Principles for working with outcomes systems


Academic references to outcomes theory

How to use these principles.

These principles are used to identify problems within outcomes systems, including the absence of key outcomes system building-blocks and key trade-offs which have to be made in setting up outcomes systems.

Outcomes theory principles.


Only some initial principles on this site so far.

There are a large number of outcomes theory principles. Some have been put up below, others are being progressively added. To be notified when they are put up, subscribe to the DoView School of Outcomes Tips newsletter.

Don't siloize means under ends

In real-world programs, one mean may contribute to more than one end (i.e. one project may contribute to more than one outcome.Therefore means should not be siloed under ends. more

Single indicator list error

Using just a 'single list' of indicators for program/organizational accountability is a mistake. Controllable indicators needs to be clearly distinguished from not-necessarily controllable indicators to avoid confusion in such lists. more

Not-necessarily controllable indicator not attributable

Just measuring a not-necessarily controllable indicator does not prove that it has changed due to a particular program or organization. more

Representing outcomes models in stakeholder discussions - 'simple but not simplistic'

Outcomes models include complex causal links and feedback loops. However at the current time they need to be represented in a 'simple but not simplisitic' form so that they do not alienate stakeholders as they slowly get used to working in a visual mode. more

Impact evaluation and controllable indicators not reaching to top of outcomes model

If controllable indicators do not reach to the top of an outcomes model, the only way to establish changes in high-level outcomes to a particular program or organization is to use impact evaluation. more

Equality of input, equality of outcome (Level playing field')

To make a fair comparison between the outcomes of different units or parties, they either need to be working with a similar level of input, or their raw outputs needs to be adjusted for any differences in input. more

The more an indicator is used for incentivization the less accurate it will be

Where indicator results can be distorted by units or parties, the more they are used to determine the level of incentives that will be provided, the less accurate they will be. more

The more collaboration the less individual attribution and accountability

There is a trade-off between attribution and accountability versus encouraging collaboration between units or parties. In general, the more collaboration that is being sought, the less one will be able to attribute high-level outcomes to a single unit or party and to hold them to account for high-level outcomes. more

Attribution and accountability only for outcomes achievable within accountability timeframes

Final outcomes can take time to occur. Where accountability is determined within periods shorter than the lenght of time it takes for an outcome to occur, units or parties should not be held accountable for high-level outcomes. more

Decisions not vindicated by their outcomes (When success is not enough)

Decisions in risk management situations should be regarded as 'right' or 'wrong' on the basis of the information that a decision-maker knows at the time. The fact that there is a positive outcome following a decision (it's 'successful') does not serve to vindicate a decision that appeared 'wrong' on the basis of information available at the time it was made. more

All levels of a program's outcomes model should be visible

Programs consist of things that are done (e.g. activities and outputs) and higher-level outcomes that are being sought. It is essential that programs show all levels of their outcomes model, including the boxes linking activities and outputs with higher-level outcomes. They should not just show the lower and higher levels and presume that stakeholders or others will be able to fill in the gap. more

Outcomes should be determined by the outside world not current organizational structure (Real-world outcomes)

Sometimes arbitary limits are put on the number of outcomes that need to be identified for a program, group or organizaiton. In the first instance, the number of outcomes should be determined by what it is hoped will be improved in the outside world, not by current organizational structure.

Always use a visual outcomes model when building an indicator set

Whenever building an indicator set (indicator framework, outcomes set, KPI list, metrics) build an outcomes (strategy) model. This lets you clarify the outcomes and steps leading to them that underpin the indicator set. more

Funder with sole provider - assist plus assess mode

Where a funder only has a single feasible provider, it can not just assess the provider and reduce funding for underperformance. It also needs to provide an assist mode. more

Overly 'tidy' indicator sets do not represent the reality under them

There is a natural tendency to produce 'tidy' indicator sets. For example, five outcome areas with five indicators under each. There is no theoretical reason to believe things would work out this way and therefore the indicators are often at different levels within the relevant outcomes (strategy) model. more