Atoms and Choice, my Master's thesis project at Rhode Island School of Design proposes a different approach to mass customization that relies heavily on curation. This approach is embodied in eight rules and was tested through a modular lighting demonstration project.
Mass customization, the production of highly differentiated products at near mass production costs, has been presented as a way to increase product value and avoid commodification since the early 1990's. Most of these customization systems have relied on collaborative strategies wherein the customer gets to actively specify the various features and aesthetic characteristics of a product. Many of these efforts have failed or been significantly diluted as the value added did not justify their cost. That said, the challenges in mass customization are not technical, but rather in understanding that different consumers have different levels of product knowledge and want to make different kinds of choices in differing amounts.
To establish a series of best practices for the mass customization of durable goods with the goal of making the process accessible to consumers who have different amounts of product knowledge and choice making styles.
One of the biggest misconceptions in popular thought about customization is that more choices are always better. In truth, there is no direct connection between the amount of choice users have and product value. More choice is valuable to someone who has specific needs and can articulate which features meet those needs. However, for someone who does not know how to correspond features to their needs, being asked what features they want can function as a deterrent to participation. The desire for more or less choice corresponds with the consumer behavior characteristics of "maximizers" and "satisficers." Maximizers tend to seek out products that represent a global optimum for their needs and will routinely look beyond local options if their specifications cannot be met. Satisficers, in contrast, tend to choose the best local option that meets their less thoroughly defined requirements. For maximizers collaborative customization provides value by making global options local so they don’t have to settle for a lesser product. For satisficers whose needs are less well-defined, additional options do not avoid any sacrifice and simply add complexity to the purchasing process. Experience, knowledge and personality determine which of these behaviors consumers are likely to display and they are inconsistent from product to product. Accommodating both of these contradictory behaviors is key to making an accessible mass customization system and the goal of this project.
I had 12 participants try existing web-based product configurators (Nike ID, Timbuk2, Mini Cooper, BMW, Ect.) and give feedback on the customization process and the resulting product.
Users who know something about the product that their assigned configurator made found the process more enjoyable. Configurators with a large amount of choices could overwhelm users, especially if they were unfamilliar with the product. This caused some to find the process tiring or to view it as a game. The users who started to view the configurators as a game gave it a higher process rating, but gave the resulting products lower ratings. The various configurators had different amounts of parameter-based choices versus curation (the use of characteristic based filters). How much the participants knew about a given product changed their preference for more curation or more independent parameters.
To follow up on these initial surveys, I attempted to create my own paper configurator system which could help the participants select backpacks through Zappos by dictating how to set the product filters with a series of printable configurator rules. I wanted to see how different levels of curation affected how satisfied people were with the process of configuration and the resulting product. I had 13 participants fill out a survey on their impressions. I created three configuration scenarios: a highly curated process, a blended semi-curated semi-parametric version, and a full parametric version. Participants were then asked about the approachability and length of the process and how closely they rated the selected backpack to their needs and aesthetics.
My initial expectation was that, in trying to accommodate maximizers and satisficers, the blended configurator might have some of the advantages of both curated and parametric strategies. This was pointedly not true. The results of this experiment were quantitatively inconclusive, except for the blended version being ranked clearly lower than the other two versions in both process and resulting product. The curated version rated marginally higher on length, approachability and enjoyability. The parametric version rated highest in meeting product need and aesthetics, but lowest in approachability. Perhaps, more important than this was an unexpected qualitative pattern that I observed. When the different 1-5 ratings of the processes and products were tallied, the resulting ratings were very neutral, not because a high instance of 3 ratings, but because there were a fairly even number of 1 and 5 ratings. The people I talked to either loved or hated the experience, not many were neutral. Trying to meet heterogeneous needs with a one-size-fits-all configuration strategy did not appear to work well. However, there did seem to be a correlation between users' goals going in and if they liked the experience. In general, the more specific backpack the participant wanted the lower the curated version ranked.
I conducted eight interviews with organizations that engage in different kinds of customization. I talked to designers, chefs, entrepreneurs, and biologists to understand their processes and the lessons that they have learned making and marketing customization goods. Selected interviews can be read here.
My interviews provided an incredible amount of information. The insights from these conversations were used to generate the eight rules for best practices in mass customization below.
The eight rules represent a summary of my initial configurator research, literature review and of the interviews I conducted. My research fundamentally suggested that moving away from a completely parameter based system to one which is primarily curated (referred to as transparent customization) with optional parametric choice later in the process could have significant advantages. Transparent customization has been used to great effect in soft goods services such as remote stylists like Stitch Fix and informational services like Google and Pandora. However, it is almost non-existent for durable goods because it is significantly harder to realize. Transparent customization functions like a personal shopper, providing unique solutions with limited user input, as compared to parameter based collaborative strategies. In the transparent mass customization system I proposed, customers were asked what their intended use of a product was and how they would like it to feel and provided the parts to achieve those goals, instead of directly asking them what components they wanted. By making the process more accessible through curation, but allowing for further refinement though limited parametric choice both maximizers and satisficers can be accommodated.
Atom lights are an extremely flexible, modular lighting system that demonstrates the transparent customization strategy laid out by the eight rules. This project aims to illustrate that the primary challenges in mass customization are identifying the functional minimal divisible unit or “atom” of a product solution space and the corresponding granularity of questions needed to get users what they want without overwhelming them. I chose a modular solution rather than a truly flexible manufacturing technique because techniques are not necessary to provide meaningful choice as long as the modules used are interchangeable parts of the core product and not simply accessories. This project employed 3D printed components made up of standardized connectors with electrical contacts, a mechanical snap and differentiated casings to fit the various electronic components. Components were connected by silver traces on the edge of acrylic panels. This project is completely non-programmable and solid state, the module’s electronics are off-the-shelf-parts and modified littleBits.
I distributed several lighting surveys which asked participants what they wanted to use the light for, where they wanted the light to come from, how they wanted to control the light, how they wanted it to feel and to include an image. The information and needs presented in these requests were used to generate the module family and eventually create three of the requested lights.
I took a three-step process to develop the basic components of my lighting system starting from a typical lamp: dissection, standardization and abstraction. Dissection involved splitting a lamp into its constituent structural and functional components. Standardization involved defining a set of components that could be applied to multiple light requests. Standardization directly addresses the third rule in identifying user application affinity groups. Finally, abstraction is the process of combining some of these basic standard components into a smaller number of atom components (the minimal divisible unit)
with complementary core functions.
From this process, the four basic units of the system were created; distributors, something that directs light and electricity; controllers, something that controls the light; power supports, something that provides power and structural support; emitters, something that emits light. Additionally, there is a fifth category of modules: auxiliary, modules that serve to connect and hold other modules.
Choice navigation in a transparent customization system consists of a series of rules to match specific user requests to the right kit of modules. In the case of this experiment the inputs were the lighting requests: What will you use the light for? Where would you like the light to come from? When and how you would you like to turn the light on and off? How do you want the light to feel? Does this light convey a particular mood? Should it emit a certain tone or color of light? From the answers I received I tried to define the parameters of physical utility, how it was going to interface with its context and user experience. This profile information was then used to define constraints, which was in turn used to create a kit of modules for a specific user context. My strategy addressed performance specifications, user needs and user experience.
In order to test the efficacy of the eight rules and the choice navigation technique I developed, I selected three of the lighting requests and sent component kits to the requestees. The kits contained only the components needed to answer their request and a photo of the assembled light. I asked the participants to take photos and notes on the assembly process and their general impressions of use in the form of text messages or emails with photos.
Very easy to figure out how to set it up.
I like that I can just slide one piece to turn it off, so I don’t have to fumble around trying to find the piece in the dark.
Things to change:
It got finicky and wouldn’t consistently come on
or stay on. It protruded a bit further into the drawer than I would have liked so it was easy to hit it.
It wasn’t bright enough to light up the whole drawer, just the stuff right in front of it. The more that it’s a narrow beam that stretches across the top of my drawer, the better.
Would you like to add any other functions?
Don’t really think I’d add anything.
Would you add anything if you were to place it somewhere else?
Hmm, I feel like the needs of the drawer light are so specific that it wouldn’t necessarily work well in another setting. I guess it could be used as a reading light, in which case it would need a better hanging system that made very little impact on the wall or bed onto which it was stuck. With that though, I think the blue light would bother me while I slept, so I’d want to be able to have that turn off.
Is that helpful?
The atom light project aimed to test the idea of a durable modular product using transparent customization. The proof-of-concept project was quite intuitive and all test users were able to assemble the lights using only a picture for instructions. Users did seem to understand that exact placement of the modules was not necessary. A couple of the lights were reassembled in a new configuration which better fit their use case after their initial assembly. However, the fact that the modules have a top (with contacts) and bottom (with a snap) was unclear to unfamiliar users, especially in the more complex configurations. The addition of some indication on the top, such as continuing the lines from the silver traces on the modules, would have helped alleviate this confusion.
The slight finickiness of the electrical connections effectively broke the fourth rule and clearly caused some confusion because it could be misunderstood as an intentional behavior, obscuring the actual function of the modules. In order for the function of each of the module atoms to be properly interpreted they had to perform their intended action and nothing else. If the function of the module atoms was explained, users were better able to troubleshoot problems and get the light working even if they were initially confused.
Generally, users liked the lights and in some cases asked to keep them longer. While the users did not select their own parts, the act of putting them together clearly added to their sense of ownership and the value they saw in the product. However, in some cases users could clearly see the compromise of standard parts since modules could not always accommodate the use case as well as a purpose built object. That said, users did seem to have an understanding that their configuration was truly personalized to their context and the light’s function came from the unique set of modules they received. When users were asked if they would like to add anything to their lights, they were not able to suggest additional features, even when some possibilities were listed. However, they were very easily able to provide qualitative improvements. This indicated that improving the product would involve repeating the initial process of asking about and observing the light in its context of use, not asking about what features users would like to add. This was a clear advantage of the more curated strategy over a parametric one. If curation was used in an on-going fashion (rule five and eight), I believe it would be reasonable to expect better product improvement as compared to simple parametric surveys, considering the user had a hard time coming up with additional features.
The promise of mass customization is to meet the needs of niche markets on an individual basis, creating value by recognizing and accommodating differences. That said, its utility has been primarily confined to niches of expert users and has remained inaccessible to interested, but layman users. The careful use of curation and active feedback can bridge this gap and prevent less knowledgeable satisficer orientated users from being overwhelmed. The ability of laymen to get as much value from customization as expert users through the use of curation has the potential to truly put the seldom achieved mass into mass customization.