Skip to the lesson
CivOps AI Academy · F21Go Live: the Go-Live Review, Handover to the Floor and the Certification Practical
0%

Chapter 1 · Ready, and proven ready

The go-live review

Going live is a decision, not a date. A short meeting with the right people checks a written list, every item green or not, and the plant manager says go or not yet. Then the platform reaches the floor one line and one shift at a time, with paper still at hand, so nothing the floor depends on rests on a single morning.

30 min12-point gateDecided and written downPilot, then parallel

By the end of this chapter you can

  • Run a go/no-go review against a written checklist with pass or fail criteria.
  • Choose a cutover strategy and explain why a pilot with a parallel week is the safe default.
  • Plan the pilot week by week, with the evidence that allows each next step.
  • Treat go-live as a managed change, with a way back to paper.

Google's site reliability engineers run every launch against a checklist, because launches fail on the items nobody thought to check, not on the clever parts [1]. A plant has the same habit for equipment: a new machine is commissioned against a list before production runs on it. Your platform deserves the same.

The checklist

Eight points come from the certification practical: they prove the platform. Four more prove the floor is ready. Every item is pass or fail; "nearly" is fail.

The go/no-go gateTwelve boxes, eight about the platform (the certification practical's points) and four about the floor, all feeding one go or no-go decision by the plant manager. Platform (the practical's eight points)The floorDeploys from the repo, by someoneelseRow-level security on every tableNew routes refused until openedThree surfaces pass the phonecheckA record travels operator →review → managerClamps and OWASP suite passRestore tested and timedIT package matches what runsPilot crew trainedPaper fallback printed at thelineSupport rota for two weeksGo-live date told to the floorGo / no-goevery box green, decided by the plantmanager, written down
The go/no-go gate. The cyan boxes are the platform, the amber boxes are the floor. One amber box red is as much a no-go as a cyan one.
Floor itemPass when
Pilot crew trainedEvery operator and supervisor on the pilot line and shift has done the four-step instruction (chapter 2) on the screens they will use
Paper fallback printedBlank paper forms for one full shift are at each pilot station, and the supervisor knows when to switch to them
Support rotaA named person is reachable on every pilot shift for the first two weeks, with the runbook to hand
Floor toldThe date, the line, the shift, why it is changing and who to ask are on the shift notice board and said at the start-up meeting

The go/no-go meeting

  • Who: the plant manager (decides), the pilot line's supervisor, the platform owner, quality, and IT if any item touches the network or accounts.
  • How long: 30 minutes. The checklist is sent the day before with evidence linked for each item.
  • How: read each item, look at its evidence, mark it green or red. Any red is "not yet", with an owner and a new date.
  • Record: the decision, the date, who decided, and the checklist as it stood, saved in the repository next to the runbook.

Cutting over

Cutover strategiesThree ways to go live drawn as timelines: big bang switches everyone at once; a pilot starts with one line and shift; pilot plus parallel also keeps paper for the first week and compares results. weeks after go-live012345678Big bang: highest riskeveryone switches on one day; paper stopsPilot: lower riskone line, one shift first; then the restPilot + parallel: lowest riskthe pilot also keeps paper for a week, and results arecomparedpaper and platform side by sidepilot line onlywhole planteveryone at once
Three ways to go live. A pilot limits who is affected; a parallel week proves the platform against the old way before the old way stops.
StrategyForAgainst
Big bangQuick; one training pushEvery problem hits every line at once; no comparison
PilotProblems hit one line; lessons improve the rolloutTwo ways of working for a few weeks
Pilot with a parallel weekResults compared day by day before paper stopsThe pilot crew does double entry for a week

The default for a plant platform is the third: a pilot on one line and one shift, with paper kept for the first week. Double entry for a week is a real cost to the pilot crew, so keep that week short, thank them, and use what they find.

The pilot planFive weeks: a dress rehearsal, then Line 2 day shift with paper in parallel, Line 2 all shifts, all lines, and paper retired. Week 0Dress rehearsalfull shift on stagingwith the pilot crewWeek 1Line 2, day shiftplatform and paper sideby side; compare dailyWeek 2Line 2, all shiftspaper stops on Line 2if week 1 matchedWeek 3All lineschampions on everyshiftWeek 4Paper retiredreview, then hand overto normal runningEach step needs the previous week's numbers to match; if they do not, stay a week longer
The pilot plan. Each step needs the previous week's comparison to match.

What "match" means

Each day of the parallel week, the supervisor compares the platform's numbers with the paper's: entries per shift, total downtime minutes, scrap counts, quality checks done. Agree the tolerance in advance (for example, the same number of entries and downtime totals within five minutes). A difference is not a failure; an unexplained difference is. Find out why before moving on.

The way back

Write down in advance what makes you go back to paper on the pilot line (for example, the platform unavailable for more than 30 minutes, or entries lost) and who decides. Going back is not a defeat; it is the paper fallback doing its job while the runbook is followed.

Exercise · Prepare and hold the go-live review60 minutes, including the meeting

You need: The checklist above, your repository, the runbook, the people listed

You will run the review that decides your go-live.

Outcome: A recorded go or not-yet decision, with the evidence behind it and a pilot plan ready.

Knowledge check

Seven of the eight platform items are green, but the paper fallback is not printed. What does the review decide?

Knowledge check

Why run a parallel week on the pilot line?

Knowledge check

In the parallel week, the platform shows 12 fewer downtime minutes than paper on Tuesday. What next?

References

  1. Google SRE book: Reliable product launches at scale. https://sre.google/sre-book/reliable-product-launches/
  2. ISO 9001:2015 Quality management systems: Requirements. https://www.iso.org/standard/62085.html
  3. OSHA 29 CFR 1910.119: Process safety management of highly hazardous chemicals. https://www.osha.gov/laws-regs/regulations/standardnumber/1910/1910.119

Chapter 2 · Training, support and who owns what

Handover to the floor

A platform the floor does not use is a cost with no return. People adopt a new way of working when they know why it is changing, can do it confidently, and get help the moment they are stuck. This chapter covers the training, the first two weeks of close support, and who owns the platform once the project is over.

30 minTWI Job InstructionTwo weeks of hypercareOne owner per activity

By the end of this chapter you can

  • Teach each screen at the line with the four-step TWI Job Instruction method.
  • Run two weeks of close support with a floor walker, line champions and a daily review.
  • Set up the support path from the floor to the platform owner and IT.
  • Agree who is responsible and accountable for the platform in normal running.

Why people adopt, or do not

Change practitioners often describe what a person needs in order to change in five steps: awareness of why, desire to take part, knowledge of how, the ability to do it, and reinforcement so it sticks (Prosci's ADKAR model) [1]. The platform's design already helps with ability: big targets, few fields, works offline. The rest is up to you: explain why at the start-up meeting, train on shift, and show the floor that what they report gets fixed.

Training at the line

Training Within Industry (TWI) was developed in the United States during the Second World War to train new workers fast, and its Job Instruction method is still used in plants today [2]. It fits a new screen well: short, at the workstation, on shift.

TWI Job InstructionFour steps in a loop: prepare the worker, present the operation, let them try it out, and follow up. 1 · Prepare the workerput them at ease; say what the job is and why it matters2 · Present the operationshow it one step at a time, stressing the key points3 · Try outthey do it and explain each step back; correct at once4 · Follow upcheck often at first, then taper off; say who to askAbout five minutes a person, on shift, at the tablet, for each screen they use
TWI Job Instruction. Four steps, about five minutes a person for each screen they use.
  1. Before: break each screen into its important steps and key points ("tap the line, then the reason; the timer starts on its own"). Print one card per station.
  2. Prepare: at the tablet, say what is changing and why: "no more typing these up at the end of the shift; your supervisor sees it straight away."
  3. Present: do it once, slowly, saying each step and key point.
  4. Try out: they do it and tell you each step; correct gently at once; repeat until they can do it and explain it.
  5. Follow up: come back during the shift, then the next day; tell them who the line champion is.

Hypercare: the first two weeks

HypercareBars for ten working days, falling from about fourteen questions a day to one. The first week has a floor walker on every shift; the second relies on line champions with the owner on call. Questions and problems raised a working day, pilot weeks (illustrative)051015day 1day 2day 3day 4day 5day 6day 7day 8day 9day 10floor walker on every shiftchampions answer; owner on call
Hypercare. Questions are many at first and fall quickly when they are answered quickly (the numbers are illustrative).
  • Week 1: a floor walker on every pilot shift, usually the platform owner or a champion, standing near the tablets for the first hour and checking back after breaks.
  • A feedback channel the floor can reach in seconds: a whiteboard by the line, or a "something's wrong" button on the operator screen that records the screen, the time and a sentence.
  • A ten-minute daily review with the supervisor: yesterday's comparison with paper, the questions raised, what was fixed. Small fixes go through the pipeline the same day and are announced at the next start-up meeting.
  • Week 2: champions answer, the owner is on call. Hypercare ends when questions are down to one or two a day and the comparison matches.

Who to ask

The support pathFive steps: operator, line champion, shift supervisor, platform owner and IT or provider, with response times growing along the path. Operatoruses the platformLine championon every shift; answershow-toShift supervisordecides: paper fallbackor waitPlatform ownerrunbook: roll back,fix, restoreIT or providernetwork, accounts,outagesat oncewithin 15 minwithin 1 h on shiftper their processProblems go one step at a time; a champion who cannot help in 15 minutes calls the supervisor
The support path. Each step tries to help before passing it on, within an agreed time.

Print the support path on the station card with names and phone numbers for each shift. The supervisor's question is always the same: can the floor carry on, on the platform or on paper? The platform owner's tool is the runbook from Session 18.

Ownership after go-live

When the project ends, the platform becomes part of how the plant runs. Write down who does what, with exactly one person accountable for each activity. A RACI chart does this on one page: Responsible, Accountable, Consulted, Informed.

RACI after handoverA grid of six activities against four roles, showing who is responsible, accountable, consulted and informed once the platform is in normal running. Plant managerPlatform ownerSupervisorsITDecide what changes nextARCIBuild and review changesIACIRun the floor on itACRIDrills, backups, keysIAICRe-review IT packageIRIAMonthly bill reviewARIIR does the work · A answers for it (one per row) · C is consulted · I is informed
RACI after handover. One A per row. The platform owner is accountable for the platform itself; the plant manager for what it is used for.
Measure of adoptionWhere to read itHealthy after a month
Entries per shift against expectedThe manager view (Session 11)Within the agreed tolerance of paper
Paper forms still usedAsk the supervisor; check the paper trayOnly during a fallback
Time from entry to supervisor reviewThe review queueWithin the shift
Requests raised and fixedThe feedback channel and merged pull requestsRequests keep coming, and most are fixed within a week
Exercise · Train the pilot crew and set up support2 hours across a shift

You need: The pilot line's tablets, printed station cards, the support path with names, the runbook

You will train the pilot shift and put the support around it.

Outcome: A trained pilot crew, a champion, a working feedback channel and a written split of ownership.

Knowledge check

Which step of TWI Job Instruction has the operator do the task and explain each step back?

Knowledge check

During hypercare an operator reports a confusing button label. What is the right response?

Knowledge check

In a RACI chart, how many people are accountable for each activity?

References

  1. Prosci: The ADKAR model. https://www.prosci.com/methodology/adkar
  2. TWI Institute: Training Within Industry. https://www.twi-institute.com/

Chapter 3 · Show it working, point by point

The certification practical

The practical is a live review: you show a CivOps reviewer your deployed platform meeting eight points, each with evidence you built during the course. Together with the unit checks, the five homework assignments and the exam, it earns the Foundation certificate. After it, the platform is simply part of how your plant runs.

25 min8 points, liveEvidence, not slidesThen 30, 60, 90 days

By the end of this chapter you can

  • Prepare the evidence for each of the practical's eight points.
  • Hand in the practical and know what the reviewer will check.
  • Know what the Foundation certificate needs: unit checks, homework, exam and practical.
  • Plan the first 90 days after go-live.

The eight points and their evidence

The practical's evidenceEight practical points, each with the evidence shown at the live review and the sessions where it was built. Practical pointEvidence you showSessionsDeployed by someone elsea merged pull request by a second person1, 18Row-level security on every tablegenerated policies and a test per role6, 7New route refusedthe default-deny test8Phone check on three surfaces390 px screenshots and the audit10, 11Record end to endone record traced live13Clamps and OWASP passthe CI run9, 17Restore tested and timedthe drill log18IT package matchesthe tag and a green drift check19
The practical's evidence. Every point was built in an earlier session; the practical shows it still holds on the live platform.
PointWhat the reviewer seesHow to prepare
Deployed from the repository by someone other than youA merged pull request by a second person, and the deployment it producedAsk your second owner to merge a small change the week before
Row-level security on every table, generated from the matrixThe generated policies, a table list with none missing, and the per-role tests passingRun the access tests; open the matrix next to the policies
A new route is refused until the matrix opens itThe default-deny test, and a request to an unlisted route refused liveKeep a test route ready that is not in the matrix
Operator, supervisor and manager surfaces pass the phone checkThe three screens on a phone at 390 px, with 44 px tap targets [2], and the audit outputRun the phone audit; bring a phone
A record travels from the operator screen through review to the manager viewOne record entered, approved and seen on the dashboard, liveUse a demo login per role on the staging data
Quality clamps and the OWASP Top 10 suite pass [1]The latest CI run, greenRe-run CI on main the day before
The runbook's restore was tested and timedThe drill log entry with date, who and minutesDo a fresh restore drill within the last month
The IT approval package matches what runsThe approved tag and the drift check green on mainCheck nothing has drifted since sign-off

Handing it in

  1. On the course page, open the practical ("your deployed spine, checked against 8 points").
  2. Hand in the deployed address, the repository link and two or three times that suit you for a live review. Give the reviewer read access to the repository through GitHub; never send passwords or keys.
  3. Prepare demo logins for each role on the staging data, and share them through your company's password manager or a one-time link, not in the hand-in notes.
  4. In the review, go point by point. If a point fails, the reviewer says what is missing; fix it and book another review.

The Foundation certificate

PartWhat it needs
Unit checksOne short check per unit, passed
HomeworkAll five assignments accepted against their rubrics
Certification exam25 questions in 45 minutes, 80% to pass; it opens once every unit check is passed
PracticalAll eight points shown on the live platform

Homework 5, due at the end of this week, collects the last pieces: the OWASP suite results (Session 17), the runbook (Session 18), the IT approval package (Session 19) and the cost comparison (Session 20).

After go-live: 30, 60, 90 days

WhenDoEvidence
Day 30Retire paper on all lines; cancel the subscriptions the platform replaced; first monthly drill and bill reviewCancelled invoices; drill log; adoption measures
Day 60Re-time the three tasks from Session 20 and update the cost case with real numbersThe updated one-page case
Day 90Review with the plant manager: what the platform changed for the decision in your intent (Session 2), and what to build nextA short written review; the next intent page

The next intent page starts the cycle again for the next problem, on the same spine: same accounts, same matrix, same pipeline, same runbook. That is what the spine was for.

Exercise · Book and pass the practical60 minutes to prepare, plus the review

You need: Your platform, repository, phone, drill log, IT package tag and the course page

You will gather the evidence for every point and book the live review.

Outcome: The practical handed in with evidence for every point, and the first 90 days planned.

Knowledge check

How does the reviewer get access to your code for the practical?

Knowledge check

Which is the strongest evidence for 'the restore was tested and timed'?

Knowledge check

What happens at day 90 after go-live?

References

  1. OWASP Top 10. https://owasp.org/Top10/2025/
  2. W3C: Understanding WCAG 2.2 success criterion 2.5.5, Target Size (Enhanced). https://www.w3.org/WAI/WCAG22/Understanding/target-size-enhanced.html
  3. Google SRE book: Reliable product launches at scale. https://sre.google/sre-book/reliable-product-launches/

Chapter 4 · 12 questions · 80% passes

Final assessment

Twelve questions across the element. Score 80% (10 of 12) to pass. Your LMS records your score and each answer; you can review the chapters and try again.

15 min12 questions≈ 15 minutesRetake allowed

Choose one answer for each question, then submit. You will see the right answer and why for every question.

1. Who makes the go/no-go decision in the course's review?
2. An item on the checklist is 'nearly' done. How is it marked?
3. Which cutover strategy does the course recommend by default?
4. In the parallel week, what counts as a failure?
5. Your platform records data a process covered by OSHA's process safety management standard relies on. What else applies?
6. What are the four steps of TWI Job Instruction?
7. What is a line champion?
8. When does hypercare end?
9. In ADKAR, what comes before knowledge of how to change?
10. How many people are accountable for each activity in a RACI chart?
11. Which is acceptable evidence for 'deployed by someone other than you'?
12. What does the Foundation certificate need besides the practical?