Developmental Evaluation

What Is Developmental Evaluation?

Developmental evaluation is an approach for programs that are still being invented rather than delivered, where the evaluator works inside the team and feeds findings back fast enough to change what happens next. Developed by Michael Quinn Patton, it occupies the space between formative evaluation, which assumes a reasonably stable program, and summative evaluation, which requires a finished one.

What separates it from other approaches is the evaluator's position and the purpose of the evidence. The evaluator is embedded rather than independent, the reporting cycle is weeks rather than years, and the question is not "did this work?" but "what are we learning, and what should we change now?" That makes it powerful for genuine innovation and inappropriate for accountability judgments, which is the distinction most teams get wrong.

When Should You Use Developmental Evaluation?

Developmental evaluation (DE) is the right approach when the program itself is still being designed, when goals and strategies are genuinely emergent, and when adapting in real-time is more important than proving what has already been built. Developed by Michael Quinn Patton, DE was designed for the space between formative evaluation (which assumes a relatively stable program) and summative evaluation (which requires a completed one).

Use it when:

• The program is a social innovation: the team is genuinely discovering what works through iteration; there is no established model to test or scale
• The environment is highly complex: political shifts, market changes, or emergent social dynamics mean the program must adapt continuously to remain relevant
• Goals are evolving: the program's theory of change is being built and tested during implementation, not applied from a design document
• Real-time feedback is more valuable than a final report: the program team needs evaluative thinking embedded in their work, not an external verdict at the end
• The learning need fits an embedded approach: confirm the intended use, evaluator role, independence safeguards, resources and any applicable commissioning commitments

DE is not appropriate for established programs with a stable theory of change, for evaluations where donor accountability requires a defined before/after comparison, or for situations where the evaluator's independence must be maintained entirely (the embedded nature of DE creates role boundary challenges).

How Does Developmental Evaluation Work?

Unlike conventional evaluation, which operates on a program-then-evaluate sequence, developmental evaluation is simultaneous with program development. The evaluator is a thinking partner, not an external observer.

Step 1: Establish the evaluator's role and boundaries

The DE evaluator is embedded in the program team - attending team meetings, contributing to strategy discussions, and providing real-time evaluative feedback. Boundaries must be explicitly negotiated: the evaluator maintains intellectual independence and evaluative perspective even while working alongside the team.

Step 2: Support theory development

In complex programs, the theory of change is itself emergent. The evaluator's first contribution is often helping the team make their implicit theory explicit and testable.

Step 3: Design real-time monitoring

Identify the critical uncertainties in the emerging theory and design lightweight, rapid data collection processes that can inform decisions in weeks, not months. This is not a full M&E system - it is targeted data collection to answer the specific questions the program team is grappling with now.

Step 4: Provide ongoing evaluative feedback

The primary output of DE is not a report - it is a continuous flow of evaluative thinking fed into program decisions. This might be a brief memo after a stakeholder consultation, a pattern analysis from field visit observations, or a synthesis of early outcome signals.

Step 5: Document learning and theory refinement

Over time, the DE process generates an evolving record of what has been tried, what was learned, and how the program theory has changed. This documentation becomes the basis for later summative evaluation and for sharing learning with the broader field.

Key Components

• Embedded evaluator: a professional evaluator working as a learning partner within the program team, not as an external reviewer
• Real-time feedback processes: lightweight data collection and synthesis methods that produce findings in days or weeks
• Living theory of change: an explicitly documented and regularly updated program theory that captures what has been learned
• Innovation documentation: systematic recording of what is being tried, adapted, and abandoned, and why
• Role clarity: explicit agreement between the evaluator, program team, and funder about what DE is and is not
• Developmental evaluation report: periodic documentation of learning and theory development (not a traditional evaluation report)
• Integration with adaptive management: DE findings must be connected to decision-making processes, not just circulated as documents

Best Practices

Clarify the DE role before you start. Developmental evaluators who are unclear about their role become either captured (they stop evaluating and just support the team) or marginalised (the team stops engaging with them). Negotiate and document the evaluator's role explicitly at program start.

Use the ToC as a working hypothesis, not a fixed frame. The program theory in DE is a hypothesis about how change will happen. Every implementation experience is an opportunity to test and refine it - not to measure performance against it.

Don't rely solely on routine monitoring. Routine data tells you what is happening, not why - and in emergent programs, the "why" is the critical question. DE requires methods that probe mechanisms, not just track indicators.

Conduct real-time evaluations during emergencies and pivots. When the program context changes dramatically - a political crisis, a funding shift, a major implementation failure - a rapid DE review can provide the evaluative thinking needed to navigate the change.

Document what was not pursued. Developmental evaluation's most underused contribution is documenting the path not taken - the hypotheses rejected, the strategies abandoned, and the reasons. This learning is often invisible in conventional evaluation reports.

Common Mistakes

Treating DE as an excuse to avoid rigour. The absence of pre-specified outcomes does not mean anything goes. DE still requires systematic data collection, transparent reasoning, and honest reporting of what is not working.

Blurring the evaluator's independence. When the evaluator becomes a de facto program staff member, they lose the evaluative distance that makes their contribution valuable. The evaluator should be able to say "this isn't working" without fear of undermining their position in the team.

Using DE language for conventional formative evaluation. DE is not just "evaluation done early." It is a specific approach for genuinely complex and emergent programs. Applying the label to a standard formative evaluation misrepresents both.

Neglecting to connect DE findings to decisions. Evaluative thinking that stays in the evaluator's notebook serves no one. Build explicit feedback loops between DE findings and program decision-making processes.

Not planning for transition to summative evaluation. DE generates valuable documentation of program development. Plan from the start how this will be used when the program reaches a point where summative evaluation is appropriate.

Examples

The following scenarios, settings, and numbers are illustrative; they are not documented program findings or standard design requirements.

Social innovation, Canada. A national foundation in Canada used developmental evaluation to support a five-year social innovation fund testing new models for youth employment in marginalised communities. The embedded DE team attended quarterly strategy meetings, conducted rapid ethnographic observations at implementation sites, and produced monthly learning briefs. Over the first 18 months, the evaluation documented seven theory revisions as grantees discovered which employer engagement strategies produced durable job placements versus short-term placements. These findings were shared across the portfolio, enabling grantees to learn from each other in near real-time.

Adaptive health program, East Africa. An adaptive health systems strengthening program in Uganda used DE to support a team working in three politically complex districts. The evaluator documented how the program theory shifted from a supply-side (training health workers) to a demand-side (community engagement) focus within 12 months as the team responded to facilities refusing to implement changes. The DE documentation informed a proposed mid-program design review. Any change to a commissioned evaluation would still need the actual approving authority and applicable agreement provisions.

Emergency response, South Asia. Following a cyclone response in Bangladesh, a major international NGO used a rapid real-time developmental evaluation to assess which coordination mechanisms were producing efficient resource allocation versus which were creating bottlenecks. The evaluation ran for six weeks alongside the response. Three coordination changes were implemented within the evaluation period based on findings, each documented and tested in real-time. The final report was completed within the response phase rather than after.

Compared To

Relevant Indicators

16 indicators across DFID, UNDP, and foundation frameworks. Key examples:

• Number of program adaptations formally documented as informed by DE findings
• Quality rating of evaluator-team engagement process (rated by both parties)
• Frequency of real-time feedback provided to program team (target: at least monthly)
• Degree of theory of change refinement documented over the evaluation period

Related Tools

• MEStudio Logic Model Builder: for developing and updating the living theory of change
• Evaluation Planner: for structuring the DE monitoring approach and real-time data collection