ep. 28: Towards a service design approach for building emerging technologies
7 min read
User-centered design (UCD) has served as a foundational product design process, helping teams remain focused on user needs, from understanding a user’s context, to specifying user requirements, to designing solutions and evaluating against those requirements. However, UCD also conventionally focuses on designing “one thing for one person”, rather than adopting a more holistic, ecosystem-level perspective.
In today’s newsletter, I discuss service design, an approach that considers how different stakeholders interact with and through solutions to achieve a desired outcome. I’ll discuss the value of a service design approach as we build increasingly complex systems (think multimodal AI agents and AI-powered smart glasses), and share three ways you can start cultivating a service design mindset.
What is service design?
Service design has traditionally been defined as planning and organizing a business’s people, props (physical or digital artifacts, e.g., products) and processes (e.g., workflows) to directly improve the employee’s experience, and indirectly, the customer’s experience. It’s how the user experience gets created, and how the internal parts of the organization align to deliver that experience. We can contrast this with the user experience (UX), which is what users encounter as they interact with a brand, such as an app, kiosk or website. The UX is typically is built by following the UCD process.
We can see the difference between service design and UX in the image above. The end user encounters the UX: the kiosk, its user interface, and a resulting mobile notification. UCD typically focuses on building and improving this UX. In contrast, service design is the ecosystem-level orchestration of technology, people, and processes that makes the UX possible: pinging the server, connecting the request with the right support agent, and documenting the outcome.
Why consider service design for building emerging technology?
To answer this question, let’s start with a brief history of HCI.
In the early days of computers (think ENIAC-era), operators developed interfaces for themselves or other experts like them. No UCD or UX was required. As computers moved from operation centers into workplaces and eventually, homes, UCD became critical. No longer was the operator designing for people like themselves - products had to work for a broad range of users. As consumer electronics like mobile phones and portable music players developed, people started to integrate these technologies into all aspects of daily life. UX design was born, helping us not only design for effective and efficient solutions, but ones that were engaging and entertaining.
We’ve entered a new era of technology, with AI-powered wearables like smart glasses, and powerful generative AI tools, like multimodal agents. Not only are these technologies integrated in every aspect of our life, but they are collecting vast amounts of data and are often interconnected with our other technology tools. In short, they are complex, interconnected systems. As such, we are rarely ever designing for a single product for one end user - instead, we’re likely designing a complex system (e.g., a platform or service), and require an ecosystem-level view to effectively design for it. A service design approach is well-suited for designing complex systems.
HCI leaders like Dr. Jodi Forlizzi have argued for years - even before the days of tools like ChatGPT - that the field of HCI needs to move beyond user-centered design (check out her elegant paper by the same name). Services and platforms are distinct from products, and therefore, require an approach beyond UCD and UX. Given their systemic nature, services and platforms must be designed with multiple stakeholders in mind, instead of one end user. For instance, a UCD framing for a ride-sharing app would focus on the passenger, whereas a service framing would also consider other important stakeholders, like ride-share drivers and taxi companies. Dr. Forlizzi also makes the case that pricing models and payment plans influence how people engage with technical systems. Serviced design considers economics as a material for design, whereas UCD or UX approaches rarely consider economics.
While a shift to service thinking can be widely beneficial to product design, I posit it’s especially relevant for designing emerging technologies. In the case of generative AI, the complexity of these systems is compounded by challenges of the technology (e.g., inference errors, hallucinations) that can lead to frustration and potentially, harm, across multiple stakeholder interactions. If we think of an example like a multimodal AI agent that help patients navigate their health plans, this complexity and risks come fully into focus.
We can examine a concrete example where service framing was not applied: Google Glass. The device was built for end users (i.e., the people wearing the glasses), without much consideration of other stakeholders, such as bystanders potentially being recorded by these smart glasses. This resulted in wearers being nicknamed “glassholes.” By applying a service framing to our emerging technology challenges, we can discover opportunities to make the tools we are building more useful and ease the adoption curve.
Why isn’t everyone doing it?
Despite the value service design can offer, we have been slow to adopt it, often defaulting to UCD and UX. Why? Researchers have identified several challenges. One is confusion around definitions of service design within HCI. Service design is often narrowly perceived as an approach to optimize an organizations’ operations to better support users, rather than a broadly applicable, ecosystem-level approach. The multidisciplinary nature of service design (e.g., touching design, marketing, management and beyond) also contributes to a broad range of definitions, whereas UCD’s aims are fairly targeted (‘focus on the end user’), making it comparatively straightforward to for a variety of stakeholders to adopt.
If we take the view that service design is an ecosystem-level approach that can help us more holistically design complex, emerging technology systems, we can attempt to move past this definitional confusion and reap its benefits.
Applying service framing to emerging technology
To close this episode, I offer three ways we can start to apply service framing to emerging technology design:
Identify stakeholders beyond the end user: Transitioning from UCD to an ecosystem-level service design approach involves broadening the focus of product design from individual user experiences to the entire ecosystem of interactions. This includes consideration of multiple stakeholders beyond the end user, such as customers, employees, and third parties. Some concrete ways to do this include:
Consider all people affected by a given problem when writing a problem statement. For example, if you’re designing a customer service AI agent, you would not only consider the end user, but other stakeholders such as customer service representatives, managers, and technical support staff.
Use a service blueprint to visualize stakeholder relationships, as well as props and processes.
Remember that we’re in an age where a stakeholder is not necessarily a human. We can envision a near-future where we’re designing for interactions between a person and two or more AI “stakeholders” (see this article from March 2024 on two AIs talking to each other). If this applies to your product space, make sure you include these digital “stakeholders” in your mapping.
Consider economics as a design material: Service design recognizes that economic models are integral to the design process. Identify how pricing models and payment plans might influence how people engage with the system you’re building.
Take inventory of existing technologies within a given ecosystem: Consider how a device like smart glasses interacts with stakeholders’ technology ecosystems, which can reveal opportunities for interoperability (i.e., ways to make your technology more useful and easier to adopt). This is also a good moment to identify what unique value (aka “superpowers”) your technology offers relative to these technologies. For a deep dive on the pitfalls of not considering people’s existing technology ecosystems, see critical feedback on the Humane AI Pin from episode 25.
Human-Computer Interaction News
Microsoft introduces SIGMA, an open-source mixed-reality system for research on physical task assistance: The SIGMA (Situated Interactive Guidance, Monitoring, and Assistance) system, currently running on HoloLens 2, provides a basis for researchers to explore, understand, and develop the capabilities required to enable in-stream task assistance in the physical world. Examples include building a bicycle or fixing a broken water heater. SIGMA is further evidence of the deepening intertwinement of mixed reality and AI, leveraging large language and vision models to guide a user through procedural tasks.
Snapchat launches new AR and ML tools for brands and advertisers: Snapchat unveiled new AR and ML tools designed to help brands and advertisers reach users on their social network with interactive experiences. The tools aim to reduce the time it takes to create AR try-on assets, which can help brands more quickly turn their 2D product catalogs into try-on experiences. Brands can also now provide a text or image prompt to generate a unique ML model that can add realistic face effects to a Lens.
Reconsidering the role of AI: valuing process over output: Information architect, author, and educator Jorge Arango reminds us that we’re far from replacing humans for many tasks being delegated to large language models (LLMs). As such, we should not expect LLM-based tools like chatbots to be “intelligent”, capable of directly generating an end product or experience. Instead we should use these AI-based tools on tasks for which they are well-suited for - the process of getting to the output (e.g., analyzing, synthesizing, manipulating data).
Interested in applying a service design approach? Sendfull can help. Reach out at hello@sendfull.com
That’s a wrap 🌯 . More human-computer interaction news from Sendfull next week.