Luis Goncalves
Last updated on | Agile Coaching/Team Coaching

Two Learning Cycle Models Every Coach Should Use

by Dennis Mansell
learning cycle


In a post on my own blog, I already considered why this subject can come up for Scrum Masters and Agile Coaches.  In this guest blog, I will take a deeper dive into which learning cycle models I use and why.  Thanks Luis for this opportunity to share my views!

Learning Models

The first thing to answer is why would you want to use a learning model at all?  Surely we all know there are different ways of learning and we have spent so many years at school and university that we should know how to learn and therefore teach by now?

Recent research into cognitive disfluency however points to the advantages of formally applying a model and pausing and committing during learning.  Basically: if you interrupt the flow of your learning just enough by including the testing, questioning and applying of knowledge, you will gain a deeper understanding more quickly and will be less prone to the drag of preconceptions.

There are a number of ‘learning cycle’ models, of which I will handle two quite similar versions.  If it is not obvious, I prefer cyclical (i.e. incremental & iterative) models over phased/developmental (i.e. waterfall/control gate) models.  This is because continuous improvement is not possible if the process has an end.  As obvious as this sounds, think of the most well known learning model: class learning in school.  That is a waterfall with one or more control gates – usually an exam – aimed at a qualification.  It will usually go something like:




The great thing about cyclical models is that they are easily adapted to specific circumstances.  Think about how you approach something like building an Ikea closet.  Some people:

  1. read the instructions first
  2. systematically check that they have everything they need
  3. build the thing
  4. post a pretty picture to internet, receiving much likes and kudos from their admirers.

However – if you are more like me – you will:

  1. blindly start constructing
  2. post a picture of your failed first attempt to internet
  3. read the instructions and then your friends’ ridicule and hopefully helpful advice
  4. discover what you did wrong
  5. redo most of what you did
  6. take another picture, etc etc.
Dennis' learning cycle
Dennis’ Learning Style

A cyclical model does not say where you should start – it is adaptable to the situation and personal preferences in each case.  It also does not have an end, you do not have to stop at a particular place or after any number of phases.  It can iterate at different speeds – a business cycle can take a year, building a closet should take at most a few hours.  

DIBBs: data, insights, beliefs, bets

Based loosely on the empirical method, I first heard about DIBBs on the (rather excellent) Boss Level podcast where Sami interviews Simon Marcus, VP of Operations at Spotify.  Apparently they use a system where they take data from a certain experiment, and see if they can glean insights from this data, they then try to induce beliefs that change or add to their theory and then they deduce bets that they can test in the next experiment.

For example:


Data from an A/B test shows that


was clicked on 870 times, whereas


was clicked on 1237 times


Both logos have similar colors and layouts, however B has only three elements whereas A has six elements.


The fact that B has only three elements is causing it to be more appealing to customers.


If we change logo A to have only three elements, it will be clicked on the same number of times.

Kolb – Experiential Learning Theory (ELT)

I actually came across the Kolb ELT model whilst training as a sailing instructor more than twenty years ago and began using it to train sailors, not software teams.  Now, based on what you know about my learning preferences and your guesses about your own, you should be able to think what the best way of learning about the Kolb model is for you:

  • Keep reading this post – a slightly jumbled mashup of two different theories, blended with my own experience.  Entertaining, inaccurate but hopefully easy to absorb.
  • Read the article on which is a good reference for the theoretical basics; enough to get up to a good working understanding of the subject and think up some training or workshop to incorporate the model.
  • Read David Kolb’s book.  That way no one will be able to say you did not research the subject properly.  Plus it will give you a chance to prove your immense perseverance and unwavering character.  Unfortunately that means reading Honey and Mumford’s book as well, but luckily that will be a breeze if you managed Kolb.

Axes and Cycles

The ELT model works in two ways that overlap: learning styles and the learning cycle.

The learning styles are set along two axes that depict two factors of your learning preference: Processing and Perception.


If you are someone who builds the closet before reading the instructions, your style is more ‘doing’.  If you start your closet building attempt by watching YouTube, you are probably more ‘watching’.


If your primary love of closet building is the laying out of all the screws and allen keys and planning your build, your style is ‘thinking’ whereas if you just want the satisfaction of the new closet and would happily have someone make it for you, your style is more ‘feeling’.


It is likely you will have a preference on both of these scales, such that you can place yourself somewhere in this field.  I have a strong preference for doing and although I enjoy planning making furniture, I am especially in it for the result.  This is of course just a preference in this case, we all possess the ability to adapt to different situations and to start learning in a less preferred style.  Even though it is not my natural inclination, I actually have read a manual before starting on a project.

Learning Cycle

Kolb sees learning as going through a cycle, which you must go through to learn about a certain topic.  Where you start can be a matter of preference but going through all four of the phases at least once is the best way to learn.  This makes Kolb different from a model like the MBTI, which places you statically on a scale – in Kolb’s ELT moving through the different phases is expressly the purpose.

The four phases of Kolb’s cycle are

  1. Concrete Experience – this is the raw data, the output of an experiment, be it the feeling you get from an Ikea closet or the numbers from an A/B test.
  2. Reflective Observation – the insights from the data, how does the data compare to your hypothesis?  What are the differences between A and B due to?  How does your closet compare to the drawing in the manual?
  3. Abstract Conceptualisation – the beliefs that logically follow from the insights.  This is our new hypothesis as to the effectiveness of a logo.  In the closet example: perhaps I should have used the long screws like it says in the instructions…
  4. Active Experimentation – test the bets that follow from our beliefs.  Doing a new A/B test with the same number of components in both logos.  Taking apart the closet and replacing all the short screws with the long ones.


You can also see the phases as the different phases of the scientific method (starting at the bottom): hypothesis, experiment, analysis, conclusion, theory/new hypothesis, experiment (etc).  The ELT cycle can therefore be overlaid on any number of Agile feedback loops.  By looking at these loops explicitly as learning cycles, we can start/facilitate continuous improvement and empirical management.  Especially, we can structure what outcomes we need at a certain point in the cycle:

  • if we have just released a piece of software for example, we can see that as a form of Active Experimentation.
  • We should go on the lookout for data – the Concrete Experience.
  • Once we have the data, we should look at whether this is what we expected it to be and why/why not – Reflective Observation
  • We should then update our theories and think up a new hypothesis for testing –  Abstract Conceptualization.

If you prefer, DIBBs is just another way of naming Kolb:


Some more examples

Actually, that is not the only cycle that can be drawn into the Kolb model.  As I said before, the nice thing about a cycle is that it can be iterated at the speed you need:



Daily Scrum



Design Thinking
Design Thinking

I am aware that this way is very deterministic – that these phases are not always run through in this order for example and that there is actually a correlation between perception and processing such that they are not true ‘factors’ in the way that a mathematician would define them.  That is however missing the point – whatever you think of the theoretical accuracy of these models, I am convinced that if you apply them in the spirit that every complex process has possible unknown outcomes, that you can always learn from new experiences, then they cannot do much harm.

Rather than labelling someone as a ‘doer’ or ‘thinker’, it helps to realise that some people learn by doing first or thinking first but eventually your team effort should be focussed on getting the most out of all the aspects of an experience.

Rate this post:


Share your point of view

  1. “Recent research into cognitive disfluency however points to the advantages of formally applying a model and pausing and committing during learning.” – Could you cite the source here? Thanks.

    1. Hi sm,
      Sorry for not answering sooner, I don’t get notifications of comments on Luis’ site. There is actually a link in the sentence that you quote to Edge magazine. Adam Alter’s paper is published here: along with lots of references to other research if you are interested. I first heard about cognitive disfluency from a chapter in the book ‘Smarter, Faster, Better’ by Charles Duhigg – which is less complete but easier reading.