How Do We Avoid Perfectionism and still Manage Quality?
I did an episode on perfectionism recently. I argued that it is wasteful and thus should be eliminated. But, in some cases, it seems to be our only way to control quality. That is, our only quality target is perfection. If we eliminate it as a goal, then are we simply left with accepting sloppy work from our Knowledge workers?
In this episode, I’ll talk about some of the challenges of quality in knowledge work. We should arrive at some actionable ideas of how to manage quality in knowledge work. Can we have quality targets as knowledge workers?
We definitely want to avoid perfectionism. And we also want to support timeboxing. We want to support the idea that this should take me four hours, and then schedule four hours, and get the work done in four hours. Versus some perfectionistic approach where we work on it, work on it, and continue to work on it until we run out of time. And then we declare it done. That’s not a thoughtful way to proceed and definitely not managing quality.
Our productivity demands that we manage quality. And we need to manage quality because we need to manage cost. Otherwise, we go out of business. But how?
How to define quality in Knowledge work
Only you, as the knowledge worker, can know what you are trying to say or deliver
Example: delivering truth when it’s bad news
What is quality?
An outcome
Quality is like school exams (at least some of the time)
If we have an objective external standard
Where does production quality fit in?
‘production quality’ above a certain threshold is not a measure of quality
Unless it is. Entertainment seems to have this property.
Kinds of flaws in knowledge work
Non-exhaustive categories of knowledge work and the associated deliverables
Factual Knowledge – what are the facts, what do we know? E.g. diagnosis
Process Knowledge – next action that should be taken, E.g. treatment regimens and delivery
Understanding Knowledge – what is our strategy? Changes to standard procedure e.g. when treatment doesn’t work
Knowledge Creation – new knowledge/processes/workflows – Insight e.g. new treatments
Problem solving – practical application of kinds of knowledge, experiments to try, “this should work” e.g. confounding symptoms
Defining quality per type of Knowledge Work
In KW, we don’t have an objective external standard – usually one-off work products
Value in use
KWs produce results that are necessarily incomplete and probably incorrect in one or more details
Factual Knowledge – high quality = correct (“true”, “accurate”) to the appropriate level of detail
The customer ‘doesn’t like it’ does not count against quality
Process knowledge – high quality = ‘doability’/ease of use/regulatory compliance
Again, ‘doesn’t like it’ doesn’t count
Understanding knowledge – high quality = reasonability, experience, convincing, case study
Knowledge creation – high quality = science
Problem Solving – high quality = results (vs. cost)
In these areas, production quality has a lower bound of comprehensibility, but improving it beyond that is probably a waste
Thinking in Bets – “resulting”
If our results have flaws, then our work was bad work?
What is ‘resulting’ and why is it bad
Redefine what a ‘good’ decision is
The key is that the problem of resulting divorces our outcomes from the quality of our effort
Why is resulting bad – it causes us to doubt, to change good processes for bad ones
Avoid resulting in decision making – follow the decision process
How to avoid resulting in Knowledge Work results – how do we know we’ve done good work
Athletes and results (three point shooting)
Knowledge work results have 1) hidden information (e.g. user needs, uncertainty about facts), and 2) risk
So, knowledge workers have done good work when we’ve followed our process
Knowledge workers have to avoid ‘resulting’
Helps avoid the perfection trap (overinvesting in quality)
Avoiding resulting in our work
It leads to perfectionism, which is wasteful
Without other standards, we default to production quality, which is probably wasteful
Avoid making our delivery processes weak (no ‘change resistance’ or consistency)
Avoid resulting’s sapping of our confidence
What to do about quality?
As knowledge workers, we need to get comfortable with ‘best effort’
Develop a process to adopt changes to our processes, don’t change our work processes at the drop of a hat
Scrum/Agile project management can help, but the user has to be highly involved
Use lots of MVPs (Minimum Viable Product)
What else to do
Develop confidence
Focused work for our target amount of time is ‘good’ work
An experiment is good to the degree that it produces usable data
Not to the degree that it supports a specific hypothesis