Summative evaluation happens at or after program completion and judges overall results. It asks "did it work, for whom, at what cost, and what was the impact?" Where formative evaluation feeds mid-course correction, summative evaluation produces the verdict.
What Summative Evaluation Asks
Summative evaluations tend to cluster around five questions:
- Were the stated objectives and targets achieved?
- What outcomes can be attributed to the program, and how much would have happened anyway?
- How did results vary across subgroups (gender, geography, vulnerability, dosage)?
- Was the program cost-effective relative to alternatives or sector benchmarks?
- What lessons are transferable to future programs, whether your own or the sector's?
Good summative evaluations answer those five cleanly. Weak ones confuse activity completion with results or restate the logframe without testing it.
When to Use Summative Evaluation
Summative evaluation is near-mandatory for any multi-year donor-funded program above a modest threshold. Specifically, it is the right fit when:
- The program is closing and accountability reporting is due to the funder and stakeholders.
- A scale-up, replication, or renewal decision depends on evidence of what worked.
- The sector is thin on evidence and a rigorous final evaluation would contribute to the public knowledge base.
- The theory of change includes outcome-level claims that cannot be judged from monitoring data alone.
For short projects under 12 months, a lighter after-action review often substitutes.
Design Features
Timing is tight. Summative evaluations run at or within three to six months of program end, with data collection timed to allow outcomes to mature without losing respondent recall or cohort traceability.
Methodologically, summative work carries full rigor: an external evaluator for programs above a certain size, a comparison or counterfactual design where feasible, mixed methods to triangulate contested claims, and sampling that supports subgroup analysis. Generalizability matters more than in formative work; the findings are expected to travel. Independence requirements are typically mandatory, with the evaluator procured separately from the implementer and reporting directly to the donor or a steering committee.
Relationship to Impact Evaluation
Summative evaluation judges overall results. Impact evaluation specifically addresses attribution: what change is caused by the program versus what would have happened anyway. Impact evaluation is a specific type of summative evaluation that uses a counterfactual design, whether a randomized controlled trial, quasi-experimental comparison, or strong contribution analysis.
Not all summative evaluations are impact evaluations. A summative without a counterfactual can still judge whether targets were met, how stakeholders experienced the program, and whether the theory of change held up, but it cannot make clean attribution claims.
Proposal Context
Virtually all multi-year bilateral and foundation-funded programs require a summative evaluation, and proposals that handle it vaguely get marked down. Specify five things:
- Timing: End-of-project or three to six months after closeout, with a clear data collection window.
- Evaluator independence: External is standard for programs above a certain size; name the procurement approach (competitive tender, pre-qualified panel) even if the firm is not yet selected.
- Methodology commitment: The final design is finalized in the inception phase, but the approach should be named now (quasi-experimental with comparison, theory-based, mixed-methods contribution analysis).
- Budget: Typically 2-4% of total program budget, higher for impact evaluations with counterfactual designs.
- Intended use: Accountability to the donor, scale-up or replication decisions, and sector learning contribution. Name the audiences.
Proposals that hit all five read as serious. Proposals that just say "a final evaluation will be conducted" do not.
Common Mistakes
- Timed too close to endline. If the summative lands after all decisions are locked, nobody uses it. Build a decision window of at least six to eight weeks between report delivery and the closeout decision point.
- No independence specification. Internal staff evaluating their own program produces predictable bias concerns from donors. State independence arrangements in the proposal, not at inception.
Related Topics
- Formative Evaluation: The mid-course counterpart to summative
- Impact Evaluation: Summative evaluation with a counterfactual design
- Evaluation - Parent concept covering all evaluation types
- Theory-Based Evaluation: Common approach to summative design
- Counterfactual: Core concept in attribution-focused summative work