Action Research (AR) is a methodology that turns a business — or any activity — into hierarchical plan → act → observe → reflect cycles. The Research Agent keeps the cycles as library documents: the reflection of one cycle becomes the plan of the next, so the work winds into a spiral instead of running in place.
The cycle
A cycle is plan → act → observe → reflect, and it never really “finishes”: continuous iteration until the desired result — and beyond. The phases are not isolated — act and observe intertwine, and reflection goes on through the whole cycle, not only at the end.
Agile in software development is one implementation of Action Research: the same spiral of short iterations with observation and adjustment. AR applies it to any business.
Three levels
- Strategic (the hour hand) — the evolution of the business as a whole: the business plan, KPIs →
research/strategic-cycles/; - Tactical (the minute hand) — the evolution of individual directions: the offer, the marketing task queue →
research/tactical-cycles/; - Applied (the second hand) — concrete tasks as mini-cycles →
research/applied-cycles/.
The clock rule: the seconds must move the minutes. Short cycles are obliged to genuinely move the long ones — the speed of your business’s evolution equals the speed of cycle iteration at every level.
Double-loop learning
Single-loop learning changes tactics while keeping the same assumptions; double-loop learning changes the assumptions themselves. The strategic cycle explicitly decides what it is doing: optimizing the model (first-order) or changing the assumptions (second-order). Above the levels runs a meta-cycle — a periodic review of the cycles themselves: which are alive, which have run out of steam, where iterations are missing.
Rules for running cycles
- the hypothesis is public and testable, with a verification metric — not “let’s try and see”;
- execution records only the facts of what was done; plans and checklists live in the planning section;
- new stages of a cycle are not written in advance — observe and reflect appear as data accumulates;
- reflection ends with a concrete decision: continue, change, or close the cycle;
- brevity, and no interaction during the process: the work runs on the best available assumptions, which are listed at the end.
Live examples
The product ships with our live cycles: strategic C1 (the business plan), C2 (the Action Research methodology), and C3 (KPIs); tactical — the offer and the marketing task queue. See Built-in Research.