Quality Function Deployment - The Quality Chart

"The iPhone is a quality product."

Let's think about that statement for a while. Do you agree with it? Why? What does the word "quality" mean for you? Do have any colleagues/friends who do agree with it? What does "quality" mean for them? What about your colleagues/friends who do not agree with it? What does quality mean for them? Hopefully it's clear that 'quality' is a subjective aspect of a product. One person's quality is not another person's quality.

This blog post looks at quality. There's a lovely methodology that helps us determine how to build quality into a product. It's called Quality Function Deployment (QFD). I'm going to be blogging a bit about the QFD methodology and will start with the first model in QFD – the quality chart. The quality chart helps us understand what our target audience mean by quality - and what elements we need to build into a product to ensure that we are achieving quality for our audience.

Along the way we look at using the quality chart to drive product development - an interesting side topic!

The Quality Chart
The Quality Chart is, basically, a matrix linking customer demanded quality items with quality elements that we can deliver. It's designed to help translate a customer's perception of quality into a set of elements in the language of the company engineers (or software developers :). Consider this example (which I just made up for a phone  - a proper analysis might have quite different things):

The items in the 'demanded quality' side of the matrix are things that have come up in customer research. To determine the quality elements that the product must have to deliver the demanded quality, we first determine all the elements (the top of the matrix) that are necessary to deliver the demanded quality and then analyze the demands to define exactly which elements are necessary to deliver each attribute of quality that is demanded.

We then rearrange this matrix to be explicit about which elements support which demanded characteristics.

• finish does not degrade - scratch resistance, waterproof
  
• easy to find features - consistent navigation; unambiguous naming scheme
  
• etc, etc, etc.

This view of the information tells us exactly which elements we need to deliver a particular attribute. The advantage of this view is that each demanded quality characteristic appears only once in the list - it may appear multiple times in the matrix due to the decomposition.


Product Selection, Creation, and Modification

A really useful thing about the demanded quality items is that we can group sets of items into products. For example, one product might have demanded quality items 1, 2, 3.1, and 4.1.2. Another product might have 1, 2, 3, 4.1.3. This lets us carve up our products into a user centered set of demanded qualities - and then determine the quality elements we will have to build into each product. As our understanding of demanded quality and quality elements changes, we can: update the matrix, determine the impacts on the quality elements for each product, and produce new versions of each product for the next release.

This process of updating our quality chart provides a product delivery cycle based on the demanded qualities of our products - quite a useful because as customer expectations (demanded quality) change, our product's features (quality elements) changes to match.

Decomposition of Demanded Quality
Demanded quality is what our customers/users/stakeholders want. We produce this though customer workshops, interviews, surveys, feedback, market research, etc, etc, etc. The key things about demanded quality items are:

• Items at the 3rd level must be SMART (specific, measurable, achievable, realistic, time-bound).
• Items should not be restricted to a particular product. Let the people determine the scope of the items.

Personas and Scenarios (as exposed by Cooper) might also be very useful in determining demanded quality.

Decomposition of Quality Elements

Quality elements are, essentially, what we can build into a product. The experience that the product team has will impact the comprehensiveness of the set of quality elements. Because creating a comprehensive list is hard, it's more important to prioritize elements than attempt comprehensive analysis.

Luckily, the ISO has defined 6 quality elements for software:

• Functionality: are the required functions available, including interoperabilithy and security
• Reliability: maturity, fault tolerance and recover-ability
  
• Usability: how easy it is to understand, learn, operate the software system
• Efficiency: performance and resource behaviour
• Maintainability: how easy is it to modify the software
• Portability: can the software easily be transfered to another environment, including installability

We can (probably) produce a standard 3 level docomposition for each of these. For example, usability might break down at the 2nd level to:

• easy to understand
• easy to learn
• avoids errors
• guessable (or 'intuitive')
• user experience
  
• etc

then we can decompose each of those further - to a level that is clearly measurable.

Is this a thinking tool?
For any product of reasonable complexity, the quality matrix is going to be large. This means that an analyst won't be able to cognitively understand and reason about the entire quality matrix. However, it does give us these abilities:

• Confidence that the features being built/designed will achieve the demanded quality
• Ability to perform an impact analysis of a particular feature not being delivered
• Ability to prioritize quality elements for design and development.

Those are all very good things. However, because of the lack of abstraction and ability to cognitively manipulate the matrix (like we can do with use cases and use case models), I call this a reasoning tool - it helps us reason about quality and determine impacts.