Logical Framework Matrix (LSM / Logframe)

The logical framework matrix describes a project so that its logic can be checked, not merely read. A donor treats it as the summary of the whole application.

The European Commission distinguishes two things that everyday usage conflates. The LFA is an analytical process – stakeholder analysis, then problem analysis, then objectives, then strategy selection. The matrix is the document summarising the result. The guidelines put it plainly: the matrix should be thought of as "an aid to thinking".

Four rows: the levels of a project

Activities – the tasks needed to deliver the planned results. Results – the direct, tangible goods and services the project delivers, "largely under project management's control". Purpose – the development outcome at the end of the project, specifically the expected benefits to the target groups. Overall objective – the broad impact at national or sectoral level to which the project contributes.

See Outputs vs Outcomes vs Impact for the result levels. Your degree of control falls with every level.

Four columns – and the exception people forget

Intervention logic, indicators, sources of verification, assumptions.

The exception in the Activities row. In the Commission's layout that row carries neither indicators nor sources of verification – optionally it carries means and costs. The activities row is itself optional in the matrix; activities can live in a separate schedule. Textbooks that draw indicators there are wrong, not simplifying.

Below the activities row sit the pre-conditions: what must hold before resources are committed.

The order of completion is not left to right

The Commission sets a 13-step sequence: the whole first column top-down (1–4), then the assumptions bottom-up (5–7), and only then indicators and sources of verification, top-down (8–13). You cannot write an indicator for a statement you are still going to change.

The two logics an evaluator checks

Vertical, as an if–then chain: if resources are provided, activities can be undertaken; if activities are undertaken, results are produced; if results are produced, the purpose is achieved; if the purpose is achieved, it contributes to the overall objective – with "and the assumptions at this level hold" added at every junction.

Test it with necessary and sufficient, and note the asymmetry: activities should be necessary and sufficient to deliver the results, but the results are only necessary and may not be sufficient to achieve the purpose. Claiming your outputs are sufficient for the outcome means you have either overstated or mis-levelled it.

Horizontal, within one row: level – indicator – source – assumption.

Indicators: QQT, and the Commission's own SMART

The Commission's formula is QQT – quantity, quality, time.

Careful with SMART. The Commission expands it differently from most training material: Specific, Measurable, Available at an acceptable cost (not Achievable), Relevant to the information needs of managers, Time-bound. Two of the five criteria are about the cost and usefulness of measurement, not about the objective. An indicator too expensive to measure must be replaced with a proxy.

SMART is not universal. The acronym does not appear at all in USAID's ADS 201/303 or in the UNDG results-based management handbook, both of which use their own indicator checklists; UNDP reads "A" as Achievable for results and Attainable for indicators.

Indicators must be independent of one another and each attach to exactly one level; "objectively verifiable" means different people measuring would get the same result; and "the trap of including too many indicators should be avoided". A source of verification must state how the information is collected, by whom, and how often – listing "project reports" without that is named as bad practice.

Assumptions: an algorithm, not intuition

Assumptions are external factors capable of influencing or determining success but outside management's direct control. They are written as positive statements of a desired state, not as risks.

The Commission gives a decision tree. Is the assumption important? If not, leave it out. If yes – will it hold? Almost certainly: leave it out. Possibly: include it. Very unlikely: can the project be redesigned to influence the factor? If yes, add activities or results and reformulate the purpose if needed. If no, the guidelines conclude without euphemism: the project may not be feasible.

The term "killer assumption" appears nowhere in the Commission's documents or the OECD-DAC glossary. It is practitioner jargon for that last branch, not methodology.

The most common wording mistake

The Commission names it directly: the purpose is written as a restatement of the sum of the results rather than as a higher-level achievement.

Bad: "Water treatment is improved and levels of direct discharge into the river reduced".
Good: "Improved quality of river water".

Other listed failures: reducing objectives to a simplistic linear chain; setting unrealistic targets too early; using the matrix as top-down control; and preparing it "mechanistically as a bureaucratic box-filling requirement", detached from the problem analysis – at which point it "becomes a fetish rather than a help".

"4×4" is USAID's phrasing, not the Commission's

The European Commission has never called the matrix "4×4". Its guidelines say "four columns and a number of rows (usually three or four)", and a footnote actively encourages adding levels where a user has good reason. It is USAID that calls its own version "the basic 4 x 4 matrix".

The current EU template has seven columns: results chain, indicator, baseline, target, current value, source and means of verification, assumptions. Levels have been Impact – Outcome – Output since 2015, with intermediary outcomes added in 2016. Activities live in a separate Activity Matrix with means and costs.

Two details of the current template are easy to miss. At Impact level the assumptions cell reads "not applicable" – by design. And renaming Results to Outputs was substantive: that row now expects output indicators only.

Note the asymmetry: the methodology is still the 2004 edition (no published successor), while the application template is from 2019. Use the template attached to your call.

Who asks for one

EU external action requires it: the logframe is an annex to the application and then forms part of the description of the action in the contract, updated with every report.

FCDO (formerly DFID) is the strictest major donor: since January 2011 a logframe is an integral part of the Business Case for every project at any value, in Excel, with impact weightings per output summing to 100%.

Horizon Europe does not: the word does not occur once in the standard application template. It uses an impact pathway and Impact canvas in Part B instead.

USAID – the most widespread false claim. The phrase "logical framework" appears zero times in ADS 201 and ADS 303: what is mandatory is a Results Framework in the country strategy, with the logframe named only as one permitted kind of logic model. Where a specific call asked for one, that was a mission decision, not agency policy. And USAID ceased to operate as an independent agency on 1 July 2025.

UN agencies work with results frameworks under results-based management rather than the classic 4×4.

The matrix and theory of change

The matrix records what will change and how it will be verified; a theory of change explains why. The OECD-DAC glossary treats theory of change not as an alternative to the logframe but as a term of the same family, alongside intervention logic and results framework.

The familiar contrast – logframe linear and static, theory of change non-linear and explanatory – comes from academic and training literature, not from donor requirements. No side-by-side official comparison exists.