Better Method.ai
Recipe 03

The post-activity outcomes report

Turn source material into a structured outcomes-report draft while keeping every figure and claim traceable.

What you’re making

A full draft outcomes report. What the gap was, what you did about it, what changed, what didn’t, and what you’d do differently. Laid out to your own template, in your own voice, with every number traceable to something you supplied.

This workflow is useful when you already have the facts but still need to turn a spreadsheet, a set of themes and a planning document into connected prose. The draft can save assembly time, while the source trace and final judgment remain yours.

What to have open

Your objectives and gap statement from the planning stage. Your numbers: attendance, completion, pre and post scores, whatever you measured. Your themes from the open text (see recipe 02).

And one more thing that matters more than people expect: a previous report you were happy with. That’s how you get your own voice instead of a generic one.

The prompt

Do this in two goes. The first sets up the shape. The second writes it.

Turn one: get your bearings
I'm drafting a post-activity outcomes report for an accredited continuing education activity. I'll give you the material, then ask you to draft.

First, read everything and tell me:
1. What story does this data actually tell? Three sentences, including the parts that aren't flattering.
2. Where does the evidence NOT support a claim I might want to make?
3. What's missing that a reviewer would reasonably expect to see?

Do not draft the report yet.

ORIGINAL PRACTICE GAP:
[paste]

LEARNING OBJECTIVES AS WRITTEN:
[paste]

ACTIVITY DETAILS:
[format, date, length, number of faculty, learner numbers and professions]

THE NUMBERS:
[paste: registrations, attendance, completion, pre and post results, self-reported change, any commitment-to-change data]

THEMES FROM THE OPEN TEXT:
[paste your themes and counts]

Use only these figures. Do not add, estimate, or infer any number I have not given you.

Read that answer properly before you go on. Question 2, where the evidence doesn’t support the claim, is the one that saves you from writing something a reviewer will pull apart. Much easier to hear it now than in a citation letter.

Turn two: draft it
Good. Now draft the full report using this structure:

[paste your template's section headings, or use the ones below]

1. Activity overview
2. Identified practice gap and needs assessment basis
3. Learning objectives
4. Design and format, and why that format suited the gap
5. Participation
6. Outcomes measured, by level
7. What the open text feedback tells us
8. What we would change
9. Implications for the next cycle

Requirements:
- Match the tone, sentence length and formality of the sample report below. Do not make it sound more promotional than the sample.
- Every number must be one I gave you. Never round to make it read better.
- Where the data supports a claim only weakly, say so in the report. Write "self-reported intent to change, not observed practice" rather than implying we observed a change.
- Where I measured at Moore level 3 or 4, do not write the report as though it shows level 5.
- Mark anywhere I need to insert something you don't have as [NEEDS: what's required].
- Don't oversell the conclusion. If the honest finding is "modest knowledge gain, format problem unresolved", write that.

SAMPLE REPORT IN OUR HOUSE VOICE:
[paste a previous report you were happy with]

What good output looks like

Two sections tell you straight away whether it worked.

Section 6 should be careful about the difference between what you measured and what you’d like to claim. “Participants demonstrated improved knowledge (pre 62%, post 84%, n=118)” is a claim about knowledge. “Participants reported they would change their titration practice (78%)” is a claim about intent. Good output keeps those two apart without being told twice, and doesn’t let the second quietly imply the first.

Section 8 should contain something that costs you a little. If “what we would change” only lists improvements you were already planning, that section wasn’t written from the data.

How to check it

Then two more things. Read section 6 against your actual measurement instrument and confirm no claim has been bumped up a level. And check that every objective from section 3 shows up somewhere in section 6 or 7. Objectives that quietly vanish between the plan and the report are exactly what reviewers look for.

Where it goes wrong

It claims a higher level than you measured. The most common problem and the most consequential. Satisfaction data written up as competence. Self-reported intent written up as changed practice. Models do this because promotional writing is the most common kind of writing that looks like this. The instruction helps. Your own read is what catches it.

It rounds to make things read nicely. 118 becomes “approximately 120”, then “around 120 participants”, in a document of record. Tell it not to, then check anyway.

It quietly attaches data to objectives you didn’t measure. If an objective wasn’t measured, the honest report says so. Left alone, a model will find something in the data to pin to it, and now you’ve claimed a measurement you never took.

It slides into marketing language. “Robust”, “highly engaging”, “significant success”. Your accreditation file is not a brochure. Pasting the sample report is the best fix. Pasting two samples is better.

Copilot
Microsoft 365 Copilot can work with supported Word and Excel files and permitted SharePoint content. Draft in sections and trace every figure back to the source.
ChatGPT
A ChatGPT Project can keep a report template, source files, examples and project instructions together. Use an approved workspace and trace every figure and claim to the source.
Claude
A paid Claude Project can keep a report template, source files and examples in its project knowledge. Use an approved workspace and trace every figure and claim to the source.

It gets much faster

Repeated reports can become faster once you have built the reusable half: your template, two approved examples in your own voice, and standing instructions about claim levels and rounding.

Save those three things somewhere permanent the first time you make them. That’s the real gain here, not any individual draft.