Success Case Method Write-Up
When To Use
The sponsor asks "did anyone actually do anything with it?" and a score cannot answer. The follow-up survey is closed, the volunteer interviews are done, and you hold notes keyed by case letter. This skill answers one question: where was the programme used, what followed, what was checked, and what stopped it where it stalled?
When Not To Use
If nobody has been asked about use yet, run Training Follow-Up Survey, then Learner Interview Guide: this skill takes notes in and writes cases out, and holds neither a survey nor an interview guide. Cases never show how common a use is or what it was worth: for figures run Before-and-After Results Table, for money run Training ROI Range.
Inputs
- The interview notes, keyed by case letter, with no name, team or location in them
- The group summary of the follow-up survey answers about use, with how many were invited, how many volunteered and how many were interviewed
- Your Critical Behaviours Map, which says what success looks like
If you have none of this, I start from the programme's behaviours and a blank case table, and mark the output as a first draft. If names arrive in the notes, I strip them, do not reproduce them, and say the notes should be anonymised before they are shared.
Approach
The Success Case Method, the approach developed by Robert Brinkerhoff and described on the BetterEvaluation page, runs in five steps: focus and plan the study, define what success should look like, survey to find cases, interview and document them, then communicate findings, conclusions and recommendations. This skill is the last part: documenting and communicating. Here the contrast is between uses that worked and uses that stalled, and barriers are read as conditions of the work, because a meta-analysis of transfer research (Blume, Ford, Baldwin and Huang 2010, correlational) relates use on the job to a supportive work environment. The failure it prevents: one glowing story from an enthusiast, read by the sponsor as proof, while the missing tool that stopped everyone else never reaches the page.
Workflow
- Ask three questions at once: how many people were invited to the follow-up survey, how many volunteered and how many did you interview? Who will read the write-up, and do any of them manage people who were interviewed? Is there a case you already doubt can be kept anonymous?
- Write the selection rule at the top, with the counts: cases were picked from follow-up survey answers about use, among volunteers, with consent. If a case was nominated by a manager, or chosen for who the person is, it comes out, and I say so in one line.
- Take "what success looks like" from the Critical Behaviours Map and sort each case against it as a use that worked or a use that stalled. Stalled uses get the same care: most of the learning is there.
- Write each case in five fields: what was done, what followed, the evidence checked, what helped, what blocked. Evidence is "seen" (someone could see it and the interviewer looked), "record" (a trace the interviewer checked) or "reported only". Opinion stays labelled as opinion.
- Anonymise, then reread as the manager would: no name, team, location, date or tell-tale detail. With a small pool, even the task can give someone away. A case that cannot be anonymised is left out and counted as left out.
- Build the barrier list across cases: the barrier as a condition (a tool, time, a process, a missing job aid), how many cases it appeared in, and the role that could remove it. A barrier that points at a person is rewritten as the condition behind it.
- Write the limits into the page itself and set the claim strength: cases alone are "suggested" at most. You check every case against your notes before anything leaves your hands.
Output Format
# Success Case Write-Up
Programme: [name] | Follow-up survey closed: [date] | Written by: [name, role] | Raw notes held by: [interviewer and L&D lead only]
## How the cases were picked
- Picked from follow-up survey answers about use, among volunteers, with consent. Never by who the person is.
- Invited: [count] | Volunteered: [count] | Interviewed: [count] | Written up: [count] | Left out, could not be anonymised: [count]
- What success looks like, from the Critical Behaviours Map: [behaviour]
## Cases
| Case | Worked or stalled | What was done | What followed | Evidence checked | What helped | What blocked |
|---|---|---|---|---|---|---|
| A | [worked / stalled] | [the use] | [what followed] | [seen / record / reported only] | [condition] | [condition] |
## Barriers across cases
| Barrier (a condition, never a person) | Cases it appeared in | Role that could remove it |
|---|---|---|
| [barrier] | [count of cases, of how many] | [role] |
## Limits
- A small, chosen sample: it shows what is possible, not how common it is. No case is typical. Claim strength: [suggested / unknown].
## Decision
[Sponsor name] decides by [date] which barriers to remove, and names the role that owns each one.
Done When
- The selection rule and its four counts sit at the top, and every case came from a volunteer who gave consent
- Each case states its evidence as seen, record or reported only
- No case carries a name, team, location or detail a manager could use to work out who it is
- Every barrier has a role that could remove it, and the limits are written on the page
Quality Bar
- Uses that stalled are written with the same care and length as uses that worked
- Never write "best performers" or "worst performers": the unit is a use of the training
- No case becomes a "typical" quote or a statistic the sponsor could present as a rate
- Raw notes never go to a manager or the sponsor; only the anonymous write-up does. Privacy points: check with a qualified adviser
- A case is a use of the programme, never a verdict on a person: voluntary, with consent, anonymous, and no detail or quote is invented
Next
Run impact-training-roi-range (Training ROI Range) when the budget holder asks for a money figure.
About the makers
This pack is made by 90n (90n.ai): AI-native ways of working and tools for Product, Marketing and HR teams, built by ex-McKinsey founders. We map where time and money leak, redesign the work around AI and stay until your people run it themselves, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).
1---2name: impact-success-case-write-up3description: Writes up success cases from interview notes you already hold, producing the selection rule with its counts, anonymous write-ups of uses that worked and uses that stalled with the evidence checked for each, and a barrier list naming the role that could remove each barrier. Use for "run impact-success-case-write-up", "success case method", "evidence the training was used", "what got in the way after training", "write up the learner interviews", "did anyone actually use the training", "stories of use on the job", "show impact without a money figure", part of the Claude Guide for Training Impact Pack by 90n.4metadata:5 source: https://skillmd.com/skills/polar-bear-org/impact-success-case-write-up6---78# Success Case Method Write-Up910## When To Use11The sponsor asks "did anyone actually do anything with it?" and a score cannot answer. The follow-up survey is closed, the volunteer interviews are done, and you hold notes keyed by case letter. This skill answers one question: where was the programme used, what followed, what was checked, and what stopped it where it stalled?1213## When Not To Use14If nobody has been asked about use yet, run Training Follow-Up Survey, then Learner Interview Guide: this skill takes notes in and writes cases out, and holds neither a survey nor an interview guide. Cases never show how common a use is or what it was worth: for figures run Before-and-After Results Table, for money run Training ROI Range.1516## Inputs17- The interview notes, keyed by case letter, with no name, team or location in them18- The group summary of the follow-up survey answers about use, with how many were invited, how many volunteered and how many were interviewed19- Your Critical Behaviours Map, which says what success looks like20If you have none of this, I start from the programme's behaviours and a blank case table, and mark the output as a first draft. If names arrive in the notes, I strip them, do not reproduce them, and say the notes should be anonymised before they are shared.2122## Approach23The Success Case Method, the approach developed by Robert Brinkerhoff and described on the BetterEvaluation page, runs in five steps: focus and plan the study, define what success should look like, survey to find cases, interview and document them, then communicate findings, conclusions and recommendations. This skill is the last part: documenting and communicating. Here the contrast is between uses that worked and uses that stalled, and barriers are read as conditions of the work, because a meta-analysis of transfer research (Blume, Ford, Baldwin and Huang 2010, correlational) relates use on the job to a supportive work environment. The failure it prevents: one glowing story from an enthusiast, read by the sponsor as proof, while the missing tool that stopped everyone else never reaches the page.2425## Workflow261. Ask three questions at once: how many people were invited to the follow-up survey, how many volunteered and how many did you interview? Who will read the write-up, and do any of them manage people who were interviewed? Is there a case you already doubt can be kept anonymous?272. Write the selection rule at the top, with the counts: cases were picked from follow-up survey answers about use, among volunteers, with consent. If a case was nominated by a manager, or chosen for who the person is, it comes out, and I say so in one line.283. Take "what success looks like" from the Critical Behaviours Map and sort each case against it as a use that worked or a use that stalled. Stalled uses get the same care: most of the learning is there.294. Write each case in five fields: what was done, what followed, the evidence checked, what helped, what blocked. Evidence is "seen" (someone could see it and the interviewer looked), "record" (a trace the interviewer checked) or "reported only". Opinion stays labelled as opinion.305. Anonymise, then reread as the manager would: no name, team, location, date or tell-tale detail. With a small pool, even the task can give someone away. A case that cannot be anonymised is left out and counted as left out.316. Build the barrier list across cases: the barrier as a condition (a tool, time, a process, a missing job aid), how many cases it appeared in, and the role that could remove it. A barrier that points at a person is rewritten as the condition behind it.327. Write the limits into the page itself and set the claim strength: cases alone are "suggested" at most. You check every case against your notes before anything leaves your hands.3334## Output Format35```markdown36# Success Case Write-Up37Programme: [name] | Follow-up survey closed: [date] | Written by: [name, role] | Raw notes held by: [interviewer and L&D lead only]38## How the cases were picked39- Picked from follow-up survey answers about use, among volunteers, with consent. Never by who the person is.40- Invited: [count] | Volunteered: [count] | Interviewed: [count] | Written up: [count] | Left out, could not be anonymised: [count]41- What success looks like, from the Critical Behaviours Map: [behaviour]42## Cases43| Case | Worked or stalled | What was done | What followed | Evidence checked | What helped | What blocked |44|---|---|---|---|---|---|---|45| A | [worked / stalled] | [the use] | [what followed] | [seen / record / reported only] | [condition] | [condition] |46## Barriers across cases47| Barrier (a condition, never a person) | Cases it appeared in | Role that could remove it |48|---|---|---|49| [barrier] | [count of cases, of how many] | [role] |50## Limits51- A small, chosen sample: it shows what is possible, not how common it is. No case is typical. Claim strength: [suggested / unknown].52## Decision53[Sponsor name] decides by [date] which barriers to remove, and names the role that owns each one.54```5556## Done When57- The selection rule and its four counts sit at the top, and every case came from a volunteer who gave consent58- Each case states its evidence as seen, record or reported only59- No case carries a name, team, location or detail a manager could use to work out who it is60- Every barrier has a role that could remove it, and the limits are written on the page6162## Quality Bar63- Uses that stalled are written with the same care and length as uses that worked64- Never write "best performers" or "worst performers": the unit is a use of the training65- No case becomes a "typical" quote or a statistic the sponsor could present as a rate66- Raw notes never go to a manager or the sponsor; only the anonymous write-up does. Privacy points: check with a qualified adviser67- A case is a use of the programme, never a verdict on a person: voluntary, with consent, anonymous, and no detail or quote is invented6869## Next70Run impact-training-roi-range (Training ROI Range) when the budget holder asks for a money figure.7172## About the makers7374This pack is made by 90n ([90n.ai](https://90n.ai)): AI-native ways of working and tools for Product, Marketing and HR teams, built by ex-McKinsey founders. We map where time and money leak, redesign the work around AI and stay until your people run it themselves, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).