Accountability governance and power analysis · Lesson 1

Decision register and accountability

6 min read

Lesson objectives

  • Apply decision register and accountability to a fictional humanitarian case and justify the decision, limits and observable output.
Reading tools

Community accountability is more than a complaints box at the end of a project. It is a relationship around decisions: who can influence the choice and conditions of a service, what information enables that choice, how people can see the effect of their input, and where they can turn when a programme gets something wrong. The 2024 Core Humanitarian Standard connects participation, safe complaints and learning from feedback. The standard does not, by itself, make every community proposal binding. A professional response states which decisions are open, weighs safety, equity and resources, and explains the reasons for the eventual choice. A promise to listen without a decision owner or a response date leaves people unable to distinguish consequential participation from data collection.

Start with one decision unit rather than a broad ambition such as improving participation. A usable question is whether to add a distribution point before the next registration cycle. Write the decision as an action, outcome and decision date. Then separate four roles that are easily confused: people entitled to describe the problem, those who provide evidence, the person authorised to choose, and the person responsible for communicating the answer. A community group can identify an access barrier without assuming responsibility for an organisation's budget or investigating a sensitive complaint. If the organisation announces one option as final before consultation, the consultation cannot alter that option. The register should show what remains negotiable and which resource or agreement constraints are already fixed. Participants can then spend effort on a real choice, and a reviewer can later see whether their input reached the authorised decision maker.

A decision register needs seven working fields: the question, affected groups, community evidence and how it was gathered, feasible alternatives, decision owner, reason for preference, and date for checking the result. Keep a separate response status: were the affected groups told what was and was not decided? Do not treat message counts as a population survey. Twenty calls from available phones may reflect one connected group while people without phones remain invisible. Check the denominator and channel coverage before labelling messages the voice of the community. Naming a manager in the owner field is also insufficient. The manager needs to know when a redacted synthesis reaches them, what they can change, and when an answer must go back. Identity data needed to follow up an individual case need not enter a collective decision record when aggregated patterns are enough.

Stress-test the register with a changed fact. If the eighteen reports were verified as distinct households across phone and oral channels, and two successive eastern trial days were fully used without increasing western waits, a permanent point would have a stronger case. Finance would then need to assess sustainable funding and effects on existing services, rather than merely ask whether plausible demand exists. If instead two people made fourteen calls because nobody answered them, correcting the response and transport arrangement should precede spending on premises. Neither possibility is assumed true. The register captures the hypothesis and the observation that could overturn it. A selected option without a review condition can quietly become permanent policy, even when its original evidence was weak.

Treat communication as part of the site decision. An honest message should not say every demand was met when the change is a weekly trial. It should state what will change and when, why no permanent office has yet opened, and how someone missed by the original channels can report an access problem. It should not reveal who raised the concern. Use channels that can reach groups missed by phone, and check receipt and understanding rather than only recording that a notice was sent. If people do not know the day has changed, a low turnout could be mistaken for lack of need. If turnout rises, that alone does not prove the participation process was good; travel may be easier or timing more suitable. Separate observed facts, interpretations and the next test. The accountability cycle then has a traceable input, a reasoned choice, bounded implementation, communication, measurement and correction.

Lesson summary

Accountability begins with a decision that people can understand and challenge, not with a feedback box detached from management.