The logical framework and SMART indicators: how to write indicators that work

A logical framework, or logframe, is a one-page picture of a project: what you will do, what should change because of it, how you will know, and what you are assuming. Written well, it is the document your whole MEAL system hangs from. Written badly, it is a table filled in for the donor and never opened again.

What a logframe contains

Most logframes are a grid. The rows are the levels of the project's logic, from the big change at the top down to the daily work at the bottom:

• Impact (or goal): the long-term change you want to contribute to. You rarely achieve it alone.
• Outcomes: the changes in behaviour, practice or condition that the project directly aims to bring about.
• Outputs: the goods and services the project delivers, such as trained people or a rehabilitated facility.
• Activities: the work that produces the outputs.

The columns say how to read each row:

• Narrative summary: what is expected at this level.
• Indicators: how you will measure it.
• Means of verification: where the evidence will come from.
• Assumptions: what must hold true outside your control for one level to lead to the next.

The logic reads upwards: if we carry out these activities, then these outputs follow; if the outputs are delivered and the assumptions hold, then the outcomes follow, and so on.

Outputs, outcomes and impact are different things

This is where most weak frameworks go wrong. Imagine a training project.

• Number of people trained is an output: it counts what you delivered.
• Trainees who now apply the practice at work is an outcome: it shows something changed because of the delivery.
• Better household income in the district is an impact: it needs time and many other factors.

If every indicator in your framework is an output, you will be able to prove you were busy, but not that anything improved.

What makes an indicator good

An indicator is something you measure repeatedly, against a target, to see whether the project is on course. The usual test is SMART:

• Specific: it says exactly what is counted and for whom. "Women aged 18 to 35 who completed all sessions", not "participants reached".
• Measurable: you can actually collect the number, with a method you can name.
• Achievable: the target is realistic for your resources and time.
• Relevant: it tells you something about the level it belongs to.
• Time-bound: it has a date by which the target should be reached.

Three more parts turn a good sentence into a working indicator:

• Baseline: where you start, and when you measured it. Without it, a later number proves nothing.
• Target: where you want to be, and by when.
• Direction: whether success means the number goes up, goes down, or stays the same. A dropout rate that should fall is a decrease indicator.

A worked example

Weak: "Improve women's participation."

Better: "Percentage of women among members attending at least three of five village committee meetings, measured quarterly. Baseline 18% in March; target 35% by December."

The second version says what is counted, who is counted, how often, where it starts, where it should end, and by when. (The percentages are only an illustration.) It also makes clear what evidence you need: attendance sheets, which become the means of verification.

Practical advice

• Fewer, better indicators. Two or three per outcome is usually enough. Each one costs time to collect.
• Include at least one outcome indicator. Otherwise you will only show activity.
• Decide how to break results down early. If a donor will ask for results by gender, age or location, plan to collect them that way from the start; it is hard to add later.
• Say who collects, how and how often. An indicator nobody is responsible for is not measured.
• Write the assumptions honestly. They tell you what to watch for when results lag.

Turning the framework into something you use

In Pulseform each indicator records a baseline, a target, a direction, a unit of measure, a data source, a frequency, and the breakdowns you want (gender, age or location). Values are entered with their collection dates, and the indicator's status compares your latest value with the target. See indicators and monitoring plans and data collections in the guide, and the glossary for the terms used here.