Whenever I examine a project, system or line of business then the first step that I normally take is to map it. Mapping is an easy process, with experience then creating a basic map should take no more than a couple of hours. To map, you start with user needs, then determine the components required to meet those needs and map according to how evolved they are (see figure 1).
Figure 1 - A Map from HS2
Now, there are lots of reasons for mapping but in this post I'd like to focus on the question of experimentation or planning? If you have a map, then the uncharted space (see diagram above from HS2 - high speed rail) is where you experiment and the industrialised space is where you plan. To make this clear, I've marked this on figure 2.
Figure 2 - Where to Plan, where to Experiment
A couple of things to note.
1) Web site is marked as commodity but using Agile development, why? The reality is the web site is a mass of components, some of which are commodity plus others (such as content / structure / data) which are novel. Hence this particular point was broken down into its own map. Whenever you deal with high level concepts then there is loss of granularity - no different with an atlas losing structure on roads / buildings etc. Maps (both geographical and this type of Wardley map) are imperfect representations of what is there.
2) All the components are evolving. What you started by experimenting with will over time become something you plan.
In general, as a rough guide, when you have a map then the following methods and purchasing techniques are applicable.
Figure 3 - Methods and Purchasing techniques.
WORDS OF WARNING
Those who are used to mapping (and that varies from Silicon Valley startups to large commercial companies to huge Government departments) don't need to be told that one size fits all methods don't work. Many others make the same mistakes over and over again.
There are entire industries of book sellers and consultants out there trying to flog you a one size fits all methods such as Agile or Lean or Six Sigma. It's misguided. You need to use all three approaches with any large complex system. They can all point to examples of how their technique beat the others, the opposing camps can do the same. All three techniques are actually useful.
The first step in an OODA loop is OBSERVE and that's the bit mapping tackles, it provides a communication mechanism to describe an environment between multiple groups. It doesn't tell you what to do or how to do it which is why once you have a map, you need to apply THOUGHT and orientate around it.
Agile & in-house development tends to be best for the uncharted space, the novel and new, the areas where you need to experiment because you lack information and will change rapidly.
Lean & off the shelf tends to work best when you have some information, you need to remove waste and focus on delivery of a product (i.e. the transitional space between the uncharted and industrialised).
Six Sigma, outsourcing and use of utility providers tends to work best for the common and well understood when you have reasonable information, need to focus on removing defects and operations at scale i.e. the industrialised space.
Any significant project will have components at each stage of evolution and require all three methods. Each method is good at what it does but apply a one size fits all method or purchasing technique and you won't run as effectively as you could. Before you say - "aren't you trying to flog me mapping?" - the entire technique is Creative Commons Share Alike. There is nothing to flog. If I'm trying to persuade you of anything then it is "look before acting" and "use all the methods according to their strengths".
The only people who can map a business are people who are running, operating and working in the business i.e. YOU. There is no need for consultants as you only need yourself and others within the company. Those who map have already discovered this. I need to emphasis this - YOU have all the power you need to map, to learn common economic patterns and to learn how the play strategic games in business.
Of course, you'll have to go and learn those individual project methods and purchasing techniques such as agile, lean and six sigma and for that then you'll probably end up requiring consultants and books but then I'd consider that investment in training in specific methods not a ONE SIZE FITS ALL to be applied across the organisation.
So when it comes to experiment or plan and which should you do, then the answer is BOTH for anything of significant scale (i.e. beyond a pet project).
Cue the endless noise from one size fits all consultants claiming their method Agile, Lean or Six Sigma is applicable everywhere and the other methods aren't.