Systems engineering and service design

Availability as experienced by users, is the tip of the ice cube. The way some services fail, the incident is changing even as you’re trying to resolve it. The underlying problems are “open, complex, dynamic and networked” (Dorst, K). It is why we need systems engineering for service design.


ice cube

Traditional RCA methods don’t account for feedback loops. People and things involve may behave and respond in ways that frustrates the recovery effort. And that would be across many different sites and locations. We really need new ways to approach new kinds of failures.

Causal loop fishbone

In the era of AI, IoT, and cloud computing, human-centered design is “necessary but not sufficient” for designing services. We also need thing-centered design.

hcd and tcd

I cover some of this in my upcoming book.

Why – II

Why do services even exist?

Because people own things.
Some things they own, some others they don’t.
Some people own things other people won’t.
Thus the economy of sharing and caring,
From the melodramatic pairing, of
Supply and demand.

Need to have and need to be,
Makes things have affinity and attraction,
Towards things that promise action,
Buffer, boost, or timely traction,
At any given time and place,
Oh, logistical grace.

Performance, affordance, actualize,
Outcome and experience, materialize,
And each side expects to realize,
Value minus the compromise,
But if transactions get too frisky,
All bets go to ‘risky’.

When people and things, do engage,
Across device, channel, platform and stage,
There can’t be painful protraction,
Costly dialog and interaction,
Unreasonable extraction,
Silly subtraction.

Flex, forgiveness, forbearance,
Reliability, recovery, and resilience,
Among the many principles of design,
In situ, impromptu, or just in time,
Four promises, and a dotted line,
Even if the print is fine.

For profit’s sake or public good,
Global scale, or the neighborhood,
Run thousands of services, many kind,
And in every instance you will find,
(A Latin square, for the curious mind)
A standard model for design.

Public transit for packages


Every now and then I’m confronted with a choice. Should I order something on or drive 20 minutes to a local store and purchase the item there. I’ve been a card-carrying member of Amazon Prime since the year it was launched. Most items are delivered to me “free of charge” (i.e. no additional payment), within the timeframe I’d like to have them in may hands. There are instances when I need something right away, so there isn’t really a choice. Even same day delivery wouldn’t be enough.

That’s when I make a short suburban trip to pay for and gain rightful ownership of a ream of paper or a pack of pencils with dark chocolate eraser tips. (I’m joking. I prefer milk chocolate).  Within the hour, I’m putting my purchase to good use. However, unless I also have other appointments or chores in the proximity of the Staples or Office Depot store, I find the go-to-the-store option costly and inefficient. I feel like a taxi driver to paper and pencils. I could be doing other things. Why couldn’t they catch the bus, tram or train?

That’s what home delivery services offer. Each package is like passenger, and the barcoded shipping label is the ticket slapped on their face. Computers and algorithms solve for the most optimal route within a set of constraints, and figure out the last mile logistics in a way that may be as environmentally-friendly as public transportation. They are public transit for your packages.

Delivery trucks deserve HOV lanes.

I’m really bothered by the amount of energy we waste without regard to long-term environmental impact. Traditional parcel delivery services like USPS, FedEx, UPS and DHL have always been a public transit service. Of course, the last mile is presently a battlegrounds of sorts with all the majors competing to be the ones who ring your door bell or leave you a note.


Amazon Prime van parked in Silver Spring, Maryland

Let’s hope that in the long term the systems for delivering goods to the door will evolve to be more efficient than ever before.



Timeframe, timeline, and tempo

A contract can be executed within minutes, or over days, months and years. It depends on the type of service and often on the level of commitment involved. Some service contracts are worth hundreds of millions of dollars, stretching over many years. Many contracts are customized, negotiated and specified between two or parties, as is the case with professional services and outsourcing. Others may last for a few seconds, involve small amounts of money, and the contract itself might be implicit or automatic.

Purchasing a ticket for a train ride includes implicit agreement on what the service will provide and at what cost. The ticket is the contract. Airline tickets are attached with conditions of carriage in fine print that rarely anyone reads, but it is all there, including a force majeure clause. Citizens and taxpayers sign up for many pure public services simply by virtue of identifying with a community, municipality, province or country. As a taxpayer, you don’t have to sign up to make use of the president, Parliament, or police, and yet they are services that operate on very large budgets.

Dialog and interaction happen within the timeframe of the service contract, commencing with engagement and concluding with fulfillment. The various elements of dialog and interaction may occur according to a timeline, scripted or unscripted. They may happen in series and in parallel, or even within a single transaction, especially when they are automatic or implicit, and involve machines and software. The engagement can be spontaneous or scheduled ahead of time; it can occur at specific intervals, periodic cycles, or as and when the need emerges. The nature of the service defines the clock speed or tempo. Each cycle may be instant, last for minutes, hours, days, or stretch over longer periods of time.

Regardless of how fast or how often, for capacity to properly interact with demand, the dialog and interaction between users and agents are critical. Users and agents conduct dialog and interaction through touch points and interfaces. Given the extraordinary expanse of the universe of services, touch points and interfaces exist in all kinds of forms, sizes, and shapes. Touchpoints and interfaces ensure alignment and fit. Alignment avoids unnecessary seek and search costs on both sides. Fit ensures solid and secure connections with a very low rate of error.

“A line is a dot that went for a walk” – Paul Klee

Empathizing with agents

Indeed, many services are poorly designed because, in their eagerness or anxiety to reduce costs, the service provider pushes some of that effort and pain to the customer and user who are often unaware of this shift in burden and are often unwilling or unprepared to take it on. That’s why many services require the use of automated systems, machines, and software to reduce the interaction costs on their side, including artificial intelligence and software bots.

Bad design results in an unbalanced, unfair, and unsustainable distribution of cost and burden, often propped up by the absence of real competition or the presence of monopolistic conditions in a particular market space. Good design has foresight on the feedback loops, unintended consequences, and unnecessary conflict that ultimate cause disruption and failure.

Good design dictates that making it easy for the user, makes it easy for the agent. Making it easy for the agent, makes it easy for the user. This creates a virtuous cycle that drives down the cost of the interaction between user and agent, to stabilize at a lower level of fixed costs. The variable costs become substantially lower.

Self-service to elf-service

Traditionally, service providers push some of the agent’s burden to users to cut down time and cost, and to scale up capacity in a less expensive way, through the division of labor. Through self-service, agents temporarily delegate some of their responsibility and authority to users who can then help themselves. They are popular with users when they prove to be quicker and less burdensome in more ways than one. But self-service is an oxymoron because you have to do-it-yourself then how is it a service? And when users struggle with self-service they become less enchanted. With bots and assistants such as Alexa on Amazon Echo becoming smarter, trustworthy, and more intelligent, it might be that self-service will no longer be necessary. It will be elf-service with users and agents as elves who understand each other. The question is, does Alexa work for Amazon or its customers? When it comes to services, conflict of interest will always be a design challenge.

It is in the mutual interest of customers and service providers to reduce the overall cost of the transaction. There is value in having fair and simple contracts with sufficient clarity and depth. Regardless of the contract value, the cost of commitment and enforcement should be as low as possible, or as close to zero. Contracts should be cheap to execute.

If there is the need for any installation or set-up, then it should be fast and cheap not just for the enterprise and the agent, but also for customers and users who do not want to bear unexpected costs of making use of a service. Dialog and interaction should not be too painful for users or agents. Depending on when and where capacity engages demand, each side should have the access they need to infrastructure and facilities the other controls. It should be effortless to give and take custody and control of artifacts and resources.

If there are difficulties or problems in using the services, it should be easy for users to get help, assistance, or remedial action. There should not be any undue or unnecessary constraints on usage of the service, and there should be flexibility and tolerance for how demand expresses itself, across users, locations and time periods. Agents should have the responsibility and authority they need to for effective decisions and decisive actions. They should be in a position to properly advise and assist users, to be able to fix problems and errors, and to be able to handle exceptions or any unexpected situations.

People, facilities, and infrastructure should be trustworthy and safe. There should be adequate capacity, continuity, and care to make sure failures and disruptions do not jeopardize the safety and security of assets customers entrust for performance; and providers entrust for affordance. Personalities and fines may deter but they don’t prevent. One side should not place on the other the undue burden of enforcing contractual obligations.

Run Majid Run

The commonly observed phenomenon of someone hurrying across a train station, lugging their luggage, and boarding the train just as the doors close, is an example of demand engaging supply. The luggage and their out-of-breath physical selves are the artifacts and events, with the need to have (a seat) and the need to be (transported).

In the Netherlands, people use the OV-chipkaart to pay for public transport via buses, trams, and trains. It can be used to rent bikes and small cars. OV is a service that facilitates intermodal journeys with transfers, across several networks, with a chip card that serves both as the ticket and a form of identification. The pre-paid pre-loaded via card terminals at stops, stations, and retail outlets. The card terminals are examples of touchpoints with interfaces. They also happen to be agents because they issue travel credits after accepting payment. In most cases the person paying is also boarding the bus, tram, or train. The user is also the customer. They are making use of the service on behalf of themselves.

Thanks to the contactless card readers on posts, hurried traveler was able to check in for travel by quickly waving the chipkaart in front of the posts. The brightly colored posts are strategically placed across the station like the ones down a ski slope. If you miss one in a hurry, there is another just ahead. The card readers are agents with the responsibility and authority to check the card for balance and validate it for travel. They stand there, day and night, never getting tired of taking a break for a soda, sandwich or smoke. It takes them less than a second to process the request and give that distinctly audible beep that signals “OK” to the hurried user who must not slow down because the train is about to leave.

As the train crosses the bridge over the Amsterdam-Rijn canal and passes by the Douwe Egberts factory, the performance is in progress. The passenger and luggage can now catch their breath. The quiet of car and comfort of the seat are part of the affordance. Stilte means silence in Dutch. The conductor greets every passenger and checks their chipkaart, wishing the, a nice day and adding alstublieft. The friendly face and polite manner hide the responsibility and authority of handhaving or control. They’re simply making sure for every payoff there will be a payment because travelers in a hurry can sometimes forget.

In many cases, it is still “OK” even when passengers don’t pay. No, they aren’t forced to watch ads that run on auto-play. Their movements aren’t tracked, their privacy not hacked, nor are their habits analyzed for product display. It happens to be, they’re architects, accountants, and ambtenaar and such, doing their jobs and traveling for work. Users themselves happen to be “agents”. Their chipkaart are zakelijk or “for business” and their employers pay for them automatically when balances are low. In this case, the employers are the payers, while the passengers are the users. The ticket conductor is the agent and the train operator is the enterprise.

This is like that

In their book, The Knowledge Creating Company, Ikujiro Nonaka, and Hirotaka Takeuchi give a really good example of the power of analogy. It’s about how a product team at Canon achieved a breakthrough in the design of the Mini-Copier by figuring out how to make the copier drum inexpensive enough to be disposable. Doing that reduced the need for maintenance, and allowed them to launch a product suitable priced for the home and small office. The inspiration came to team leader Hiroshi Tanaka from the beer can he was holding, during a discussion on how to solve the problem. Tanaka could associate the designs of the beer can and copier drum because they were analogous in material and shape. Therefore, his team could learn from the process of manufacturing inexpensive aluminum cans.

In an essay in WIRED magazine, Chef David Chang explains the thinking that leads to the discovery of several of his most successful dishes, such as the Momofuku pork bun. He talks about how it is possible to deconstruct dishes to a set of base patterns that people respond to and subconsciously identify with, across cultures and cuisines. So, for example, a morsel of the signature dish Spicy Pork Sausages and Rice Cakes (in the mouth), evokes the sensations of mapo tofu and Bolognese, in the mind. In another dish called Ceci e Pepe, fermented chickpeas (in the mouth) evoke the sensation of Pecorino Romano in the mind. Because of flavor compounds that are common across ingredients, diners at Chef Chang’s restaurants are associating deep memories of dishes they’ve enjoyed since childhood, with the sensations of something entirely new on their plates.

“The key to analogical problem solving is to find known problems that have the same structure as the problem being solved.” Arthur Markman, Kristin Wood, Julie Linsey, Jeremy Murphy and Jeffrey Laux, in Supporting Innovation Through Analogical Reasoning, Rotman on Design, 2013

A battery stores energy. So do fat molecules, flywheels, and springs. They are solutions to problems with the same structure: Conservation or storage of energy through local conversion. AquaBells are dumbbells you can pack for travel. When you are ready for exercise, you fill them with water to add the necessary weight. Like air mattresses, they are compact, flexible and light and make use of an inexpensive local resource that’s readily available and discardable.

Known solutions are also useful in imagining patterns of possible failure, and therefore in understanding the costs and risks of design decisions that might lead to similar kinds of failure. Known solutions can be services, and innovation can be less risky and more inexpensive when it is possible to use existing services as the basis or components of a new service in another sector or problem domain. But first, we need a system or method for drawing and extending analogies that is suitable for services. As Markman and colleagues suggest, there needs to be a way to describe the essence of a service in a way that relates to other services. Performance and affordance provide that relational essence.

If you visit the National Air and Space Museum in Washington DC, one of the exhibits from the missions to the moon is the DSKY, the numeric display, and keyboard for the Apollo Guidance Computer (AGC). Two of the most worn out buttons are the ones used to issue commands. They have the labels VERB and NOUN. Within the context of a program, every command was a verb-noun combination. For example, the sequence NOUN06 VERB33 ENTERwould issue the command to display the time of the next engine burn. The verbs and nouns make sense within the context of programs or procedures during various phases of a mission. The essence of the 06-33 command is to get the computer to display-time.


Similarly, it is possible to express capacity and demand in a way describes the essence of a service, in terms of performance and affordance. Simple verb-noun combinations can express demand for action (AC), demand for access (XR), capacity to perform (CA), and capacity to provide (RX).