Someone should turn on the lights

The Hawthorne experiments revealed that productivity increases were not due to physical conditions but rather the psychological effect of being observed and chosen. This highlights the importance of observation over predefined success metrics in experiments. Flexibility and openness to unexpected insights can lead to transformative changes in human behavior and organizational processes. #HawthorneEffect, #Productivity, #Psychology, #Observation, #BehavioralScience, #OrganizationalChange, #Experimentation

Episode Transcript

(00:00):
Welcome to the Colig Experience Podcast. In this episode, we're diving into bold insights,

(00:10):
innovative experimentation, and groundbreaking leadership strategies that can revolutionise
how you approach challenges.
Today we'll explore the fascinating findings of the Hawthorne experiments from the 1920s,
revealing how observation can unlock unexpected insights and transform organisational success.

(00:31):
Let's dive in.
Someone should turn on the lights. The Hawthorne experiments are the collective name for a
series of sociological experiments conducted in the 1920s at a factory with the unsurprising
name of Hawthorne. In one of these experiments, researchers wanted to test the effect of lighting

(00:51):
on the productivity of a particular unit in the factory. In the first phase of the experiment,
they examined what happens to productivity when lighting levels are increased and they
discovered that productivity went up. In the second phase of the experiment, they lowered
the lighting levels and it turned out that the productivity of that same unit continued

(01:12):
to rise. These surprising results led to a deeper investigation of the phenomenon and
what became clear was that the factor causing the increase in productivity of that unit
wasn't related to the physical conditions but to the fact that they were chosen from
among all the factory workers to be part of the experiment. The very fact that they were

(01:33):
specifically chosen to serve as part of the experiment, along with the close monitoring
by the research team from Harvard University, caused the workers in that unit to dedicate
themselves more to their work and achieve higher performance. This story is a wonderful
example of how the very existence of an experiment influences what we learn from its results.

(01:56):
By the way, for us, the connection between experiments and our understanding of the reality
around us seems natural and self-evident. But the truth is that the requirement to prove
or disprove ideas through experimentation is relatively young, only about 300 to 400 years
old. As strange as it sounds, those who dealt in ancient times with developing theories about

(02:21):
how nature operates were freed from the need to prove them. And not only that, if the theory
contradicted observations made in nature, this wasn't considered a refutation of the theory,
just a light blow to the wing at best. This leads us to an important distinction between
observation and experimentation. When we conduct an experiment, we usually assume expected results

(02:46):
in accordance with our thesis about what happens in reality, that same thesis that the experiment
is supposed to prove or contradict. Yes, yes, it's true that here and there within the framework
of such experiments, significant breakthroughs were discovered in fields that weren't directly
related to the experiment, serendipity, penicillin, and so on. Observation, on the other hand,

(03:11):
is looking at reality in order to derive from it insight, explanation, or prediction about what
is happening or will happen. Does this sound like a theoretical and irrelevant issue to you?
Think again. The term pilot is surely familiar to you. It's a term borrowed from the technological
arena into the organizational and process arena for the sake of more careful and wise decision

(03:36):
making. On the table lies a proposal for significant organizational or process change.
On one hand, it holds promise for positive improvement, but on the other hand, it's
unclear whether it will work and whether all the assumptions that need to be fulfilled for it to
succeed will actually occur in reality. So what do we do? We test. And how can we test? We do a pilot.

(04:02):
That is, we conduct a small experiment and examine the results. If they support the general idea,
we can move forward, and if they don't, we can go back to the drawing board to think about a new
solution. Pure science, really. The thing is that, unlike technology, when dealing with processes
that involve human behavior, it's more appropriate to move from experimentation to observation.

(04:29):
Here comes the explanation. If we had a shekel for every time we participated in a discussion about
a pilot where the question comes up, how will we know the experiment succeeded? What are the KPIs?
The metrics. And we ask, why is this important? Why not do observation and see what we learn

(04:50):
from the proof of concept? After all, if we're already doing the pilot, why define in advance
what we'll do at its conclusion? Why not wait for its completion and examine the new reality?
Maybe we'll learn new things from it that wouldn't have occurred to us at the beginning of the
experiment. When we define in advance what constitutes successful results of the experiment,

(05:12):
and what constitutes unsuccessful results, we're assuming that we know in advance all the possible
outcomes. This assumption is what causes us to focus on very specific aspects of the experiment
and ignore everything else. We suggest a slightly different process. Our assumption is that when
we introduce change, even the smallest one, there can be interesting and important effects that

(05:38):
aren't necessarily related to what we knew at the time of deciding on the experiment.
Therefore, we encourage not trying to control the experiment by defining in advance standard
operating procedures for every result, but rather to let it develop, stop after some time,
and examine the new reality in relation to itself and not in relation to what we thought when we

(06:00):
set out on the journey. Instead of dealing with what was important to us when we set out on the
journey with the experiment, to focus on what we know after the experiment when we're wiser and
can make a decision about the next experiment, let us share two examples from our consulting
work that illustrate this approach. We once worked with a large media agency that wanted to pilot a

(06:24):
new client communication system. Instead of defining success as response time under two hours or
client satisfaction score above eight, we encouraged them to simply implement the system with one client
and observe. What they discovered was completely unexpected. The new system didn't just improve

(06:46):
response times. It fundamentally changed how account managers prepared for client meetings,
leading to more strategic conversations. This insight became far more valuable than any
predefined metric would have captured. Similarly, we worked with a software company that wanted to
test a new code review process. Rather than measuring bugs reduced by 30%, we suggested they

(07:12):
observe how team dynamics evolved. They found that junior developers became more confident
contributors and senior developers started documenting their knowledge differently.
These behavioral changes ultimately proved more transformative than any bug reduction metric.
Why is it still important for people to define in advance what success of the experiment will be?

(07:36):
Sometimes the experiment is a way to figure out who's right. When there's disagreement about which
way to go, there will be those who want to establish the definition of success in advance,
because that will be their way of saying, we were right. And it's likely that they will also
help the experiment succeed. In parentheses, let's say that if the experiment comes to

(07:58):
decide on a dispute about one work method or another, then even if success metrics for the
experiment are defined in advance, it will often be impossible to agree on them as well.
Each side will explain why the results justify or don't justify their claim, whether they're
relevant or not relevant to the discussion, and so on. We don't know many cases where an

(08:22):
experiment led one side to become convinced. Therefore, if the problem is a dispute about
way A or way B, think about another mechanism to resolve the dispute. An experiment won't help in
most cases. Close parentheses. Sometimes the desire to define in advance what will count as
success of the experiment stems from naivety, from thinking that the human factor behaves in a way

(08:46):
that can be analyzed unambiguously, and from which unambiguous insights can be derived.
Fortunately, this isn't the case, at least for now, until the algorithm learns to predict this too.
And sometimes it stems from our natural need to control uncertainty.
There's uncertainty about the future? It's unclear what the appropriate decision is.

(09:11):
It's hard to know what the experiment will yield. We define in advance a space of results and
possible response to each result. This seemingly reduces uncertainty and provides a pleasant
illusion of control. As we said, we're in favor of being in constant dialogue with reality,
to change, to try, and then to feel the new reality and move on to the next experiment.

(09:37):
Organizations that develop the ability to constantly move between experiments and
observations become more flexible, open to changes, and also more interesting.
You're welcome to try this at home.
And that's a wrap for today's podcast. We've explored how the Hawthorne experiments

(10:00):
illustrate that observing and adapting to real-world outcomes can yield more valuable insights
than sticking to predefined success measures. Don't forget to like, subscribe, and share this
episode with your friends and colleagues, so they can also stay updated on the latest news
and gain powerful insights. Stay tuned for more updates.