Lazy User Model

THE BACKGROUND

I have been working for a number of years on issues having to do with mobile business and business models for mobile and e-business.What I have encountered everywhere is great uncertainty in the estimation of success of the new ventures – it actually feels like one has tolaunch to test the demand – this means “invest first and then pray” type of business model. Am I talking about mobile services? Internet services? Other hi-tech services? Yes all of these and sure enough all of them belong to the group of invest & pray businesses. Am I talking about start-up companies? Large established companies? or what – what seems to be the most interesting issue here is that there does not seem to be any marked difference whatsoever in the amount of ex-ante demand knowledge of this type of services between companies of any size. This means that small companies loose millions and large companies loose tens or even hundreds of millions on their new servicesthat just don’t fly (there are multiple examples on this). I find this very interesting, as we still seem to know very much about technology and even able to launch market studies and other research into how much demand different new products and services seem to ex-ante have. Marketing is also a subject that has been widely researched.

So if we don’t know ex-ante if a service / product will fly – what can we turn to for help?

Some seem to think that we can use existing models for technology adoption and acceptance for help. My comment on that is that there is a huge overall disappointment with existing models & methods that try to explain technology acceptance and adoption especially when one tries to use them to predict the ex-ante success of a new services & technologies. In other words, these models are worthless in predicting success => and using them makes your dumber that you already are = you think you know, which is worse that knowing that you don’t know.

A common answer of the advocates of the models is “but the models are not meant for ex-ante prediction of success, but acceptance and adoption”. Success IS that someone wants to adopt & accept new technology and products & services, so if the methods are trying to explain and predict acceptance & adoption then, unfortunately, they are failing at that.

BIRTH OF THE MODEL

I had been working together with Franck Tétard (and a small group of others) on a project that looked into using medical databases installed in mobile phones to support, the work of medical doctors during their military service. We tried to figure out if the system was actually helpful to the military doctors or if it was just “nice to have” thing, and if the system was deemed good then what made it good and what could be better (HCI focused efforts). After this joint project with Frank I started to think about why the existing user adoption/acceptance models fail in predicting success of new technology consumer or B2C products and services and started to sketch what in my mind were the things that matter.

What I always ended up with were two things: The user and the “cost” of using. If the user does not need the service she will not use it. If the service costs too much the user will not use it even if there is a need. => so there is need and there is cost. What else is there? There is “who the user is”, i.e. the user characteristics. Then I started thinking about the existing models, I found that many of them have as factors the same things that I was looking at, but mostly the models were not focused on the user, but on the technology. AND what I found very interesting was that they were almost always focused only on ONE technology = as if there was only one alternative in the world. I found that that is a VERY big obstacle for any kind of prediction ability that is, including only one technology in the analysis of adoption – it is CLEAR by common logic that a person looking at a new technology will compare it to the existing (old) technology and take the one that is better => so I started to figure out what affects the selection of a product/service if the alternatives given all fulfill the need of the user?

The first thing that came to my mind was the cost, but not only the monetary cost, but the overall cost, or better yet the “lowest level of effort” that we can understand as a combination of monetary cost + use of time + physical / mental effort needed to use the solution to the need (problem). My hypothesis is that it is that combination that is, by far, the most important driver in determining how we select products / services from the “list” of choices that we have. There are indications to this effect from psychology and from information seeking – I ran into for example articles from Zipf when going through the literature. Also it is not a secret that even water tries to fined the path of least resistance when flowing, so this least effort issue may not be “lazy” behavior at all but natural and normal ;)

Lazy is just so catchy….

To simplify these thoughts I drew a model on a piece of paper, worked on it a bit and the first version of the “lazy user model of solution selection” was born. I very fast wrote the first working paper on the model and showed it to Franck Tétard who I have been working with on the model ever since.

THE MODEL

The model “starts” from the observation that there is a user need, i.e. we expect (for modeling purposes) that there is a “clearly definiable, fully satisfiable want” that the user want’s satisfied (we can also say that the user has a problem and she wants the problem solved). So there is a place for a solution / product / service. The user need defines the set of possible solutions (products, service etc.) that fulfill the user need. For the purposes of simplification we like needs that are 100% satisfiable and services that 100% satisfy the needs. Only the solutions that solve the problem are relevant here, so this logically means that the need defines the possible satisfying solutions. Here we can see that we really are talking about a SET of solutions (many different products / services) that all can fulfill the user need, hence we are not limiting ourselves to looking at one solution separately.

All of the solutions in the set that fulfill the need have their own characteristics; some are good and suitable for the user, others unsuitable and unacceptable – for example if the user is in a train and wants to know what the result from a tennis match is right now, she may only use the types of solutions to the problem that are available to her = the user state determines the set of available / suitable solutions for the user and thus limits the (available) set of possible solutions to fulfill the user need. The user state is a very wide concept, it is the user characteristics at the time of the need.


Figure: The lazy user model of solution selection

It is age, wealth, location… whatever, that determines the state of the user in relation with the solutions in the set of the possible solutions to fulfill the user need. The point here is that after the user need has defined the set of possible solutions that fulfill the user need, and the user state has limited the set to the available plausible solutions that fulfill the user need, the user will select a solution from the set to fulfill the need. Obviously if the set is empty the user does not have a way to fulfill the need. The model assumes that the user will make the selection from the limited set based on the lowest level of effort (cost), understood as already discussed above as the combination of monetary cost + time needed + physical / mental effort needed.

CASE EXAMPLE OF THE MODEL 1: RESULT OF THE GAME

A sports interested mobile telephone owner user has made a bet on the result of the game and knows that the game has ended. She wants to know, as soon as possible, if she has won. The user need is, therefore, information on the end result of the game, as soon as possible. The overall possible ways of getting the information are numerous, however, if we consider two user states a) the user is at home watching TV on the sofa and b) the user is at an airport abroad waiting, the set of possible ways to obtain the result of the game are different. In user state a) we assume that the user has eight different possible solutions (radio, TV-news, teletext, call friend and ask, newspaper next morning, internet, mobile Internet, and SMS result service). In user state b) the set of solutions are limited to the possibilities offered by the mobile phone (call friend and ask, mobile internet, SMS result service) and Internet at the airport at an elevated cost.

Figure 2: result of the game

In user state a) the user choices that offer the least effort are teletext (the user is sitting on a sofa with a remote control nearby), an SMS result-service and TV-news. Depending on chance the TV-news may be showing the result instantaneously, which would make it the least effort solution, however, if this is not the case and the user is an experienced user of teletext, then teletext would be the least effort solution. However, if the user is not experienced with teletext and there are no TV-News that would show the result, then an SMS service would be the least effort solution. It seems that there may be a set of solutions that offer very similar low levels of effort, which makes the selection of the solution difficult to the user. In such cases the user familiarity with the solution may be the deciding factor, e.g., if the user is not accustomed to using teletext and is accustomed to using the SMS service, then the SMS service may be the least effort solution even if the user is sitting next to the television. In any case, it is most likely that the user will select one of the three solutions identified here as the least effort solutions.

CASE EXAMPLE OF THE MODEL 2: TRAM TICKET IN HELSINKI

In our second example we discuss the Helsinki City Transport Company’s mTicket that enables mobile phone users to pay for their tram, metro and bus tickets with an SMS. We expect that the user is not a holder of a tram pass, that she has a mobile telephone capable of sending and receiving SMS, and that she is waiting at the tram stop. The user need is to get the ticket for the tram. We are considering two different user states a) the user is in a hurry and does not have cash and b) the user has all the time in the world and is carrying cash.

The set of solutions for buying the ticket are to buy one from the tram (with cash), to buy one by using the SMS service (information on every tram stop), to buy one from a kiosk (non – evenly distributed throughout the city), or to buy one from a vending machine (available at metro stops).In user state a) the user choices to fulfilll the need are reduced to buying the mTicket, as the user has no cash (trams accept only cash) and as she has no time to buy with a credit card from a kiosk, or a vending machine, both located at a distance.

 Figure 3: tram ticket in Helsinki

In user state b) the user choices are all the four possible solutions. According to the lazy user theory the user selects the solution with the lowest level of effort. In user state b) the least effort is to buy the ticket from the tram with cash, or to buy the mTicket. Buying the ticket from the tram means that the user must walk to the front of the tram and buy the ticket from the driver; buying the mTicket means the user must take her mobile phone and send an SMS to the correct number. Even if the user would have unlimited time (and can afford to miss the next tram) it is unlikely that buying the ticket from a kiosk, or from a vending machine, would under any circumstances be the least effort solution. If the tram does not come instantly and the user has spare time to buy the mTicket (and at the arrival of the tram just walk in the tram), the least effort solution will most likely be to use the mTicket.

User attachment to mTicket can be enhanced by advertising the service, e.g., at the tram stops – potential service users that have time to wait for the tram are likely to adopt due to it being the least effort solution. Further, there are a number of other possibilities to enhance the attachment of users to the service, e.g., the pricing policy of mTicket can be made such that it gives an incentive to use, which reduces the workload of the drivers and contributes the trams ability to keep the tight timetables (service quality). Additionally, if the mTicket is available as a shortcut, e.g., in the menu of the mobile phone as a “one-button-solution” the effort will be even further reduced and possibly make the mTicket clearly the least effort solution. The above mentioned issues are also usable indicators for service design more generally.

On a related note, in 2006, in Stockholm, Sweden, referring to safety concerns bus drivers refused to accept cash payments after a series of ticket payment robberies. This resulted in losses for the City of Stockholm – an mTicket type solution would possibly have solved the problem.

Implications to understanding switching costs and their importance to starting learning

Franck Tétard has also been involved in working on the issues connected to learning within the context of the lazy model, here an important issue are the switching costs; the paper presented at HICSS (see below) discusses this issue. More on the issue will be posted here later.

We published a paper in the Lecture Notes on Business Information Processing vol. 86 that discusses the learning issues and switching costs. The paper was presented at the Scandinavian Conference on Information Systems in Turku 2011.

Research on the method validity

We have started to test the validity of the model by doing a survey on the model and its parts, we have also gone forward with a simple applicationto use the type of analysis the model can offer to analyse the success factors / ex-ante success of new mobile services (more on that will also be added here soon).

From the first small scale survey it is already evident that respondents attach a lot of importance to the combination: [ time consumption/effort/€ cost] when selecting services. This is an important indication about the credibility of the model – some results can be found below in Anna Fredriksson’s master’s thesis (downloadable below) and the GMR Conference paper.