Orchard Rollout: Lessons Learned

Lessons learned presentation that names the causes, not the people

Orchard Rollout was a warehouse ERP project planned for 20 weeks that took 26 and cost 12% more than approved. This is the closing review deck: it puts planned against actual on slide two, walks the timeline of what happened, and ends on five lessons each attached to a change the business has already made.

Create a presentation with OneCraft10 slides, read in order

Every slide, in order

The whole deck as it renders, one slide after another. Read it the way the audience would.

ORCHARD.

Lessons learned from the warehouse ERP rollout

OR

Project close review

OCTOBER 2026 · ALL HANDS INVITED

Slide 1 · cover · The project name and nothing defensive: a close-out review owns its numbers from slide one.

Planned against actual

The honest table first. The rest of the deck explains these rows.

Dimension
Planned
Actual
Duration
20 weeks
26 weeks
Budget
$840k approved
$941k, 12% over
Scope
3 warehouse sites
3 sites, unchanged
Go-live defects
40 tolerated
118 logged
User training hours
300 planned
410 delivered
Planned is the figure approved at kickoff, before any change request.

02

Slide 2 · comparison · Planned against actual on slide two. Every later slide explains one of these rows.

What happened

Six milestones, three of them slips, in the order they landed.

Week 0

Kickoff with all three sites and the vendor implementation team.

Week 6

Data migration slips 3 weeks: the item master held 9,000 duplicates nobody had profiled.

Week 11

Vendor swaps out its lead consultant mid-build; two weeks of context rebuild.

Week 15

A second UAT round is added after round one fails 61 scripts, buying 2 more weeks.

Week 24

Cutover weekend lands clean: zero downtime, all three sites live Monday.

Week 26

Hypercare runs a week long, closing 118 defects down to 9 open.

03

Slide 3 · timeline · What actually happened, in six milestones, slips included with their week counts.

Where the 12% went

Approved $840k to final $941k, with each overrun named.

Approved
Migration rework
Second UAT round
Contractor extension
Training venue saving
Final spend
0$k500$k1000$k1500$k2000$k

04

Slide 4 · metrics · The budget bridge from $840k approved to $941k spent, overrun by named cause.

Why six weeks late

Causes, not culprits. Every bone was reviewed with the people who did the work.

6 weeks late
Data
9,000 duplicate items
No data owner named
Process
61 failed scripts
No entry criteria
Vendor
Lead swapped week 11
No named-staff clause
People
Pick season clash
Backfill unfunded
Governance
Fortnightly steering
No decision log

05

Slide 5 · framework · The fishbone asks why the project ran late, with causes on the bones, never names.

What went well

Three things worked and should be copied into the next rollout unchanged. Each carries its evidence, same as the failures.

The Cutover Weekend

  • Zero downtime across all three sites
  • Rehearsed twice, timed to the minute
  • Rollback plan printed and never needed
  • Monday orders shipped on schedule

Training That Stuck

  • 98% completion before go-live
  • 410 hours, floor based, not classroom
  • Super users trained three weeks early
  • Help desk tickets halved by week two

Change Champions

  • One champion per shift per site
  • Caught 30 defects before UAT did
  • Ran the go-live floor walks
  • Kept the rumour mill factual

06

Slide 6 · pillars · What went well, with evidence, because a review that only lists failure teaches half.

Five lessons

Written as instructions for the next rollout, and each one maps to a change on the next slide.

Profile before you plan

Profile the data before the plan is priced. The 9,000 duplicates were findable in week zero for a day of work.

Two UAT rounds, always

Baseline two rounds with entry criteria. A single round only ever discovers that you needed two.

Name the vendor staff

Put named key staff and a substitution clause in the contract, priced, not promised.

Steer weekly, log decisions

Weekly steering with a decision log. Fortnightly meetings cost the project three separate weeks of waiting.

Budget hypercare

Hypercare is project work. Fund it as a line item so the team that built it is still there to fix it.

07

Slide 7 · checklist · The five lessons, written as instructions to the next project, not observations.

Already changed

These are not proposals. The project method changed before this meeting, and the next rollout inherits every row.

How Orchard ran

The method now

Data quality assumed good until migration proved otherwise

One UAT round in every baseline plan

Vendor staffing handled by relationship

Fortnightly steering, decisions in minutes nobody reread

A funded data profiling spike before any plan is priced

Two UAT rounds with entry and exit criteria in the template

Named staff and substitution clauses in every vendor contract

Weekly steering with a public decision log, hypercare funded

08

Slide 8 · comparison · The changes already adopted, was and now, so the review provably produced something.

FOR THE NEXT ROLLOUT

Four rules Orchard paid for.

01

Profile the data first.

No plan is priced until the data has been profiled. Estimates built on unprofiled data are fiction with a spreadsheet.

02

Plan the second round.

UAT round two is in the baseline, not the contingency. Contingency is for surprises, and round two is not a surprise.

03

Contract the people.

The vendor sells a team, not a logo. If the names can change silently, the price should too.

04

Decide weekly.

A decision that waits a fortnight costs a week on average. The log makes waiting visible, and visible waiting gets fixed.

09

Slide 9 · takeaway · Four rules for the next rollout, stated as imperatives on a dark slide.

Discussion

The full 40 page close report is on the project wiki under Orchard.

10

Slide 10 · qa · Discussion, and where the full report lives, because ten slides summarise forty pages.

The structure

What each slide is doing, so you can reuse the order even with different content.

Slide 1cover
The project name and nothing defensive: a close-out review owns its numbers from slide one.
Slide 2comparison
Planned against actual on slide two. Every later slide explains one of these rows.
Slide 3timeline
What actually happened, in six milestones, slips included with their week counts.
Slide 4metrics
The budget bridge from $840k approved to $941k spent, overrun by named cause.
Slide 5framework
The fishbone asks why the project ran late, with causes on the bones, never names.
Slide 6pillars
What went well, with evidence, because a review that only lists failure teaches half.
Slide 7checklist
The five lessons, written as instructions to the next project, not observations.
Slide 8comparison
The changes already adopted, was and now, so the review provably produced something.
Slide 9takeaway
Four rules for the next rollout, stated as imperatives on a dark slide.
Slide 10qa
Discussion, and where the full report lives, because ten slides summarise forty pages.

How to adapt this deck

A sprint retrospective keeps slides two, five and seven, planned versus actual, causes and lessons, and drops the budget bridge; twenty minutes, team only. An incident review swaps the milestone road for a minute by minute timeline and tightens the fishbone to the single failure, which is the postmortem shape the SRE literature describes. For a project that went well, the deck matters even more and almost nobody writes it: keep the same slides, and let the fishbone explain why it worked, because unexamined success is how the next project inherits luck instead of method. Orchard is the same fictional project the project kickoff deck example opens, so the two pages bracket a full project life. Whichever version you run, write the lessons as instructions rather than observations. A lesson that reads we should communicate better changes nothing, and a lesson that reads the site survey happens before the quote goes out changes the next project.

What makes this deck work

Planned versus actual opens the deck

Slide two states 20 weeks planned against 26 actual, $840k against $941k, 40 tolerated defects against 118 logged. Every later slide explains one of those rows: the timeline explains the weeks, the bridge explains the dollars, the fishbone explains both. A review that buries this table is managing perception, not learning.

The fishbone carries causes, not names

Five bones, data, process, vendor, people and governance, each with two specific causes like the 9,000 duplicate items and the fortnightly steering cadence. Not one bone is a person. That is what makes the five lessons presentable in a room containing everyone who worked the project.

Every lesson has a change already made

The five lessons map one to one onto the was and now slide: profile before pricing became a funded spike, one UAT round became two with entry criteria, the staffing handshake became a contract clause. A review that ends in adopted changes rather than recommendations has already paid for its meeting.

Questions people ask

What is a lessons learned presentation?

The closing review of a project, presented to the team and its sponsors: what was planned, what happened, why the gaps opened, what went well, and what the organisation changes as a result. This example reviews a late, over budget ERP rollout in ten slides without naming a single culprit.

How do I structure a lessons learned meeting?

Follow the deck order: the honest numbers first, then the narrative timeline, then cause analysis, then the positives, then lessons and changes. Sixty to ninety minutes with the whole delivery team present. Circulating the planned versus actual table beforehand stops the meeting starting in dispute.

How do I present failure without blame?

Put causes on systems and artefacts, never on roles or names, the way this fishbone hangs the delay on an unprofiled item master and a fortnightly steering cadence. Blameless does not mean consequence free; it means the fixes land on contracts, templates and cadences, which are the things that persist.

What should planned versus actual compare?

Duration, budget, scope and quality at minimum. This deck adds training hours because the overrun there was a good decision worth seeing. Include a row where the plan held, scope stayed at three sites here, so the table reads as measurement rather than confession.

How soon after the project should the review run?

Two to six weeks after close: late enough for hypercare data like the defect count to settle, early enough that memories and the team are still in place. Orchard closed hypercare in week 26 and reviewed in week 30. A review scheduled next quarter is a review that never happens.

Is this the same as a sprint retrospective?

Same intent, different scale. A retrospective inspects a two week window with the team only and skips the money; this reviews a whole project with sponsors present, so it carries a budget bridge and contract changes. The adaptation notes below cover shrinking this shape to sprint size.

Build your own in about a minute

The button below opens the generator with this use case already described. Change the wording to match your own, generate, then edit anything you like.

Make my lessons learned presentation that names the causes, not the people

Other presentation examples

Want the steps in the builder? Read Add charts, maps and diagrams to slides, then Edit your deck with the AI assistant. For everything this generator can do, see the presentation maker.

Sources

Written and checked by the OneCraft team. Last checked .