C4, Lesson 7: The Control Tower and Your Roadmap: Seeing Everything, Fixing Something
Tags: Sourced / General / Our view / Illustrative. About 14 minutes.
1. The story: six days in the Suez Canal (March 2021)
On March 23, 2021, the 400-metre container ship Ever Given ran aground in the Suez Canal in high winds and wedged across the channel. It was refloated on March 29. Reuters reported that more than 400 ships were waiting, and quoted the Suez Canal Authority chairman, Osama Rabie, saying the backlog of 422 ships could be cleared in about 3.5 days. [Sourced: Reuters, March 29, 2021, https://www.reuters.com/world/stranded-suez-canal-ship-re-floated-inch-cape-2021-03-29/ ; refloating announcement: Suez Canal Authority, March 29, 2021, https://www.suezcanal.gov.eg/English/MediaCenter/News/Pages/nav_29-03-2021.aspx] Why tell it in a lesson about AI? Because the first problem for every shipper was not "what is the best plan" but "which of my goods are on those ships, and what do I do about each one?" Companies with a live picture of their shipments could re-route, re-promise and warn customers within days. Companies without one found out by calling around. [Our view; the Canal facts are sourced, the lesson drawn is ours]
2. The one idea
A "control tower" is a shared screen and a set of habits: one live view of orders, shipments and stock, with alerts, and named people who act on them. The software is the small part. A tower that shows problems nobody owns is a TV in an empty room. [General]
3. The kitchen-table version
You have 20 parcels out, 10 supplier orders open, and 3 big customers. A wall chart shows where each one is and flags anything late. That is a control tower. AI's part is reading the incoming messages, tracking data and news, guessing which flags matter, and drafting a response. You still decide. [General]
4. What the pieces are
- Data feeds: orders, shipments, stock, supplier confirmations, carrier tracking.
- Alerts: late, short, damaged, at risk.
- Predictions: estimated arrival, probability of a stockout (builds on Lessons 2 and 3).
- Playbooks: if X happens, who does what within what time.
- Owners: a named person per alert type.
5. How far have companies got? (survey evidence)
McKinsey's supply chain leader surveys report: 67% of respondents said they had implemented digital dashboards for end-to-end visibility in 2022, and 79% in 2023. In 2022 only 17% said they had a good picture of their supply chains down to the third tier or beyond (2% a year earlier). In 2023, 71% expected to revise planning processes and governance within three years, which the authors tie to visibility exposing process weaknesses. [Sourced, secondary: McKinsey Global Supply Chain Leader Survey 2022 and 2023, self-reported by executives at large companies; https://www.mckinsey.com/capabilities/operations/our-insights/taking-the-pulse-of-shifting-supply-chains and https://www.mckinsey.com/capabilities/operations/our-insights/tech-and-regionalization-bolster-supply-chains-but-complacency-looms] Read this as: most large companies have a screen; few can see deep suppliers. Survey definitions of "visibility" are the respondents' own. [Our view]
6. Worked example: more alerts is not better
All numbers [Illustrative], computed by script. A tower raises 200 candidate alerts a day; 20 (10%) are real problems. Each alert takes 5 minutes to check.
- Loose settings: catches 95% of real problems (19) but also flags 40% of the non-problems (72 false alarms). Only 21% of alerts are real. Checking takes about 7.6 hours a day. Misses 1 real problem.
- Tight settings: catches 70% of real problems (14) with 5% false alarms (9). 61% of alerts are real. Checking takes about 1.9 hours. Misses 6 real problems. Neither is "right." The choice depends on what a missed problem costs versus what a planner's hour costs. If one missed problem costs more than several hours of checking, loosen. If staff stop reading, tighten. Ask any vendor for these two numbers on your data: how many real problems it catches and how many alerts are false. [Our view]
7. What it costs and what it breaks
- Data plumbing: connecting systems is usually most of the cost and delay.
- Alert fatigue: see the example.
- Unowned alerts: a flag with no named responder changes nothing.
- False certainty: a clean screen only means no problem was recorded.
- Dependence on partners: tracking data from carriers and suppliers is only as timely as they make it.
- Vendor lock-in: keep your data exportable.
8. Your 90-day roadmap for AI in the supply chain
Using what this course covered. Pick the smallest step that proves value. [Our view]
- Days 1-30, see: list products, suppliers, lanes (Lesson 6 checklist). Track on-time delivery and stockouts in a plain sheet (Lessons 1-3). Write down unwritten rules (Lesson 5).
- Days 31-60, pilot one thing: choose one of: a better forecast for 20 top items (Lesson 2), a safety-stock review (Lesson 3), nearest-next-stop routing (Lesson 5), or an alert list for your top 10 suppliers (Lesson 6). Set a baseline first and a stop date.
- Days 61-90, decide: compare to baseline. Keep, change, or drop. Name the owner. Only then ask about software. Do not start with a control tower product. Start with the decisions it would speed up. [Our view]
9. Monday checklist
- Pick one of the four pilots and write its baseline number today.
- Name who acts when something is late.
- Decide what a missed problem costs you in dollars.
- Ask each vendor for catch rate and false alarm rate on your data.
- Write the stop date for the pilot.
10. Try it yourself (no code, 30 minutes)
Imagine your biggest supplier's shipment is late a week. Write the first three calls you would make, who makes them, and what data you would need to see first. Whatever you lacked is your first data project.
11. Self-check
- How many ships did Reuters say were waiting when the Suez Canal reopened? More than 400.
- What makes a control tower useful beyond the screen? Named owners and playbooks.
- What did McKinsey's 2022 survey find about third-tier supplier visibility? Only 17% said they had a good picture, up from 2%.
- In the example, what is the cost of loose alert settings? 7.6 hours of checking a day, 79% false alerts.
- What should come before buying software? A baseline and a named owner for one decision.
12. Where this is going
Better tools will draft the response as well as raise the flag, and may act for you inside limits you set. Whether they should is a question of trust, limits and who is accountable. Some people expect far more capable AI systems later; for supply chains the near-term constraints remain shared data and clear ownership. [Our view]
Sources
- Reuters, March 29, 2021: https://www.reuters.com/world/stranded-suez-canal-ship-re-floated-inch-cape-2021-03-29/
- Suez Canal Authority, March 29, 2021: https://www.suezcanal.gov.eg/English/MediaCenter/News/Pages/nav_29-03-2021.aspx
- McKinsey 2022: https://www.mckinsey.com/capabilities/operations/our-insights/taking-the-pulse-of-shifting-supply-chains
- McKinsey 2023: https://www.mckinsey.com/capabilities/operations/our-insights/tech-and-regionalization-bolster-supply-chains-but-complacency-looms
Review flags
- RESOLVED (Oct 7, 2026): both McKinsey articles were read at mckinsey.com. 2022 pulse survey (113 supply chain leaders, Mar-Apr 2022): 67% have digital dashboards for end-to-end visibility; 17% have a good picture to the third tier or beyond, up from 2% the year before; 71% expect to revise inventory policies. 2023 survey (101 respondents, Apr-May 2023): 79% have dashboards; 71% expect to revise planning processes. Figures are self-reported by executives in small samples, and the different sample sizes confirm the two years are not a panel. One correction to the ledger below: the 17% and 2% are in the 2022 pulse article, as stated in the next flag.
- The lesson states no cause or cost for the Ever Given grounding. ACCEPTED LIMITATION: not needed for any claim made. The "shippers who had visibility did better" point is labeled our inference, not sourced. McKinsey's 2022 survey separately says respondents with dashboards were twice as likely as others to have avoided supply chain challenges that year (self-reported).
- The alert example uses invented rates.
- The 90-day roadmap is our recommendation, untested.
- The 17% and 2% are both from the 2022 McKinsey article (2022 vs 2021). The 67% (2022) to 79% (2023) comparison uses two different respondent sets, not a panel; do not read it as the same companies improving.
Claim ledger (C4 Lesson 7)
| # | Claim | Source | Verified | Status | Re-check |
|---|---|---|---|---|---|
| 1 | Grounded Mar 23, refloated Mar 29, 2021; 400+ ships waiting; backlog 422 clearable in 3.5 days (Reuters quoting SCA chairman Rabie) | Reuters; SCA announcement for refloating | 2026-10-06 | Read (Reuters excerpt, SCA page) | none |
| 2 | 67% (2022), 79% (2023) have visibility dashboards; 17% third-tier (2022) vs 2% | McKinsey surveys | 2026-10-06 | Excerpt only; self-reported | Read full articles |
| 3 | 71% expect to revise planning within three years | McKinsey 2023 | 2026-10-06 | Excerpt only | Read full article |
| 4 | Alert triage arithmetic | Our computation | 2026-10-06 | Illustrative, script-checked | none |