M&E Studio

Status: Authored worked illustration using the fictional source pack at worked-project-input.md. It is not output from a skill run, generated test, client example, or practitioner trial. This guide shows how three M&E jobs can use the same supplied evidence without turning plans into findings or suggestions into approvals.

The complete illustrative artifacts provide the full questions and matrix, TOR, review/change record, indicator sheet, workload plan, coherence check, evidence report, revision and proposed actions. The excerpts in this guide orient the reader to that bundle.

The project in one view

Hillside Kitchen Gardens ran a five-session course at two community gardens during January-March 2026. The director will decide whether to repeat the course format and whether to consider follow-up visits. The supplied target is 32 of 40 enrolled participants completing at least four sessions. The attendance register records 31. A feedback form reached 24 participants; 18 returned it, with 14 Yes and 4 No responses. The pack contains one period only and no measure of later practice use.

The first job is to define useful questions and methods. The second is to make an operational measurement plan that fits stated capacity. The third is to report only what the records support and propose decisions without assigning them.

Start with the supplied files

Use the fictional source pack as the whole evidence set. Its IDs, HP-01 through HP-07, anchor the examples below. Keep unprovided information marked Not supplied, Unconfirmed, or Decision required. Do not fill a gap with a plausible value.

The canonical methods are maintained in their individual skill files. This guide links to them rather than copying their full procedures:

Job	Method and workflow entry	Output
Frame evaluation decisions	evaluation-questions, question-method-alignment, evaluation-tor-draft, evaluation workflow	In order: evaluation-questions, question-method-alignment, evaluation-tor-draft, then the review skill `me-review:tor-review`. Files: questions.md, matrix.md, tor-draft.md, review.md, tor-revised.md, change-record.md
Plan practical measurement	measurement-feasibility, indicator-reference-sheets, me-plan-draft, measurement workflow	In order: the review skill `me-review:indicator-quality`, then measurement-feasibility, indicator-reference-sheets, me-plan-draft, then the review skill `me-review:me-plan-review` for the coherence check. Files: indicator-review.md, burden.md, reference-sheets.md, me-plan.md, coherence.md, decisions.md
Check evidence and draft action	progress-evidence, findings-to-actions, progress-report-review, reporting workflow	In order: progress-evidence, `me-review:progress-report-review`, then findings-to-actions. Files: evidence.md, report-draft.md (the supplied draft report, saved unchanged), review.md, report-revised.md, actions.md, change-record.md

Each stage below is a summary. For the full steps, open that workflow's START.md and use its starter prompt; it sets the stage order, the file names and the checks. Save each stage in its own folder (evaluation/, measurement/, reporting/), because some file names, such as review.md and change-record.md, repeat across stages. You may combine stage files into fewer files if each stage stays clearly headed and the change record says where each stage went.

1. Evaluation planning

Select questions for the decision

HP-01 says the director will decide whether to repeat the course format and consider follow-up visits. HP-02 and HP-03 support a question about attendance and course completion. HP-04 supports a limited description of the feedback respondents. No supplied record measures whether participants later use the practices.

Priority	Question	Evidence and method	Boundary
1	How many of the 40 enrolled participants met the approved completion definition, and how does that compare with the cohort target?	Recount the attendance register using the approved four-of-five-session definition; compare with the 32-person target. HP-02, HP-03.	This establishes a register-based count, not whether the course caused later practice use. Confirm any discrepancies before reporting.
2	How did the participants who returned the feedback form answer whether the demonstration was useful?	Summarize the 18 returned forms and report the six nonreturns separately. HP-04.	The selection reason for inviting 24 is unknown. Do not generalize respondents' answers to all 40 participants.
Deferred	Did participants use the garden practices after the course?	No approved definition or follow-up data are supplied. HP-04, HP-07.	Decide whether to define a follow-up measure and obtain contact/access authorization. Do not add this as a finding or imply a causal design.

This question set is a proposed way to inform the decision. It is not approval of the questions or of follow-up visits.

Map questions to evidence before choosing methods

A compact matrix keeps a question from being mistaken for proof that a source can answer it.

Question	Supplied source	Method	What it can support	What remains unknown
Completion and target	Attendance register; approved definition and target	Recount participants with at least four recorded sessions	A cohort count and comparison with the end-of-period target	Whether the register is complete and whether any ID/session entry needs correction
Usefulness among respondents	18 returned forms from 24 invitations	Tabulate Yes/No among returned forms; record nonresponse	Respondent-only answers and the response count	Why six did not reply, how the 24 were selected, and whether answers represent all participants
Later practice use	None supplied	No current method can answer this	Only that it is not measured in this evidence pack	Definition, observation window, permission, source, and any feasible follow-up design

Draft the TOR without filling unknown sections

A TOR can be useful before every detail is known. It should show the decisions still needed, not make them disappear. The excerpt below is abbreviated: the evaluation-tor-draft skill returns all ten TOR sections, with placeholders where the source is silent.

Purpose and use: Provide the director with evidence on course delivery and participant feedback to inform whether to repeat the current format. Whether to add follow-up visits remains a decision for the director.

Scope: January-March 2026 cohort at the two community gardens. Participants are the 40 enrolled adults. The completion measure uses the supplied four-of-five-session definition.

Methods and evidence: Reconcile the attendance register to the approved definition and target. Summarize the 18 returned feedback forms, retaining the 6 nonreturns and the unknown invitation-selection process. No later-practice or causal claim is supported by the supplied records.

To be confirmed: Review date, decision owner beyond the named director, any additional fieldwork, permissions, budget, and whether a later-use measure should be designed.

If the scope, timing, team effort, budget, or access is not supplied, keep it open. A review of this draft can flag gaps, but cannot grant authority or access.

2. Feasible measurement

Preserve the approved measure and separate proposed measures

Measure	Status and supplied definition	Source/frequency	Planning note
Course completion	Approved for this cohort: enrolled participant attended at least 4 of 5 sessions; target 32 of 40 by March 31	Attendance register; sessions recorded during the 12 scheduled course weeks	Keep the definition and target. Reconcile the 31 reported completers against the register before final reporting.
Demonstration usefulness	A feedback question, not an approved outcome indicator: “Did you find the garden-practice demonstration useful?”	24 invitations after the final session; 18 returned forms	Report the respondent count and nonresponse. Do not create a target or claim representativeness.
Later practice use	Not measured; no definition or authorization supplied	No source supplied	Any follow-up is a proposal requiring an approved definition, timing, access/consent decision, and capacity check.

Check workload in matching units

Use supplied volumes and time-per-task only. For the two facilitators, weekly attendance work is:

Actual estimate: 2 facilitators × 15 minutes/week × 12 weeks = 360 minutes = 6 hours for the quarter.
Available capacity: 2 facilitators × 30 minutes/week × 12 weeks = 720 minutes = 12 hours for the quarter.
Each facilitator uses an estimated 15 of 30 available minutes per week. This is a planning comparison, not a finding that all attendance work is complete or that extra work is approved.

For the monitoring officer, compare each month with the stated 90-minute monthly capacity:

Month	Supplied task estimate	Capacity	Remaining capacity
January	20 minutes to consolidate logs	90 minutes	70 minutes
February	20 minutes to consolidate logs	90 minutes	70 minutes
March	20 consolidation + 45 reconciliation/reporting = 65 minutes	90 minutes	25 minutes
Quarter	105 minutes (1 hour 45 minutes)	270 minutes (4 hours 30 minutes)	165 minutes (2 hours 45 minutes)

The monthly comparison matters: March has less remaining time than January or February. These figures do not include a new follow-up visit, contact attempts, travel, tool design, or additional analysis. The source pack supplies no time or budget for those tasks, so their feasibility is unresolved.

Keep the plan operational

An M&E plan should connect the indicator ID and definition to the source, population, period, frequency, role, checks, reporting, and decision use. For this example:

Continue the attendance register as the supplied source for the completion measure.
Reconcile participant IDs and session records before finalizing the 31-person count.
Keep the feedback summary respondent-only and display the six nonreturns.
Do not add a later-use indicator until its definition, observation window, access, and approval are decided.
Keep any extra follow-up as a proposal; do not charge it to the available staff capacity without a time estimate and authorization.

A coherence review should compare the same definitions and periods across the indicator sheet, plan, evidence report and any proposed action. A changed definition needs a visible decision and a comparability note.

3. Reporting to action

Reconcile before interpreting

The completion calculation is:

Reported: 31 / 40 = 77.5% of enrolled participants.
End-of-period target: 32 / 40 = 80%.
Difference: 31 - 32 = -1 participant, or 77.5% - 80% = -2.5 percentage points.

The report therefore shows one fewer completer than the supplied target. This count must still be checked against the attendance register because HP-03 says no discrepancy review is recorded.

For feedback:

Response rate: 18 / 24 = 75%; six invited participants did not return a form.
Among respondents, 14 / 18 = 77.8% answered Yes and 4 / 18 = 22.2% answered No.
Report those as respondent-only shares. Do not treat the six unknown responses as No, and do not claim that 77.8% of all 40 participants found the demonstration useful.

Refuse the unsupported trend claim

The draft says completion is “increasing steadily and is on track.” Only one cohort and one period are supplied. There is no prior value for a trend and no time-phased expectation for an on-track judgment. The end-of-period target supports an attainment comparison, not a trajectory claim. Mark trend as not assessable from supplied records and remove “increasing steadily” and “on track.”

Example of a bounded report revision

Course completion: The register summary reports that 31 of 40 enrolled participants attended at least four of five sessions (77.5%). The supplied end-of-period target is 32 of 40 (80%), so the reported count is one participant below target. No reconciliation record is supplied, so confirm the count before finalizing.

Demonstration feedback: Eighteen of 24 invited participants returned a form. Fourteen respondents answered Yes and four answered No to whether the demonstration was useful. Among respondents, 14/18 (77.8%) answered Yes. Six responses are missing, and the invitation-selection reason is not recorded; these results are not an estimate for all 40 participants.

Trend and later use: A trend and “on track” status cannot be assessed from this single period. No later-practice-use data were supplied.

Keep proposed actions separate from management decisions

Supported issue	Proposed next step	Status/owner/date
Completion-count reconciliation is not documented in the supplied records.	Recount the 40 attendance records using the approved four-of-five-session rule, then confirm the report value.	Proposal; owner and date not supplied.
Six feedback forms were not returned; invitation selection is undocumented.	Record the response denominator and clarify the invitation method before making broader claims.	Proposal; no approval or assigned owner supplied.
Later practice use is not measured, and follow-up permission is not supplied.	Decide whether to define a later-use measure and whether contact is authorized; estimate workload before adding it to the next cohort plan.	Decision required; follow-up approval is not supplied; status is unconfirmed.

Do not state that management accepted these actions or assign an owner or deadline without source evidence.

Resume from saved work

Keep the original source pack with each output. A handover should identify completed files, evidence IDs, changed inputs, decisions, and exact next step. For this example, a safe handover would say:

Source: attachments/worked-project-input.md, HP-01 through HP-07.
Completed: draft questions, matrix, full draft TOR, workload plan, evidence report, and proposed action table.
Confirmed: target definition and target are approved for this cohort; the register summary reports 31/40; feedback source reports 14 Yes, 4 No, 6 nonreturns.
Unconfirmed: register reconciliation; representativeness of the 24 invitees; later-use definition, permissions, approval, owner and date.
Corrected claim: “Increasing steadily” and “on track” were removed from the draft report because the pack provides no prior period or time-phased expectation. This is not evidence that completion did or did not increase.
Next step: supply any updated register, approval, or definition; compare changed facts with the original source and record what changed. Do not treat a prior proposal as an approved action.

If a new session receives only the handover and not the source pack or saved artifacts, it should ask for them rather than assume access to an earlier folder or chat.

Quality checkpoints

Before using a real project, check that:

Each factual statement has an evidence source and period.
The denominator, unit and reporting basis match the calculation.
Missing or conflicting information stays visible.
A respondent-only percentage is not described as a whole-population estimate.
One period is not called a trend; target attainment is separate from trajectory.
A plan, proposal, recommendation and approved commitment remain distinct.
Privacy, contact, site access and fieldwork permissions are not assumed.
Saved artifacts and handover are supplied when work resumes.

What This Shows and What It Does Not

This is a hand-written illustration with a fictional program, not the output of a skill run or a practitioner trial. It does not show that every route works for every account. Claude Code is the route that has been checked. One skill (evaluation questions) was also checked by browser upload on a paid plan. The Anthropic help page says Free and paid plans can use uploaded skills once code execution is turned on; we did not test a Free account. Practitioner feedback has not been gathered yet. These are AI-assisted drafts: a person who knows the program should check every output before it is shared.

Source: M&E Studio, monitoringevaluationstudio.com. Guide and fictional attachment use is governed by the existing free-content terms; public plugin files retain their separate MIT license.

See usage and credit for how to credit M&E Studio.

A Worked Project