Baseline Design

When to Use

Baseline design is the right approach when you need to establish a clear starting point for measuring change. Use it when:

• Setting indicator targets: Design teams often lack adequate information to confidently propose specific targets during proposal writing. Set provisional targets from the available evidence, label assumptions, and review them with baseline findings through an agreed change process.
• Measuring change: You need a comparable pre-intervention reference. A before-after difference alone does not establish that the intervention caused the change.
• Informing project adaptation: You want baseline data that directly informs critical decisions during implementation, not just donor reporting.
• Documenting agreed measurement needs: Check baseline deliverables and timing in the applicable funding agreement or evaluation design.
• Any project duration: When initial conditions matter, measure before participants receive the intervention and account for seasonality. A short intervention may need baseline data just as much as a long one.

Baseline design is less useful when:

• The design does not need pre-intervention measurement: A valid post-test randomized comparison may estimate a treatment effect without a baseline, though baseline data can improve precision and implementation understanding.
• You only need routine monitoring: For ongoing activity tracking, routine monitoring data collection follows a different workflow.
• Rapid emergency response: For immediate emergency response, a rapid needs assessment with a different, faster methodology is required.

How It Works

Baseline design follows several core principles that distinguish it from other data collection activities:

Decision-utility first. Every data point collected must serve a specific, identified decision. If you cannot articulate which specific project decision each indicator will inform, question its necessity. This principle prevents the common dysfunction of data-rich but insight-poor baselines that collect excessive, non-essential data.

Count the burden. Collecting data is not free. Every indicator added to a baseline survey represents costs in terms of time, money, and human resources. Assess the burden (time, cost, complexity, participant fatigue) of collecting each data point. If the burden outweighs the value derived from informing a critical decision, challenge or discard the indicator. This is a core application of the burden-consciousness principle.

Trace to structure. Ensure data collection aligns with project logic, defined indicators, and any relevant structural requirements (e.g., donor modules). The baseline should support subsequent analysis and reporting, not create data that is difficult to code and analyze.

Comparability is key. Design baseline and endline methods to be as identical as possible to enable valid comparisons. Document any necessary deviations. If methods change, assess whether harmonization or a bridging study can preserve comparability; disclose residual limitations.

Baseline and endline are not evaluations. Baseline and endline studies are not evaluations themselves, but an important part of assessing change. They usually contribute to project/program evaluation (e.g., a final or impact evaluation), but can also contribute to monitoring changes on longer-term projects/programs.

Key Components

A well-constructed baseline design includes these essential elements:

• Decision identification: A documented, prioritized list of key project decisions that the baseline data will inform. This moves beyond broad categories to precise questions about what information is needed for each decision.
• Indicator selection: A lean, documented list of essential, decision-linked indicators. Each selected indicator should be directly linked to a prioritized decision, with the burden of collecting data for each indicator demonstrably justified by its decision-relevance.
• Data collection methodology: The most appropriate and cost-effective methods for collecting each data point (e.g., household surveys, key informant interviews, focus group discussions, direct observation, secondary data review). Methods should be clear, concise, unambiguous, and culturally sensitive.
• Sampling strategy: A representative sampling frame with appropriate sample size calculations, respondent selection procedures, and geographic coverage. For indicators requiring statistically sound data, ensure representative sampling and confidence intervals.
• Data collection tools: Draft tools (questionnaires, interview guides, checklists) that are clear, concise, easy to administer, and minimally burdensome for respondents and enumerators.
• Data management protocols: How data will be collected, entered, cleaned, stored, and secured. Specify data formats, validation rules, and quality checks.
• Quality assurance plan: Procedures for enumerator training, supervision, data verification, and spot-checking to ensure data accuracy and reliability.
• Informed consent procedures: Clear, simple, and culturally appropriate informed consent forms and verbal consent scripts that explain the purpose of data collection, how data will be used, confidentiality measures, potential risks and benefits, and the participant's right to refuse or withdraw at any time without penalty.
• Ethics and authorization: Determine the applicable review requirements and obtain required authorization before collection. Do not treat this guide as ethics approval.

Best Practices

Collect baseline data at the right time. Plan pre-intervention measurement before exposure, including seasonality and the design's comparison requirements. If collection begins after exposure, label the date and limitations honestly. Valid existing pre-intervention records may supply a reference; do not relabel an exposed measurement as an untouched baseline.

Link every indicator to a decision. For each proposed baseline indicator, ask: "If this data shows X, what decision will we make differently?" If you cannot answer this, the indicator is likely a candidate for removal or revision. This ensures the baseline serves as a dynamic tool for adaptation, not just a historical record.

Specify baselines where needed. Change indicators need a justified reference value or reconstruction. Some counts or post-test randomized estimates do not require a measured baseline. Mark missing references unknown rather than zero.

Incorporate baseline plans in project documents. Record the required reference measurements, timing, resources, and responsibilities in the project's measurement plan and tracking table.

Refine provisional targets with baseline evidence. Design teams will often not have adequate information to confidently propose specific targets. Set provisional targets from the available evidence, label assumptions, and review them with baseline findings through an agreed change process. Upon completion of a project baseline assessment, project decision makers set indicator targets. These targets should be informed by baseline results, the project timeline, human and financial resources dedicated to the project, and the permissiveness or difficulty of the context, including levels of uncertainty.

Select appropriate data collection methods. Select baseline data collection methods that align with and support the overall evaluation design, are appropriate for the given indicators, and meet any applicable requirements documented in the commissioning brief or agreement. Methods should be the most efficient and effective for gathering the required data while minimizing respondent and enumerator burden.

Document baseline values properly. Once the baseline value for an indicator has been determined, record the value and date collected in the 'baseline value' column of the Indicator Plan. Projects should document the baseline values in the Indicator Performance Tracking Table, or IPTT, or donor equivalent.

Measure initial conditions before exposure. Record the measurement date and intervention exposure for the population concerned. Baseline data are helpful for many evaluation designs, but are not a universal prerequisite for credible impact evaluation.

Common Mistakes

Collecting data without decision utility. The most common failure is designing a baseline that collects data solely for donor reporting without a clear plan for how it will inform decisions. When every data point cannot be linked to a specific project decision, you've created a burden of excess that wastes resources.

Using different methodologies at baseline and endline. Changes in methods, tools, or definitions require a comparability assessment and may prevent valid change estimation. Design both time points to be as identical as possible to enable valid comparisons. Document any necessary deviations.

Skipping pilot testing. Rushing to deploy without pilot testing leads to major, costly errors in the field. A pilot test should systematically identify issues with tool clarity, flow, length, and cultural appropriateness before full deployment.

Underestimating ethical review timelines. Ethical review timelines can vary significantly and may require proactive follow-up. Underestimating the time required for review and approval leads to project schedule slippage.

Mislabeling post-exposure data as baseline. Record when each population was exposed. Existing pre-intervention records may support a reference if definitions and coverage are suitable; retrospective reconstruction needs evidence and explicit uncertainty.

Not documenting baseline values. Once baseline values are determined, they must be recorded in the Indicator Plan. Without proper documentation, you cannot track progress against targets or demonstrate change to donors.

Treating baseline as an evaluation. Baseline and endline studies are not evaluations themselves. They are important parts of assessing change and contribute to project/program evaluation, but they do not constitute a full evaluation.

Examples

These scenarios are illustrative and do not describe verified program findings.

Agricultural Livelihoods - East Africa

A 5-year agricultural resilience program in Kenya and Uganda developed a baseline design that explicitly linked each indicator to a critical project decision. The team identified five key decisions: which crops to promote for market linkage, what training types farmers need, what market access barriers exist, which farmer groups are most receptive to new techniques, and how to allocate resources. For each decision, they collected only the data needed to inform it. When mid-term monitoring revealed land tenure was the binding constraint (not seed availability as expected), the program redirected resources accordingly. The baseline functioned as a diagnostic tool, not just a design artefact.

WASH - South Asia

In a hypothetical WASH program, comparable baseline and endline surveys describe changes in health outcomes. They do not identify separate effects of behavior communication and infrastructure. A claim that one pathway explains 60% and another 40% would require a credible identification design, measurements of exposure and confounders, and assumptions about interactions. No such empirical decomposition is established here.

Protection - West Africa

A protection program in Sierra Leone initially planned a full baseline with 80 indicators. Using a decision-first approach, the team mapped each indicator to a project decision and found that only 25 indicators directly informed critical choices. They collected the remaining 55 indicators as "nice-to-know" data through secondary sources where possible, reducing survey time from 90 minutes to 45 minutes and freeing resources for deeper analysis of essential data.

Compared To

Baseline design is one of several approaches to establishing initial conditions. The key differences:

Relevant Indicators

18 indicators across 4 major donor frameworks (USAID, CRS, FCDO, ECHO) relate to baseline design and use:

• Baseline documentation: "Proportion of indicators with documented baseline values collected before project start" (USAID)
• Decision linkage: "Percentage of baseline data collection linked to specific project decisions" (FCDO)
• Baseline timing: "Baseline assessment completed within 6 months of project start for projects 18+ months" (CRS)
• IPTT documentation: "Baseline values documented in Indicator Performance Tracking Table (IPTT)" (USAID)

Method Source

World Bank and IDB, Impact Evaluation in Practice, second edition (2016), chapters 4 and 15. This is methodological guidance, not a funding requirement.