reasonable to assume that they’re referring to player activity on the pitch.  However, this weekend in the UK we had an off-pitch incident involving Managers which the sports press somewhat predictably named “Spygate”.

Photo by Thomas Serer on Unsplash

Frank Lampard,  the Manager of Derby County, accused Marcelo Bielsa, the Manager of rivals Leeds United, of having sent a spy to secretly watch a Derby training session ahead of a game between both clubs.  A game that Derby went on to lose 2-0.

After the match, Lampard said that he felt that Bielsa had not acted in the spirit of the game. In response, Bielsa freely admitted that he had sent someone to spy and that he had been doing it for a number of years.  Bielsa, who is Argentinian, started working in the UK in the middle of last year having previously worked in a number of other countries where, he claimed, it was a common ‘tactic’. In his post-match interview he said:

“I found out by calling him that he thought I didn’t respect the fair play rules. If we take into account this fact, Frank Lampard is right, Derby County are right and their fans are right. The conclusion that has been drawn is that I had a sporting advantage and that my behaviour was not right. I can explain my behaviour but I don’t justify it because I have to respect the norms that are applied in the country where I work.”

Source: Yorkshire Evening Post

Note that word “norms”, we’ll come back to it later. Reinforcing what Bielsa said, Leeds United also put out a statement saying that they had subsequently reminded him of  “integrity and honesty which are the foundations of the Club.”

Bielsa hasn’t yet been charged with anything by the Football Association (FA) and I suspect that it’s unlikely he has actually broken any rules.  But it’s clear from the widespread condemnation he received and the Club’s response, that he has breached an unwritten code of conduct.

I’m fascinated by this from a Human Risk perspective.  During my hiatus from blogging, I’ve been working on “Bringing Behavioural Science (BeSci) to Compliance”; deploying BeSci to deliver better outcomes from the two disciplines I cover: Compliance and Operational Risk.  Both are largely focussed on influencing human decision-making; Compliance because organisations can’t be compliant of their own accord, its the people within them that determine that status.  Meanwhile, the largest single factor, whether root cause or accelerant, behind Operational Risk is people.

The Bielsa case highlights something I’ve been thinking about a lot: how can we get people to do what we want them to? In particular, how to write rules and policies that are more effective at influencing human behaviour.

Your perspective or mine?

Photo by Nadine Shaabana on Unsplash

Rules are most commonly written from the perspective of the organisation writing them, rather than the perspective of the target audience.  The implicit presumption being that  because a particular outcome is required by the organisation; “if we write it, they will comply”.  In predictable environments, this approach makes perfect sense. I want the ground and aircrews of the plane I’m about to fly on, to have completed standard pre-flight checklists. Regardless of whether or not they think it’s a good idea. Though obviously those flying with me have a huge incentive to do so.

However, in less predictable environments,this can lead to rules that are model answers to exam questions that codify a theoretical “black and white” world, rather than the “grey” real one in which people find themselves.  

To achieve this, the rule-writers often err on the side of writing longer, rather than shorter rules.  This makes logical sense.  Having lengthy all-encompassing rules seems like a good way to mitigate risk.  After all, if you cover everything, there’s no room for confusion. Ironically though, the more prescriptive the rules, the greater the potential risks.  

In a fast-moving and more complex world, there will increasingly be situations that people find themselves in, which won’t have been in the direct contemplation of the rule-writers.  In “rule-centric” environments, people are incentivised to behave like electricity seeking the path of least resistance and seek the “rule of least resistance”. 

As a result, “good” people can end up doing the wrong thing having tried to do the right thing by following the rules, whilst “bad” people can use this dynamic for arbitrage.

Perception is everything

Even where the rules do contemplate a particular situation, what is often also ignored is the target audience’s pre-existing perception of the situation and the authority that is seeking to control their behaviour.  If the target audience already has a view on how it would approach a situation then a rule that supports this is likely to gain more natural traction with them than a rule that goes against that.  But equally a rule that codifies something the target audience would do of their own accord can also backfire, if there’s a sense that it is patronising or trying to force them to do something they would do naturally of their own accord.  And if the target audience thinks the rule-makers have overstepped their authority, then there’s a likelihood of a backlash. 

This doesn’t mean just writing rules the target audience like.  What it does mean is giving careful thought to how they are likely to react and finding smart ways to codify rules that will maximise their propensity to comply.  If we want rules to be effective; then they need to be as user-centric as possible.

I recognise that in global organisations, this can be challenging. Not least because Social Norms may differ according to culture. But just because it’s difficult, doesn’t mean we shouldn’t think about it. By employing people, these dynamics exist whether we want them to or not.

In the Knowledge Economy, there is a trend towards hiring people to do the things that machines can’t.  These tasks involve cognitively advanced skills like judgement, nuance and intuition.  If we want smart people that have been hired for those skills to perform at their best, then we need to craft rulebooks that give due consideration to them as “thinking” rather than “unthinking” users.

The best example of the implementation of “thinking compliance” in a user-centric framework that I’ve come across is the one deployed by Netflix.  Their Freedom & Responsibility culture begins with the principle that there are only two kinds of non-negotiable rules: those that prevent irrevocable disaster; and those that cover moral, ethical and legal issues (for example bullying). Other rules can “shapeshift” according to circumstances. Hence their Expensing, Entertainment, Gifts & Travel policy is 5 words long: Act In Netflix’s Best Interests. Obviously, this leaves huge room for interpretation, but it’s not a rule that can be arbitraged or that any employee can deem unreasonable. Compliance becomes a matter of judgement and requires employees to justify their behaviour in absolute terms, rather than relative to a prescribed norm in the rule. It’s admittedly very bold and not something every company will be able to adopt, but it shows how rules can be crafted in a more user-centric way. For more on this, I recommend Powerful: Building a Culture of Freedom and Responsibility by Patty McCord, the former Netflix Chief Talent Officer.

The reason I suspect there’s no rule against what Bielsa did, is that the FA probably intuitively engaged in some Netflix style thinking. There’s no rule, because no-one ever thought it was necessary to have one.  After all, if the Social Norm is well understood, then why would you need a rule?  The answer might be that with an increasingly global market for players and managers, a domestic Social Norm might no longer be understood by everyone.

 To understand whether a rule is needed and if it is, how it is best written, it is necessary to have a good appreciation of the perception and perspective of the target audience.

Assuming I’m right that there isn’t a rule against what he did and the FA decided it did want to write one, how would they do it?  One thing’s for sure; it wouldn’t be easy. Don’t take my word for it.  Frank Lampard revealed precisely that in his post-match interview: 

“I don’t know what the rules are. I believe there’s not an absolute clear-cut rule about it but we can’t open the door to this thing happening every week. What kind of farce would that be, of everyone sending undercover people, drones, whatever, into training. It would be farcical. So something has to be done, I just don’t know what it is and it’s not my decision.”

Source: Yorkshire Evening Post

Perhaps unsurprisingly, what I’ve been discovering is that it’s very easy to write rules.  It’s a lot harder to write rules that are effective in delivering the intended outcome in all situations.  To misquote a well-worn aphorism: I’m sorry my policy is so hard to comply with, but I didn’t have time to write one that made it easy.

Bringing Science to Compliance

Fortunately, I’ve also been discovering that my hunch (that BeSci can help us solve this and other Compliance challenges) which set me off on my quest, is being proven to be correct.  In my research, I’ve come across some really innovative thinking.  Here are just three examples:

The first comes from a practitioner friend of mine, Maarten Hoekstra who has created the wonderfully named Broccoli Model for Compliance.

(c) Maarten Hoekstra

This user-centric approach delivers Compliance objectives by helping people to do the right thing, rather than just telling them what to do. By focusing on Moments, real-life situations in which people find themselves, and factoring in Social Norms, the Broccoli Model works with the grain of human thinking rather than against it.  To find out more and why Maarten gave it this unusual name, watch his presentation

The second is from the field of Behavioural Law. In The Law Of Good People, a book I highly recommend, Professor Yuval Feldman explores the factors we need to consider about “the rule of law in a world populated by individuals with different levels of awareness of their own unethicality”.

Two things he covered that gave me particular pause for thought.

Firstly, there is a temptation to talk and think about Compliance as if all requirements are designed with the same intent. They’re not. Take Whistleblowing as an example; you want people to feel incentivised to whistleblow if, and only if, they come across situations that necessitate it. You don’t want everyone doing it all of the time. In Feldman’s words, this is a situation where “it only takes one to help”. Contrast that with other things which you need everyone to do all of the time, because “it only takes one to harm”. What we’re asking people to do in these two instances is very different, yet we tend to incentivise both behaviours in the same way. Not least in the language we use to write the rules.

Secondly, Feldman opened my eyes to the concept of “Quality” of Compliance which I’d never really thought about before. Some tasks are binary in nature; you either are or are not, compliant. But others, have a quality component. Whistleblowing is one. At least with whistleblowing, you can assess that quality after the event. For some things, you can’t. Many Firms ask people to track their activity; either for client billing or resource allocation purposes. For this exercise to have any value, people need to feel incentivised to take it seriously and provide accurate data. That isn’t always easy to do and the individuals tracking their activity will probably be aware of that. Whilst it is possible to do some form of Quality Assurance (QA) on the reported activity of individuals doing predictable tasks or working on specific clients or activities, it is much harder for those that don’t. Particularly in countries like Germany with strong data privacy laws. It’s the people with the least QAable data, whose input potentially offers the most interesting perspectives. After all, if we have no means of QAing it, then what they’re giving us is by default new insight. So for this and other activities where we need people to undertake tasks involving an element of quality, there is a need to ensure they’re positively engaged in it and not doing it under duress.

Finally, I am indebted to Dr Roger Miles for bringing something called The Table of Eleven (T11) to my attention in his highly readable book on Conduct Risk Management.

T11 is a framework created by the Dutch government that helps legislators consider potential reasons for non-compliance. By analysing the reasons why people might not want to do something, it can help legislators think about how they might craft better rules with a greater likelihood of people willingly complying with them.

I know some readers will see this thinking as overly complex and will prefer a simpler world where rules are simply written and obeyed.  Experience tells us that this doesn’t work and that the billions that are spent on writing, maintaining and policing rules, isn’t necessarily money well spent.  By making e it easier for people to do the right thing, we’ll get better outcomes and almost certainly save money. People don’t do things just because we tell them to or just because we employ them. And if we need quality of Compliance, then we risk not achieving it by thinking in those terms.

To rule or not to rule…

It will be interesting to see whether the FA actually has a rule that covers spying on other clubs.  If it does then Frank Lampard doesn’t know about it.  But then he doesn’t need to; he’s compliant because of a Social Norm.  His behaviour is intrinsically motivated. That’s far more powerful than the extrinsic motivation of doing it because of a rule that tells you to.

If there isn’t a rule, then the FA should think hard about whether one is really necessary.  Write a rule that suggests you think there isn’t a powerful Social Norm on the topic and you risk undermining the one that’s already there. Not only has the publicity surrounding this case demonstrated quite how powerful that Norm is, but it also theoretically means that no-one in English football can be unaware of the issue.  Which is probably enough to impose the Social Norm amongst those that weren’t already abiding by it.  At least in the short term.  Except perhaps foreign managers who come in the future.  Though I’d argue that there are probably better ways to get to them, than through the rulebook.

If there’s one thing I think the Bielsa case illustrates, it’s that if we want to get the right outcomes, then my mission to Bring Behavioural Science to Compliance is the right way forward.