The client asks for a training-effectiveness report. The supplier has attendance, a satisfaction form and a few positive comments. The easiest response is to assemble them into a success story.
That report may be accurate as far as it goes. It becomes misleading when completion, satisfaction, skill and workplace change are treated as the same evidence. A training project can have full attendance and still leave the target workflow unchanged. It can also produce a strong learner artifact without proving revenue, productivity or organisational impact.
For an overseas expert delivering training in China, the safer approach is to keep four evidence layers separate.
The public documents show why layers matter
A Ministry of Housing and Urban-Rural Development training award describes a specified project that includes a learning-effect survey and a training-quality evaluation report. An Xiamen SME digital-transformation project document separates research, a diagnosis report, process verification and an acceptance report.
These sources do not establish a universal reporting standard. They do show that a project may name different evidence and delivery stages rather than treating “training completed” as the end of every question.
The documents cannot prove that a training programme changed workplace performance. They cannot authorize a public case study. They cannot be used to claim that an overseas expert has a normal China reporting template or that the buyer’s budget represents a market price.
Four layers, four different claims
Layer one: completion
Who attended, what was delivered, and which sessions or materials were made available? Attendance, access logs and delivery receipts belong here. They can establish that a defined activity occurred.
They cannot establish that participants understood the content or used it later.
Layer two: participant feedback
What did participants report about clarity, relevance, confidence or perceived usefulness? Feedback can identify confusion and improve the next session. It can be collected immediately or after a defined interval.
Feedback is self-report. A positive comment is not a skill assessment and not a business result. A negative comment is not automatically evidence that the curriculum failed; context and task conditions matter.
Layer three: demonstrated skill
Can a participant complete a defined task under stated conditions? A submitted workflow comparison, annotated exercise, diagnosis worksheet or reviewed output may be evidence here. The task and rubric must be named.
Demonstrated skill is stronger than attendance, but it is still bounded. A learner who completes one supplied exercise has not necessarily changed an organisation’s routine.
Layer four: workplace or business change
Did a defined work process change after the training, under what conditions, and what evidence supports the claim? This may require a separate operational measurement, a baseline, a time window, an owner and permission to inspect the relevant workflow.
This layer is the most likely to be overclaimed. It cannot be inferred from a smile sheet, a completion badge or a single success anecdote.
A scene: the report that jumped two layers
An AI training supplier delivers a workshop. Ninety percent of attendees complete the feedback form and several write that the examples were useful. The supplier writes: “The programme improved AI adoption across the organisation.”
The evidence supports completion and participant feedback. It may support a limited statement about perceived usefulness. It says nothing about adoption unless the organisation defined adoption and collected evidence after the workshop.
The scenario is constructed, not a reported client case. The correction is simple: write the report in layers and label the missing evidence.
The Xiamen project is not a universal evaluation template
The Xiamen SME digital-transformation document is useful because it distinguishes research, diagnosis, process verification and acceptance. That sequence can help an expert ask whether a training project has a defined before-and-after work question.
It does not mean every training project needs the same reports. It does not mean a public project’s acceptance report proves a course changed a company. It does not establish an overseas supplier’s qualification, a private buyer’s demand or a standard commercial process.
Use the document as a source for separating evidence stages, not as proof of a market or a case study.
What to write in the client-facing report
Start each section with a verb that matches the evidence:
- Delivered: the agreed sessions, files or access were provided.
- Reported: participants described relevance, clarity or confidence.
- Demonstrated: a defined learner task was completed and reviewed.
- Observed later: a defined workflow or metric changed under stated conditions.
Avoid using “proved” when the evidence only shows “reported”. Avoid “transformed” when the record shows “attended”. Avoid “case study” when the participant or buyer has not approved public reuse.
If a layer is missing, say so. An honest report can close a project while leaving workplace impact as unmeasured.
The case-study boundary
Even if a project has data, public use is a separate decision. A training report may contain participant names, comments, internal work examples or buyer information. It should not become a website case study without permission and a clear attribution or anonymisation boundary.
No source in this article grants such permission. Public procurement documents are not invitations to reuse private project facts. Keep the evidence record and the public-marketing decision separate.
A counterexample: when completion is the intended result
Some engagements only promise a seminar, orientation or delivered set of materials. In that case, completion may be the relevant acceptance evidence. The report should describe that limited promise accurately rather than inventing a skills or business outcome.
Likewise, a project may explicitly test one task and nothing more. Then a reviewed task artifact can be meaningful evidence at layer three. It still should not be expanded into an organisational result without layer-four evidence.
The four-layer report template
Before sending the report, ask:
- What was delivered and completed?
- What did participants report?
- What task did they demonstrate, and how was it reviewed?
- What workplace change was actually measured, by whom and when?
Add a fifth line: what cannot be published or generalized without permission and further evidence.
The sentence worth carrying forward is: a completion record closes an activity; it does not close the question of changed work.
If you need to design an evidence-bound training report for a China-facing engagement, request a China validation review. Bring the delivery promise, learner artifact and any approved measurement plan separately.
Related reading
- Product Portability Audit
- AI Agent Workshops in China Need Workflow Evaluation, Not Tool Demos
- When Topic Popularity Is Not Buyer Demand
- A China Training Procurement Brief Is a Service Scope, Not a Course
Sources and evidence boundary
- Ministry of Housing and Urban-Rural Development technical capability training award, checked 2026-08-23. Shows learning-effect survey and quality-evaluation report items in one specified project; does not establish workplace impact or general reporting standards.
- Xiamen SME digital-transformation pilot project document, checked 2026-08-23. Shows separate research, diagnosis, process verification and acceptance stages; does not establish training outcomes, demand or public case permission.
Image brief: Four stacked evidence cards: completion, participant feedback, skill artifact and workplace change, with a clear gap between each layer. No satisfaction scores, client logos or success arrows. Alt text: “Four evidence layers separate training completion from skills and workplace change.”