Friday 2 March 2012

Get them hooked: how can you turn a one-off experiment into long term usage?

Profits up despite the recession in Europe
Looking for a good investment opportunity in Europe? Why not buy some shares in British American Tobacco (BAT). Amidst the turmoil of the EU debt crisis, BAT is performing better than ever. Its profits were up 12.4% last year. Three cheers for good old fashioned British enterprise!

What we can learn from cigarette companies
The reason BAT is kicking arse is because its business model is geared around long term use (aka addiction). BAT doesn't make much from the first pack you buy. But over a lifetime, they make a killing! Even though you might question the ethics of a company like BAT, it's obvious they're doing something right to still be around and still making profits despite such a harsh regulatory environment. The take-away message is it's not adoption that matters, it's long term usage of your service


How to get your users hooked on your service
Let's use cigarettes as a metaphor to come up with some general principles around creating long term users of your service. The reason cigarettes are addictive is because nicotine causes your brain to produce more dopamine, which in turn activates the pleasure centres in the brain. Unfortunately unless your service involves direct brain stimulation, you're not going to be able to immediately activate people's pleasures centres, but you can do it indirectly. All you have to do is make people feel good.

Making people feel good: the Expectation Confirmation Model
Way back in the 80s, a dude called Richard Oliver came up with a model that explains why people get addicted to cigarettes and why people become long term users of a service. The Expectation Confirmation Model states that if people's initial expectations are met, they'll keep using your service. Since then it's been validated empirically many a time (e.g. here), racking up over 4500 citations.

Oliver's initial theory is a nice overview. But it opens up three big questions:
1. What influences people's initial expectations?
2. How do people decide if their expectations have been met?
3. How do people's expectations change over time?

What influences people's initial expectations?
Ventakesh et al (2011) sought to answer this question for IT systems building on a wealth of previous research. Their Unified Theory of Acceptance and Usage of Technology (UTAUT) model proposes five key factors:


FactorQuestions people ask themselves
Performance expectancy
Is this service actually worth using? How is it going to make my life better?
Effort expectancy
How much effort will it take to learn how to use this system? How much time and money will I have to spend to get the benefit? Is it worth my while?
Social influence
Are my friends using this service? Will this help me impress women if I start using this service?  Will my boss be pleased if I use this service?
Facilitating condition
How much support is there? Is there a hotline I can call if I get stuck? Do I know anyone who can help me use this service?
TrustAre these guys going to sell my email address to spammers? Are they doing this for the right reasons or do they just want to make a quick buck? Does the company actually care about me or am I just another number

Based on empirical evidence, these five factors are the necessary ingredients to make your service go viral. Im et al (2011) looked at the spread of MP3 players and internet banking in Korea and the US and found that the UTAUT model explained close to 80% of the variance in the data.

How do people decide if their expectations have been met?
Notice the operational term "their expectations". It doesn't matter if you can prove that your service has delivered the benefit that you believe it delivers. What matters for satisfaction is whether people perceive that their needs have been met. Cadotte et al (1987) give a good explanation of satisfaction. The basic message is can the hype so you don't disappoint people. People get really pissed off if you make a big promise and don't deliver on it. Conversely, they get really excited and evangelical when you exceed their expectations.

An idea I have for this is to explicitly lay out what you won't promise and won't deliver. I haven't seen a lot of products or services do this. Probably because it requires humility and many entrepreneurs are sorely lacking in that area:P

How do people's expectations change over time?
Ease of use doesn't really matter much after a while
Bhattacherjee and Barfar (2011) point out some issues with UTAUT. One of their main objections is that UTAUT doesn't include a time component and a sliding scale on each of the factors. They acknowledge that while all of the factors in the UTAUT might be equally important at the beginning, over time, it's likely that some factors will become less important. For example, Karahanna et al (1999) show that once people have used the system for a while, ease of use really doesn't matter to them anymore. It could be the least user friendly interface ever, yet because they're familiar enough with it, they'll keep using it. This explains why a lot of big companies that have been around for a while still have MS-DOS systems:S

Role of habit
Another factor that UTAUT doesn't consider is the role of habit. Limayem and Cheung (2008) show that it doesn't matter how good a service is and how resolutely people intend to start using the new service if their habits prevent them from using it. Let's take CoCoRide for example: even if we can check off all of the five factors in the UTAUT, people aren't going to share rides if they're locked in the habit of driving to work by themselves.

An expanded version of Expectation Confirmation Theory
Because of those two fundamental flaws with the UTAUT in explaining long term use of a service, Bhattacherjee and Barfar (2011) suggest completely separating out initial adoption (which would be explained by the UTAUT) and long term use (which is better explained by the Expectation Confirmation Theory - ECT). They then propose an expansion to the ECT: habit. In the end, their model looks like this:

Take away message
You've got to underpromise and overdeliver and then you have to make it so easy to keep using your service that people will take it on as a habit.

Where to from here?
The next thing I want to look at is habit formation. 

No comments: