Workflow template for operations and process improvement
AI pilot result write-up
When an AI workflow's pilot ends, the pilot owner and the leader who decides get two pages: what was tested and on how many cases, what passed, what failed and why, what nobody tested, the bar set beside the results, and the decision left open. It recommends nothing.
Listed under
Add it to a workflow in the FullyPowered Planner, beside the teams and roles that will run it. Book a fit call
The outcome
The end of a pilot turns into a decision the leader can make on the facts. The leader who decides reads, on two pages, exactly what the pilot showed: how many cases were tested, which passed, which failed and for what reason, and how the workflow did on real work in the pilot weeks. Every count is shown as a number of cases, so a strong result on a handful of runs reads as a handful of runs.
Failures are grouped by the causes your own people wrote down when they judged them, with the cases behind each group, and a run someone had to correct is shown beside them rather than hidden among the passes. What nobody tested sits in its own section, so a gap is never mistaken for a pass. The bar the owner wrote before the pilot is set beside the results line by line, with the working for each.
Then the page lays out what the whole team would need: the access still missing, the instructions to fix, the records to tidy and who owns the workflow from here. The decision stays open as three options, roll out, change and re-test, or stop, each with what it would take. The write-up recommends none of them, claims no saving and no business effect, and names roles, never people.
Who starts the work, and when
The pilot owner starts it at the end of the pilot period, once the last test run has been judged and the run log is closed: usually whoever leads the team that ran it, such as an operations lead, department head or team lead. The leader who decides whether the workflow goes to the whole team, such as the head of operations, reads it before the decision meeting. The expert who judged the test runs answers any questions it raises.
What they supply
- The acceptance test results: each test case and run, what was expected, what came out, pass or fail, and the role that judged it, with the cause of each failure in the judge's words.
- The run log from the pilot weeks: each piece of real work the workflow prepared, its date and version, and whether it was used as prepared, edited or set aside, with the reason.
- The bar the owner wrote before the pilot: each line of what has to be true for the workflow to go to the whole team.
- The list of what was deliberately not tested, with why.
- The people who took part, by role, and what each role did.
- Open items noted during the pilot, on access, instructions and records, and who owns the workflow afterwards: business ownership, technical maintenance, user support and change approval.
- The settings: the final version, the pilot dates, the pilot owner, the leader who decides and the team it would go to.
What comes out
One draft write-up in seven sections: what was tested and on how many cases; what passed; what failed, grouped by cause with the cases behind each; what was not tested and so is not known; the bar set beside the results, each line met, not met, not shown or left for the leader to judge; what extending the workflow to the whole team would take; and the decision as three options with what each would need. Numbered questions for the pilot owner follow.
When a column does not match, a result is recorded in words outside the agreed list, or the pilot has not ended, the work stops and asks. It never counts a failed or unjudged run as a pass, never writes a bar after the fact, and sends nothing.
An example
Illustrative example with sample data. The company, its orders, its people and every date and result below are illustrative. It is the same illustrative 60-person software company as the pilot acceptance-test preparation example: its client success team piloted a workflow that prepares the handoff record a client success manager reads before a new client's kickoff. People appear by role only. Nothing here happened.
The pilot ran from 5 to 30 October 2026. On Monday 2 November 2026 the client success lead ran the write-up for the head of customer operations, who decides whether the workflow goes to the whole client success team. The inputs come first, in full, so the workflow template can be run on them. The draft it returned follows.
The inputs
1. Settings (typed by the client success lead)
Workflow: Client-onboarding handoff record
Final workflow version: 1.1
Pilot start: 5 October 2026
Pilot end: 30 October 2026
Pilot owner: Client success lead
Deciding leader: Head of customer operations
Whole team: the client success team, 9 client success managers
Few-cases line: 10
Record kept at: the pilot folder on the shared drive
Today: 2 November 2026
2. Acceptance test results: test_results.csv
case_id,case_kind,run,date,workflow_version,expected,output_summary,result,reviewer_role,cause
Order 3317,ordinary,1,2026-10-05,1.0,Complete record with every item sourced,Complete record with every item sourced,pass,Senior client success manager,
Order 3317,ordinary,2,2026-10-05,1.0,Complete record with every item sourced,Complete record with every item sourced,pass,Senior client success manager,
Client C,missing,1,2026-10-05,1.0,Marked incomplete with one question for the account executive,Marked incomplete with one question,pass,Senior client success manager,
Client C,missing,2,2026-10-05,1.0,Marked incomplete with one question for the account executive,Marked incomplete with one question,pass,Senior client success manager,
Order 3320,contradictory,1,2026-10-05,1.0,Both start dates shown with sources and the start marked unconfirmed,Order form date only,fail,Senior client success manager,used one start date without showing the other
Order 3320,contradictory,2,2026-10-05,1.0,Both start dates shown with sources and the start marked unconfirmed,Order form date only,fail,Senior client success manager,used one start date without showing the other
Client D,ambiguous,1,2026-10-05,1.0,Both project leads shown and a question asked,First project lead chosen,fail,Senior client success manager,picked one contact without asking
Client D,ambiguous,2,2026-10-05,1.0,Both project leads shown and a question asked,Both shown with a question,pass,Senior client success manager,
Order 3325,stop and ask,1,2026-10-05,1.0,Stops and refers the discount to the sales lead,Stopped and referred to the sales lead,pass,Senior client success manager,
Order 3325,stop and ask,2,2026-10-05,1.0,Stops and refers the discount to the sales lead,Stopped and referred to the sales lead,pass,Senior client success manager,
Order 3317,ordinary,1,2026-10-19,1.1,Complete record with every item sourced,Complete record with every item sourced,pass,Senior client success manager,
Order 3317,ordinary,2,2026-10-19,1.1,Complete record with every item sourced,Complete record with every item sourced,pass,Senior client success manager,
Client C,missing,1,2026-10-19,1.1,Marked incomplete with one question for the account executive,Marked incomplete with one question,pass,Senior client success manager,
Client C,missing,2,2026-10-19,1.1,Marked incomplete with one question for the account executive,Marked incomplete with one question,pass,Senior client success manager,
Order 3320,contradictory,1,2026-10-19,1.1,Both start dates shown with sources and the start marked unconfirmed,Both dates shown and start unconfirmed,pass,Senior client success manager,
Order 3320,contradictory,2,2026-10-19,1.1,Both start dates shown with sources and the start marked unconfirmed,Both dates shown and start unconfirmed,pass,Senior client success manager,
Client D,ambiguous,1,2026-10-19,1.1,Both project leads shown and a question asked,Both shown with a question,pass,Senior client success manager,
Client D,ambiguous,2,2026-10-19,1.1,Both project leads shown and a question asked,Second project lead chosen,fail,Senior client success manager,picked one contact without asking
Order 3325,stop and ask,1,2026-10-19,1.1,Stops and refers the discount to the sales lead,Stopped and referred to the sales lead,pass,Senior client success manager,
Order 3325,stop and ask,2,2026-10-19,1.1,Stops and refers the discount to the sales lead,Stopped and referred to the sales lead,pass,Senior client success manager,
3. Run log: run_log.csv
run_id,date,workflow_version,item,outcome,cause,run_by_role
R-01,2026-10-06,1.0,Order 3341,used as prepared,,Client success manager
R-02,2026-10-07,1.0,Order 3344,used as prepared,,Client success manager
R-03,2026-10-08,1.0,Order 3346,edited,missed a promise in the call notes,Client success manager
R-04,2026-10-09,1.0,Order 3349,used as prepared,,Client success manager
R-05,2026-10-13,1.0,Order 3352,set aside,order form not attached,Client success manager
R-06,2026-10-14,1.0,Order 3355,used as prepared,,Client success manager
R-07,2026-10-16,1.0,Order 3357,edited,wrong billing contact,Client success manager
R-08,2026-10-20,1.1,Order 3360,used as prepared,,Client success manager
R-09,2026-10-21,1.1,Order 3362,edited,missed a promise in the call notes,Client success manager
R-10,2026-10-22,1.1,Order 3365,used as prepared,,Client success manager
R-11,2026-10-23,1.1,Order 3367,set aside,two start dates on a multi-year order,Client success manager
R-12,2026-10-26,1.1,Order 3370,used as prepared,,Client success manager
R-13,2026-10-27,1.1,Order 3372,used as prepared,,Client success manager
R-14,2026-10-29,1.1,Order 3375,used as prepared,,Client success manager
R-15,2026-11-02,1.1,Order 3378,used as prepared,,Client success manager
4. The bar, written by the client success lead before the pilot: bar.csv
bar_id,condition,measure,scope,threshold,written_on
B1,Every acceptance test run passes on the final version,test pass share,all,100,2026-09-28
B2,No stop-and-ask run fails on the final version,no fail in kind,stop and ask,,2026-09-28
B3,At least 20 real handoffs prepared during the pilot,minimum real runs,,20,2026-09-28
B4,At least 60% of real handoffs used as prepared,used as prepared share,,60,2026-09-28
B5,The client success managers who used it would keep using it,other,,,2026-09-28
5. What was not tested: not_tested.csv
area,why_not_tested
Renewals and upgrades for existing clients,Outside the workflow's purpose for this pilot
Orders priced in a currency other than US dollars,None came in during the pilot
Use by client success managers outside the pilot group,Only three managers took part
Time taken to prepare a handoff,Not measured in the pilot
6. Participants: participants.csv
role,count,part_taken
Client success lead,1,Pilot owner
Senior client success manager,1,Expert; judged every test run
Client success manager,3,Ran the workflow on real handoffs
Account executive,2,Answered questions the handoff records raised
IT manager,1,Set up access to the AI workspace for the pilot group
7. Open items noted during the pilot: open_items.csv
area,item,raised_by_role
access,Six client success managers have no access to the approved AI workspace,IT manager
access,Client success managers cannot open the order form folder and paste order forms by hand,Client success manager
instructions,The instructions do not say what to do when an order has two start dates,Senior client success manager
records,Billing contacts in the account record are out of date for older clients,Client success manager
8. Ownership after the pilot: ownership.csv
responsibility,role
business ownership,Client success lead
user support,Senior client success manager
The output
AI PILOT RESULT WRITE-UP: Client-onboarding handoff record, version 1.1
For: Client success lead and Head of customer operations. Prepared 2 November 2026 from: acceptance test results (20 rows); run log (15 rows); the bar (5 lines); the not-tested list (4 areas); participants (5 roles); open items (4); ownership (2 of 4 named).
Pilot period: 5 October 2026 to 30 October 2026. Whole team: the client success team, 9 client success managers. Few cases: below 10.
1. What was tested
- Acceptance test, version 1.1: 5 cases, 10 runs: 9 passed, 1 failed, 0 unjudged.
- By kind, version 1.1: ordinary 2 of 2 (few cases); missing 2 of 2 (few cases); contradictory 2 of 2 (few cases); ambiguous 1 of 2 (few cases); stop and ask 2 of 2 (few cases).
- Acceptance test, version 1.0: 10 runs: 7 passed, 3 failed, 0 unjudged.
- Real work: 14 runs from 6 October 2026 to 29 October 2026 (7 on version 1.0; 7 on version 1.1): 9 used as prepared, 3 edited, 2 set aside.
- Left out: R-15, dated 2 November 2026, after the pilot ended. See question 1.
2. What passed
- Cases that passed every run on version 1.1: Order 3317 (ordinary), 2 of 2 (few cases); Client C (missing), 2 of 2 (few cases); Order 3320 (contradictory), 2 of 2 (few cases); Order 3325 (stop and ask), 2 of 2 (few cases).
- Real runs used as prepared: 9 of 14 (64.3%).
3. What failed, grouped by cause
- F1 "used one start date without showing the other": 2 failed test runs (Order 3320 version 1.0 run 1; Order 3320 version 1.0 run 2); 0 real runs set aside; 0 real runs edited. Seen on version 1.1: no.
- F2 "picked one contact without asking": 2 failed test runs (Client D version 1.0 run 1; Client D version 1.1 run 2); 0 real runs set aside; 0 real runs edited. Seen on version 1.1: yes.
- F3 "missed a promise in the call notes": 0 failed test runs; 0 real runs set aside; 2 real runs edited (R-03, R-09). Seen on version 1.1: yes.
- F4 "order form not attached": 0 failed test runs; 1 real run set aside (R-05); 0 real runs edited. Seen on version 1.1: no.
- F5 "wrong billing contact": 0 failed test runs; 0 real runs set aside; 1 real run edited (R-07). Seen on version 1.1: no.
- F6 "two start dates on a multi-year order": 0 failed test runs; 1 real run set aside (R-11); 0 real runs edited. Seen on version 1.1: yes.
Edited runs appear here because a person corrected them; they are not counted as failed.
4. What was not tested, and so is not known
- Renewals and upgrades for existing clients: Outside the workflow's purpose for this pilot.
- Orders priced in a currency other than US dollars: None came in during the pilot.
- Use by client success managers outside the pilot group: Only three managers took part.
- Time taken to prepare a handoff: Not measured in the pilot.
5. The bar set beside the results
- B1 "Every acceptance test run passes on the final version" (written 28 September 2026): 9 ÷ (9 + 1 + 0) × 100 = 90.0%, against 100. NOT MET.
- B2 "No stop-and-ask run fails on the final version" (written 28 September 2026): stop-and-ask runs on version 1.1: 2, of which 0 failed and 0 unjudged. MET on few cases.
- B3 "At least 20 real handoffs prepared during the pilot" (written 28 September 2026): 14 real runs in the pilot period, against 20. NOT MET.
- B4 "At least 60% of real handoffs used as prepared" (written 28 September 2026): 9 ÷ 14 × 100 = 64.3%, against 60. MET.
- B5 "The client success managers who used it would keep using it" (written 28 September 2026): nothing computed. FOR THE HEAD OF CUSTOMER OPERATIONS TO JUDGE: the inputs hold no record of this.
6. What extending it to the whole team would take
- Access: Six client success managers have no access to the approved AI workspace (raised by IT manager); Client success managers cannot open the order form folder and paste order forms by hand (raised by Client success manager).
- Instructions: The instructions do not say what to do when an order has two start dates (raised by Senior client success manager). Failure groups seen on version 1.1: F2, F3, F6.
- Records to tidy: Billing contacts in the account record are out of date for older clients (raised by Client success manager).
- Other: None noted.
- Who owns it: business ownership, Client success lead; technical maintenance, not named; user support, Senior client success manager; change approval, not named.
- Whole team: the client success team, 9 client success managers; 3 of its people took part, as client success managers.
7. The decision, for the Head of customer operations
- A. Roll out to the client success team, 9 client success managers. Would need: bar lines B1 and B3, not met, and B5, to judge; the two access items in section 6; a role for technical maintenance and for change approval, both not named.
- B. Change and re-test. Would need: a change for F2, F3 and F6, approved by change approval, not named; a rerun of Client D, and of Order 3325 for bar line B2; 6 more real runs to reach bar line B3; the not-tested areas, for the Client success lead to choose whether to add (4 listed).
- C. Stop. Would need: the Head of customer operations' reason, recorded with the pilot; the pilot's access withdrawn by technical maintenance, not named (access to the AI workspace set up for the pilot group, by the IT manager); the pilot record kept at the pilot folder on the shared drive.
This write-up recommends none of the three.
Questions for the Client success lead
- R-15 (Order 3378) is dated 2 November 2026, after the pilot ended on 30 October 2026, and is left out of every count. Was it part of the pilot, or the first run after it?
- No role is named for technical maintenance after the pilot. Who holds technical maintenance after the pilot?
- No role is named for change approval. Who approves a change to the workflow's instructions before it reaches the team?
Draft for the Client success lead and the Head of customer operations. Every count traces to the rows named above. No saving and no business effect is claimed. Nothing has been sent and no record has been changed.
Where it fits
- Operations and process improvement, which runs the pilot and owns the record of what it showed.
- Leadership, where the leader who decides needs the facts on one sheet before committing a whole team to a new way of working.
- Software technology, where client success, support and sales teams pilot workflows on handoffs and account records before every manager uses them.
- Professional services, where a firm pilots a workflow with a few practitioners before the whole practice relies on it with clients.
- Law firms, where the firm administrator brings the result of an intake or matter-opening pilot to the partners who decide.
- Growth & scaling, when each new team should inherit workflows that have shown what they can do.
- Established businesses, where a pilot has to earn its place beside a way of working people already trust.
It closes step two, develop the workflows, and opens step three, deploy, in how organizations plan their use of AI.
What has to change for your company
Map your test record sheet and run log to the columns the workflow template expects, and set the case kinds your acceptance tests use. Have the pilot owner write the bar before the pilot starts, in measures the template can check, such as a share of test runs passed or a number of real runs; anything else is left for the leader to judge. Set the few-cases line, which is 10 by default. Name who runs the write-up, who decides, and who will hold each of the four responsibilities after the pilot. Whether the workflow is then used and useful across the team is measured later, as Installed, used or useful? sets out. It runs in ChatGPT, Claude, Copilot or Gemini, from the files you supply; it needs no connection to any system.
What has been checked
FullyPowered AI wrote this workflow template. It was editorially reviewed on 19 November 2026. It was tried on its example on 19 November 2026 in two runs, the second published with it. It has not yet been tested in a client environment, and it has not yet been tested as a packaged version in ChatGPT, Claude, Copilot or Gemini.
Testing in your environment means running it on one or two past pilots whose results your pilot owner already knows, with hard cases written in advance: a pass with no reviewer recorded, a count of seven real runs, a bar line written after the pilot began, a failure with no cause, and a request to pick an option. The pilot owner checks every count against the rows before anyone relies on the write-up. Pilot acceptance-test preparation prepares the test whose results this reads, and what we test before we call an AI workflow ready describes the record we keep.
Who decides
Your people keep
- Whether the workflow goes to the whole team
- Which of roll out, change and re-test, or stop
- Each line the leader is left to judge
- Who owns the workflow after the pilot
It does not
- Recommends none of the three options
- Treats a failed or unjudged run as what it is, never as a pass
- Never writes a bar after the fact
- Claims no saving and no business effect
- Names roles, never people
- Sends nothing
Questions people ask
Who starts it and when?
The pilot owner starts it at the end of the pilot period, once the last test run has been judged and the run log is closed. The leader who decides reads it before the decision meeting.
What do I need to supply?
The acceptance test results, the run log from the pilot weeks, the bar the owner wrote before the pilot, the list of what was not tested, the people who took part by role, open items and who owns the workflow afterwards.
Does it say whether to roll out?
No. The decision stays open as three options, roll out, change and re-test, or stop, each with what it would take. The write-up recommends none of them.
How does it treat a result on a few runs?
Every count is shown as a number of cases, so a strong result on a handful of runs reads as a handful of runs.
What has been checked?
It was editorially reviewed on 19 November 2026. It was tried on its example on 19 November 2026 in two runs, the second published with it. It has not yet been tested in a client environment.
The full workflow template
Everything above is the outline. The complete workflow template is the working version: the roles, the inputs, the procedure, the exceptions and stop conditions, the output format, the reviewer checklist, the acceptance tests, the instructions for an AI assistant and what to customize for your company. It comes with a download and the link to its folder in FullyPowered’s public repository.
Free with an account. Sign in by email: no password and no card.
Next step
Put this workflow template to work for your team: on your plan, or built with you in a sprint.
- 01
The full workflow template
Free with an account. Sign in by email: no password and no card.
- 02
FullyPowered Planner
Add it to a workflow in the FullyPowered Planner, beside the teams and roles that will run it.
- 03
Build this in a sprint
In a sprint, FullyPowered adapts this workflow template to your rules, proves it on your own work and puts it in the hands of the roles that own it.
These workflow templates are examples and a starting point, not a complete list. We can build any workflow for your company, including one that runs several steps, in a Workflow Sprint or a scoped implementation.
Rather talk it through first? Book a fit callDiscuss adapting this
FullyPowered Planner. A card to start, everything unlocked, nothing charged for seven days.Security & your informationPlanner termsRefunds