Decision Records & Retrospectives
Decision records help sustain institutional memory of the rationale for decisions. When combined with decision retrospectives, they help an organization make better decisions by learning from previous ones.
What is it?
It is a record of key decisions along with arguments (in favor/against/alternatives etc.) along the way. It is useful for the categories of decisions that result in substantial deployment of resources and/or have proven to be controversial or painful in the recent past.
How does it help?
It helps the organization learn to make better decisions faster without too many meetings. It allows for real empowerment via autonomy for decision making and balances it with a culture of accountability.
How does it work?
We maintain a record of important decisions and hold periodic decision retrospectives to learn from hindsight and to reflect if the decisions we made were the best given the information available at the time.
First, an executive sponsor comes up with a set of decision categories that could benefit from a decision record. Examples of common categories are product strategy, funding, sourcing, build-vs-buy decisions and staffing.
Next, they establish decision owners and a circle-of-input providers for each decision category. Decision owners have the autonomy to make decisions and are held accountable for them. They are required to solicit written input from their circle-of-input but not mandated to go by the input. This gives them autonomy for decision-making.
In order to balance decision-making autonomy with accountability, we have the decision record. It dissuades decision owners them from disregarding input and helps the group retrospect and learn from previous decisions. To make this process effective, we capture along with decisions, their expected effects and expected time-frame for the effects to play out. This information is used during retrospectives.
The decision owner or anyone from the circle-of-input can table a topic as something that requires a decision. After a few days of offline deliberations, the decision owner makes a decision along with a short description of what effect the decision is expected to have by when. The decision owner may also nominate someone from the circle-of-input as decision-maker on a case by case basis.
Recording discussions vs. decisions
Although it is a good first step, only capturing decisions does not ensure that all relevant parties are heard. For that, we a discussion record and a decision record. The best discussion records are those that aren’t captured after a discussion (as in, minutes of meeting) but captured as they happen. This calls for a culture of discussing in writing, not easy but doable with executive sponsorship and change management.
What tools could we use?
If we only wish to record decisions, and not the discussions leading up to them, we could just use the corporate document repository. For discussion records, we could start simple with a corporate discussion forum per decision-making group. Once the practice catches on, we could graduate to structured discussion tools such as Kialo.
The picture above is a visual of a structured discussion (go to info/stats/topology in the menu on the linked page) on the topic of decision records. The grey circle at the center represents a proposed decision. The green and red areas surrounding it represent arguments for and against (a) the proposal (first layer) or (b) the arguments (subsequent layers).
Decision Retrospectives
Decision records become powerful when we supplement them with the practice of retrospecting on decisions. Decision retrospectives are similar in spirit to Agile retrospectives. Once a quarter, each decision group picks the decisions that haven’t been covered in retrospectives so far and whose expected effect time-frames have elapsed. For each decision, the group compares actual effects and time-frames with what was recorded as expected. The reflect if the decision was the best one under the circumstances and discuss potential learnings.
Expectations
For non-urgent decisions, decision-owners shall allow for a minimum of days (e.g. 5 days) for discussion before making a decision.
For urgent decisions, it might work better to have a meeting and record the key arguments later. The decision owner posts the summary and input providers reply with corrections, if any.
Decision owners are expected to read threads initiated by others within a certain maximum number of days (e.g. 3 days) of posting. While there is no requirement to respond, decision owners will be assumed to have taken the input into consideration in future settings.
Mechanics
A new discussion thread can be:
Initiated by decision owner
Soliciting input towards a decision to be made
Soliciting feedback for a decision that’s nearly made
Soliciting input for change of earlier decision
Soliciting feedback for a proposed change of decision
OR
Initiated by input provider
Propose a new topic that requires a decision
Suggest a change to a decision
Affirm a decision with new information
When a decision is made, the decision owner shall state the decision in the thread. Threads that don’t culminate in a decision may represent stalemates and a potential indicator of organizational dysfunction. If a decision goes against certain inputs, it is good practice to explain why the decision was made despite the input.
Where has it been used?
Decision records based on Cleararchy have been used in several situations:
Prioritization decisions: How do we decide what projects to invest in or what to prioritize in a product roadmap? Traditionally, these decisions are made by a committee that reviews project or feature proposals. Unfortunately, these decisions are often subject to the power dynamics within the committee. The combination of decision records and retrospectives helps create traceability and accountability while at the same time also offering the opportunity to learn from previous decisions.
Product strategy: Sometimes, we arrive at a fork in the road in terms of product strategy and are forced to choose. In one organization, the product group choice often differed from that of an innovation lab. Cleararchy based decision records helped break decision deadlocks and reduce animosity between the two groups.
It also been used to break through differences of opinion between sales and engineering as to the timing of a release, between architects on architectural choices, and between managers on changes to team staffing.