Chapter 8. What Makes a Product Great
Now that we’ve covered goals and the purpose of AI for both people and business, let’s turn our attention to how can we create an AI vision around building great products that are able to achieve these goals.
This chapter is meant to give you an overview of the concept of importance versus satisfaction, the four components that I think make products great, and also Lean and Agile development concepts that we can use when building AI solutions.
Importance versus Satisfaction
Let’s begin by discussing the concept of importance versus satisfaction. This provides helpful context to keep in mind as we cover the four components that make products great in the next section.
We introduced the “Jobs to Be Done” Framework previously, which is part of a larger strategy and process called outcome-driven innovation (ODI). As discussed people hire businesses, products, and services to get a “job” done. The reasons behind people hiring a product to get a job done are often not obvious, or immediately understood by the people doing the “hiring,” nor are they easily explainable.
Sometimes, the reasons are because one product makes them “feel” a certain way, which is a bit intangible but can be certainly real. Examples include hiring a bowl of ice cream to make you feel better, a toothbrush to get your teeth cleaning job done, an accountant’s services to get your annual tax filing job done, and finally, an email client such as Gmail to get your emailing jobs done.
Well-known author and Harvard professor Clayton M. Christensen, along with his coauthors, points out that there can be powerful social and emotional factors beyond just the functional that influence people’s perception of whether a solution does a job well or not, and the experiential aspect can be very important as well. Ultimately, people tend to continue working with companies, products, and services that they’ve “hired” if they do the job very well, otherwise, they might get “fired” if they do not.
Further, all jobs consist of steps, and the Jobs to Be Done framework is an organized way to brainstorm and unlock innovative ideas around how to make the steps easier, faster, or unnecessary. This represents a very big change in the way people normally create products. It shifts the focus from business metrics to customer metrics, and from making the product better to making the jobs and outcomes better for people.
In other words, the focus is on the jobs to be done and not the product itself. It’s also on the voice of the customer instead of assuming to know what the customer needs or wants. Product ideas should follow naturally, and this approach enables both innovation and greater business success. A happy customer means a more successful business. In a few moments, we examine in detail additional concepts and approaches to human-centered product design and development.
Jobs to Be Done also heavily emphasizes the concept of opportunity and outcomes as a function of importance to customers and their satisfaction with existing alternatives. Figure 8-1 shows this relationship.
Figure 8-1 shows three different market segments, particularly those where the outcomes are underserved or overserved. This provides a graphical way to segment customers based on understanding a job that they’re trying to get done and determining the degree of opportunity that the customer’s desired outcome represents. The biggest opportunity is where the customer’s needs are largely unmet, which in this context means that the importance of getting the job done and achieving the desired outcome is high, whereas the customer’s satisfaction with alternative ways of getting the job done is low (lower-right quadrant).
Figure 8-1. Importance versus satisfaction (from Anthony Ulwick, “A Three-Segment Solution,” The Marketing Journal, http://bit.ly/2HS4IvT, accessed February 23, 2019)
It’s worth noting that we can apply the importance versus satisfaction relationship to individual features as well as entire products. This is one way to help prioritize and order which features to build in benefit-driven product development. With our understanding of why people “hire” businesses, products, and services in order to get a “job” done, let’s now discuss what makes products great, which ultimately prevents them from being “fired.”
The Four Ingredients of a Great Product
What makes a product great? What causes people to use one product over another? What causes people to use a product daily as opposed to once a month? What causes people to engage and interact with a product more over time, and for longer durations with each use?
It boils down to four primary ingredients:
-
Products that just work
-
Ability to meet human needs, wants, and likes
-
Design and usability
-
Delight and stickiness
Let’s discuss each of these ingredients in turn.
Products That Just Work
Nothing makes me want to stop using a product more than it not working or not working well. Quality matters, and you should spend time and effort ensuring it, and yet this isn’t always the case. Products that don’t work might have bugs, crash regularly, or need workarounds. Conversely, products that “just work” do exactly what they are meant to do, in the exact way that they’re meant to do it, and without error. Or at least that’s the case for products that are not error-based. Let me explain further.
Not all but most AI and machine learning models are error based in some way, which means that the models are trained using a training dataset until the chosen performance metric (e.g., accuracy) is within acceptable range when the model is tested against a test dataset. AI and machine learning solutions aren’t perfect and aren’t able to be correct 100% of the time. Acceptable range is another way of saying that the performance metric (and subsequent error-level reduction) is “good enough.”
Unfortunately, good enough is not a very quantitative measure. Also, sometimes reaching a target performance level, or even an acceptable performance level, can be very difficult depending on many factors, and certain errors can have life or death consequences, all topics that we discuss later in this book.
That said, often the benefits of “good enough” far outweigh any potential downsides, which makes it a reasonable goal. Also, some applications can get pretty close to perfect. Business folks, domain experts, and AI practitioners must work collaboratively to determine what errors are acceptable (good enough) for a specific application. When all of the criteria in this section are met, including “good enough,” an AI application should “just work,” and that’s an important goal.
Ability to Meet Human Needs, Wants, and Likes
A product’s ability to actually meet human needs, wants, or likes is critical and should be somewhat obvious, but it’s definitely worth discussing. There is a lot of nuance here that will be considered throughout this section, including the difference between the three.
To begin, the more that human needs, wants, or likes are prioritized and understood when developing technology solutions, including those built with AI, the more successful the product and better the human experience will be for users. Humans don’t necessarily care what technologies or underlying nuts and bolts are being used as long as they solve a problem or meet certain needs, wants, or likes in a delightful way. The best applications of AI are those for which the use of AI is abstracted away and the user knows only that a product meets one or more of these things and is better than the alternatives.
To better understand concepts around human needs, let’s briefly go over Maslow’s hierarchy of needs.
Maslow’s Hierarchy of Needs
Many of us are familiar with, or at least have heard of, Maslow’s hierarchy of needs. These needs are significant drivers of human motivation, and are largely geared toward physical and mental survival and health, as well as self-actualization. The point of bringing it up here is to provide a brief recap and discuss these needs in the context of technology and AI specifically.
Figure 8-2 shows Maslow’s hierarchy of needs as a five-level hierarchy, with the need categories in order of importance, from bottom to top: physiological, safety, belongingness and love, esteem, and self-actualization.
Figure 8-2. Maslow’s hierarchy of needs
The bottom four layers of the hierarchy are referred to as deficiency needs. Humans are most motivated to fill needs lower in the deficiency part of the hierarchy because they are required for fundamental physical survival, and humans become more motivated to fill these needs the more that they are deprived. As those needs are met, motivation shifts to fill needs that are more social and mental. Interestingly, the framework suggests that motivation again increases, potentially significantly, after deficiency needs are met in order to meet self-actualization needs (top of the hierarchy). Unlike deficiency needs, which are based on lacking fulfillment in certain areas, self-actualization needs are driven by desire for personal growth and fulfillment. Examples include professional growth, becoming an SME, climbing Mount Everest, or learning to play a musical instrument.
Originally Maslow stated that the needs must be met in order, but later clarified that motivation decreases for a specific need once it’s “more or less” met, and focus will shift to the next unmet set of needs (salient needs). Although people naturally try to fulfill needs from the bottom to the top of the hierarchy, life events and other circumstances (e.g., disease diagnosis, divorce, getting laid off) often make that an iterative and ever-changing dynamic.
Maslow also thought that due to fundamental differences between people and what motivates them, certain needs higher up the hierarchy can become more important than even basic physiological or other lower needs (anorexia, as an example). Additionally, people can be motivated by multiple needs simultaneously. Social media is a great example. People often use social media because it provides a sense of love and belonging while also helping with their personal self-esteem.
It’s also worth noting that there are certain criticisms of Maslow’s hierarchy. Some include the lack of spiritual, altruistic (putting others needs ahead of ones own), and societal (self-centered versus society-centered cultures) needs in his framework, all of which can obviously be very powerful.
The difference between needs, wants, and likes
An important distinction to make is between human needs, wants, and likes. They are all similar and related, but not the same.
Needs such as those included in Maslow’s hierarchy represent things humans need for survival, strong mental health, and self growth in the case of self-actualization. Wants on the other hand represent things that humans want to have, but don’t necessarily need to survive mentally and physically.
The farther from a need a certain want is, the more people’s focus is placed on concepts like utility, usability, and delight (discussed further shortly). Also, humans might want (desire) something because they think it will meet one of their noncritical needs, or they think they’ll like it after they have it, but that’s not always the case. Think of all the toys that kids want at first, but that never see the light of day after the first use or so.
Likes are the end result—the degree to which someone likes something (e.g., product, food, trip destination) whether they wanted it or not. We’re often presented with opportunities to try things we didn’t know about or want (e.g., grocery store samples) but then realize how much we like or dislike it.
Human needs, wants, and likes are powerful forces that drive human decisions, technological innovation, and the difference between great products and not great products. With all of this in mind, let’s turn our discussion to actually meeting needs, wants, and likes and the factors that we must consider in order to do so.
Human-centered over business-centered products and features
Actually meeting human needs and wants with technology products is often much easier said than done. There are many reasons why, with the most important arguably being the difference between designing and building human-centered versus business-centered products and features. Products are all too often made in a business-centered as opposed to a human-centered way, which usually results in products that don’t properly address the user’s needs, wants, and likes.
Businesses owners and workers aren’t the customer or users of their products despite how often many create product ideas and make product decisions as if they were. This results in business-centered as opposed to human-centered products and features, and quite simply, the less human-centered any user-facing product or feature is, the less successful it will be.
A related and contributing factor is the so-called highest-paid person’s opinion problem (aka the HiPPO problem, which usually refers to company owners and senior executives). The HiPPO problem occurs when product decisions become disproportionately weighted based on a HiPPO over actual customers or users. Another manifestation of this problem is when the HiPPO overrides those with greater expertise in making great products (e.g., UX designers, product managers).
Lastly, business stakeholders typically have differing goals and incentives related to their business function, and therefore tend to compete with one another for product features that can benefit them directly. If not managed carefully or left unchecked, this can become a runaway scenario in which a product winds up including everything but the kitchen sink. This will ultimately result in a bad UX and diminish the ability of a product to meet human needs, wants, and likes as well because the desired functionality can become buried in a sea of nonessential features and UI elements.
Products should always be benefit-driven, not feature-driven, and, most important, have the user’s needs, wants, and likes as the driver of everything else. There are many ways to ensure this and avoid the issues discussed; for example through UX research and design, human-centered design, user-centered design, and design thinking in particular.
An example is empathetic research and design as baked into the empathy stage of the Design Thinking process (discussed in Chapter 9). Empathetic design centers on observing consumers and understanding their needs, as opposed to relying solely on market research or nonconsumer perspectives.
Also, Google started an effort it calls human-centered machine learning (HCML) to handle some of these considerations, along with a focus on how to leverage AI and machine learning in inclusive ways. The effort emphasizes the importance of the UX axiom that says “you are not the user,” and provides a lens that allows them to “look across products to see how ML can stay grounded in human needs while solving them in unique ways only possible through ML.”
As a final note to this section, there are many ways to measure whether a product meets human needs, wants, and likes. They include product performance analysis (e.g., sales, customer acquisition), customer engagement and retention analysis, and analyzing user feedback for example.
Design and Usability
Design is an often-underestimated area when creating a product vision and strategy, and yet it is critical to product success. Design is a very broad term, though, so a little bit of specificity is in order. This book is about harnessing emerging technologies such as AI to innovate; therefore, we focus on design as it relates to innovation and technology.
Dan Olsen, a product management consultant and author of The Lean Product Playbook,1 created two frameworks that we discuss in this chapter. The first is called The UX Design Iceberg and the second is called The Product-Market Fit Pyramid (covered later in this chapter). Let’s cover The UX Design Iceberg first, shown in Figure 8-3.
Figure 8-3. The UX Design Iceberg
The part of the iceberg that is above the water surface is what users see and interact with; or, in other words, is the visual and interactive part of the UX. The layers of the iceberg below the water is what the UX is based on—it’s the design foundation. Let’s briefly discuss each of these foundational pieces, starting from the bottom.
Conceptual design represents the earliest stage of the design process. It is where a high-level understanding of user needs is generated, and subsequently turned into initial concepts of what the solution or product will look and feel like. This is the beginning from the design perspective of mapping the problem space to the solution space.
Information architecture is the part of the design process during which the designer determines how the information is logically organized in the product and what the path or flow of information retrieval is like for the user. The logical organization includes the structure and layout of information across the entire product. Product navigation and subnavigation order and options, along with content layout in a feed, for example, are all aspects of information architecture.
Interaction design is the design of user interactions with the product. This can take the form of navigation and flow throughout the product as well as interact with individual elements of the UI, such as inputing information, speaking (becoming more dominant), selection, clicking, swiping, and pinching. Interaction design also includes interaction feedback design in the form of messages, notifications, and errors (e.g., validation), for example.
The final user-facing design layer is visual design. This is the aesthetic aspect of design—how the product’s UI looks. This includes colors, contrast, fonts, typography, logotypes, graphical elements, positioning, and sizing. This is not a complete list, but hopefully this helps establish many of the key areas of visual design.
Designing and implementing everything as described by the UX design iceberg does not mean that the design or UX will be good or effective. This is true even if you have met the first two criteria of great products: the product “just works” and is able to meet human needs or wants. People need to understand the information that the product presents, and also how to interact with it. In most cases, people experience a technology product for the first time without having read a user manual or receiving training on it (think of most of the mobile apps you’ve tried).
This is where the very important concept of usability comes in. This is an entire field consisting of important research, concepts, and testing methodologies, an in-depth discussion of which is out of scope here. That said, let’s discuss the most important concept and key takeaway for usability.
In the book Don’t Make Me Think,2 author Steve Krug talks about how, as much as is possible, the purpose and functionality of a web page or app should require almost zero mental effort for the user to understand and use; the user should just “get it.” He uses terms like self-evident, obvious, and self-explanatory interchangeably to describe this.
Although he explains usability using the terms self-evident and self-explanatory as being the same, I think of it slightly differently. I usually explain usability to people using three categories of degree of usability, from suboptimal to optimal. These categories are “Explanation Required” (suboptimal), “Self-Explanatory,” and “Self-Evident” (optimal). We can apply these categories to any technology that provides an interface for a user to interact with directly (e.g., web app, mobile app, smart home). Note that the categories represent more of a spectrum than hard divisions.
The “Explanation Required” category means that the UX is not particularly “usable.” This means that for a user to understand and use a technology interface, some degree of explanation is required. This explanation can be in the form of training or documentation, for example. This is suboptimal and you should avoid it if at all possible.
The next level is “Self-explanatory,” which is pretty good, and where most good (not great) products are in terms of usability. “Self-explanatory” means that although not immediately obvious, a little careful observation, text reading, and context studying in the UI should shed the necessary light on what everything means and how to use it.
“Self-evident” is the optimal case. It means that zero explanation is required because everything about the UX is completely and immediately obvious. This is a very high bar to achieve and obviously won’t be the case for every user, but hopefully it will be for most. This is also a distinguishing characteristic of truly great products.
Delight and Stickiness
Delight and stickiness are two extremely important concepts when it comes to great products, and both lead to successful products that enjoy highly engaged and retained users.
Delight is a concept that is highlighted well by a framework known as the Kano Model, a product development and customer satisfaction theory developed in the 1980s by Noriaki Kano, a Tokyo University of Science’s professor of quality management.3 Delight is also a psychological and emotional concept, and it is incredibly important to the success of a product or service. People prefer to use products that are pleasurable to use and that makes them feel good (delightful). This is true even if it means giving up bells and whistles (features). Simple and delightful products are almost always better than the opposite alternatives.
Stickiness basically means that users perpetually continue to use a product after their first use; from the business perspective, this is a measure of user engagement and retention. We are all very familiar with the difference between sticky products and those that are not, even without being aware of this terminology.
Let’s begin by talking about delight. Product features which the Kano model categorizes as “delighters” are those that customers don’t expect and which provide a huge boost of enjoyment and an element of surprise. Delighters are the secret sauce; they are key product differentiators and generators of competitive advantage. Customer satisfaction increases exponentially with the introduction and enhanced implementation of delighter features. These are the true differentiators that get people to use your product over others, and also produce a great experience and level of enjoyment at the same time.
All companies should try to determine what the delighters for their products are, and then implement them. The potential ROI is huge, and can easily result in dedicated customers that also advocate and evangelize a company’s products on their behalf. Assuming that you’ve built a great product that is delightful, eventually the word will spread and the product will be a great success. As Dharmesh Shah said, “Don’t make customers happy. Make happy customers.”
One interesting thing to note is that today’s delighters are tomorrow’s must haves. Think about the iPhone. There are few products that I can think of that were introduced with not just one delighter, but many. A beautiful relatively high resolution touch screen with multiple interactions (e.g., pinching to zoom), a camera and photo library, music player and library, apps, and more. This was delighter city, and phones like this were called smart phones.
Now we’ve dropped the “smart” designation and all of these features are table stakes for all mobile phones—must haves. Also interesting is that AI is now becoming the source of delight in mobile phones with features such as automatic image categorization, facial recognition-based security, animoji, and photo optimization. Thinking ahead, I can easily see the time when phones, tablets, and computers no longer have keyboard interfaces and all interaction is speech driven. The concept of today’s delighters becoming tomorrow’s must haves is also in line with a concept called the AI effect; that is, when certain applications no longer seem to be powered by AI as they become more commonplace (e.g., Google Search). We’re seeing this happen now more and more.
Let’s discuss stickiness next. You might have a hundred or more apps on your phone. The question is, how many of those do you use every day? Further, how many do you use multiple times per day? How many once per week, and how many once per year? We all have apps on our phones that fall into one of these different groups of usage and frequency. Sticky apps (and products in general) are those that you actually use, and use often. Like delight, stickiness can be a huge driver of product success. In fact, the prospect of understanding what makes products sticky has resulted in people researching and writing about it.
In Nir Eyal’s book Hooked: How to Build Habit-Forming Products, he presents the Hook Model, which he created for building what he calls habit-forming technology. The model consists of four components of a feedback loop. These components are trigger, action, variable reward, and investment. At a high level, the idea is that triggers cause people to perform certain actions and behaviors, which in turn should be rewarded in a nonpredictable and desire-generating way (a concept similar to delight given the element of surprise), and then ending with the user making some sort of investment into improving the feedback loop the next time around. As Nir puts it, “These investments can be leveraged to make the trigger more engaging, the action easier, and the reward more exciting with every pass through the Hook”.
Delight and stickiness should not be ignored or underestimated when building AI-based products, or any products and services in general. In addition to the benefits of innovation and being able to successfully use emerging technologies, these are the key generators of differentiation and competitive advantage.
Now that we’ve covered the four components that great products should have, let’s discuss Netflix in this context.
Netflix and the Focus on What Matters Most
As noted in a MathWork’s whitepaper, Netflix points out that as compared to ultimate [AI] model performance—usage, UX, user satisfaction, and user retention are what it finds most important and better aligned with its business goals.
From a product perspective, these statements are extremely interesting, in that optimization for Netflix is much more user centric, as opposed to performance focused. This makes sense, particularly given that Netflix can afford some level of nonperfect performance because the company is not using machine learning for cancer diagnostics, for example.
The really interesting takeaway here is that the four things Netflix points out are directly related to the four things I am proposing that make a product great. Things that just work results mainly in retention, usability results in all three (great UX, satisfaction, and retention), delight results in high satisfaction and retention, and ability to meet human needs and wants results in high satisfaction and retention, as well. Always ask why and look to optimize the cause to achieve the desired effect, as opposed to the other way around.
Lean and Agile Product Development
Now that we’ve established what makes products great and what their purpose is from a human perspective, how do you ensure that you build successful, great products quickly, efficiently, and with minimal risk? It really comes down to two things. The first is ensuring that you have product–market fit and that your product is better than the alternatives, where “better” can mean price, functionality, and delight, for example. The second is that you’re actually able to successfully execute and deliver an innovative product that achieves product–market fit.
Figure 8-4 shows The Product-Market Fit Pyramid, the second framework created by Dan Olsen. Achieving product–market fit is especially important when creating innovative new products such as those powered by AI, and particularly products intended to capture new markets.
The Product-Market Fit Pyramid is logically split into two sections: the market and the product. When all of the levels of the pyramid are satisfied between a product and target customer market, the product is said to have product-market fit. This is what ensures product success.
The bottom two market levels indicate that identification of your target customer and their underserved needs is the foundation by which your product and possibly entire business should be based.
Figure 8-4. The Product-Market Fit Pyramid
The product then is built on this foundation, starting with the product’s value proposition. This is the why that I’ve focused on so much throughout this book. It can be framed in the context of needs, wants, goals, or whatever you choose, but ultimately this is the North Star that guides everything else. The value proposition is also where you need to determine why this product is going to be better, as mentioned.
From there, a set of product features and their associated functionality is determined to make the value proposition a reality, and then the UX is created. The UX is what the customer or user interacts with to experience and enjoy the value that the product provides.
Defining everything represented by The Product-Market Fit Pyramid involves creating hypothesis and making key assumptions that must be tested. This is where Lean and Agile product development comes in. In the broader product development sense, Lean and Agile methods such as kanban and scrum are intended to build and iterate quickly in order to test risky assumptions and build working software instead of extensive documentation. The working software mentioned then is able to be tested by actual customers, or nonbusiness people who are as close to customers as possible.
These methods also allow companies to test for product–market fit as quickly as possible and make changes or pivot as needed for success. There are many concepts and frameworks that people have created to help guide this process. They go by names such as failing fast, the build-measure-learn feedback loop, and the hypotheses-experiments-test-insights feedback loop.
Usually, the early part of this iterative process is based on the concept of a minimum viable product (MVP). As discussed earlier in the book, an MVP provides a mechanism to help mitigate risk and avoid unnecessary expenditure of time and cost. The primary idea is that the minimum amount of UX and software functionality should be built and put in users hands as quickly as possible, to properly test the riskiest assumptions and validate product–market fit and user delight.
From a qualitative perspective, an MVP should be delightful, usable, reliable, and functional. Another interesting concept is that of the minimum lovable product (MLP). Sam Altman of Y Combinator has said, “It’s better to build something that a small number of users love, than a large number of users like.” Whether it’s an MVP, MLP, pilot, proof of concept, or prototype, the purposes as covered are the same.
Summary
Great products aren’t great because of the technology that they’re built on alone. Great products should “just work,” meet one or more human needs or wants, are easy to use and understand, and are delightful and sticky. Creating an AI vision and strategy to achieve all of these goals will help ensure AI solution success.
In Ben Horowitz’s book, Hard Thing About Hard Things,4 he quotes a former boss as saying, “We take care of the people, the products and the profits…in that order.” Even though he’s referring to company workers, I think this 100% applies to building great products, as well. If products are built with customers and users in mind first and foremost, success and profits will follow.
After these considerations have been incorporated into your AI vision and strategy, it’s important to determine how to test your riskiest assumptions and user delight by defining and building an MVP or comparable testable solution. You can effectively facilitate this via Lean and Agile methodologies that will allow you to learn the most information and make any needed changes in the shortest time possible.
Now let’s turn our attention to the UX of AI and apply what we’ve learned, specifically, to creating better human experiences, a primary goal of AIPB.
1 Olsen, Dan. The Lean Product Playbook: How to Innovate with Minimum Viable Products and Rapid Customer Feedback. New Jersey: Wiley, 2015. https://dan-olsen.com
2 Krug, Steve. Don’t Make Me Think, Revisited: A Common Sense Approach to Web Usability. 3rd ed., New Riders, 2014.
3 Coppenhaver, Robert. From Voices to Results - Voice of Customer Questions, Tools, and Analysis: Proven techniques for understanding and engaging with your customers. Packt Publishing, 2018. http://bit.ly/2VK9CQ8
4 Horowitz, Ben. The Hard Thing About Hard Things: Building a Business When There Are No Easy Answers. New York: HarperCollins, 2014.