"Who made that call?" — the week the client contact changes, the project restarts
An NT$1.8M order management system, five months in, and the client's project owner resigns. The new contact arrives with a requirements list of their own — eleven items the previous owner had explicitly rejected. Two weeks of clarification, one change order, NT$220,000 added, six weeks late. That is not a technical risk or a requirements risk. It is the risk of knowledge attached to a person.
Four industry myths
- Myth 1: "Meeting minutes protect us from turnover." Minutes carry conclusions without reasons, and a conclusion with no reason gets overturned.
- Myth 2: "The SOW locks the scope." An SOW locks deliverables, not interpretation: "supports multiple warehouses" has two readings.
- Myth 3: "Turnover is the client's problem." Rebuilding consensus eats 3–6% of total project hours, absorbed by the vendor.
- Myth 4: "A few more meetings will catch them up." Meetings cannot rebuild context. A searchable decision log can.
The core framework: a decision log
Record every decision that affects scope, cost or schedule in six columns:
- ID (D-001) — so it can be cited
- Decision — one sentence, verb first
- Reason — finance requires manual verification of each entry
- Rejected alternatives — and why they lost
- Decision maker and date — a named person, client side
- Impact — saves 12 dev hours, adds 6 manual hours a month
Rules: update it the day of the meeting, cite IDs in every estimate, hand it to a new contact first. A five-month project runs 30–60 entries at 20 minutes a week.
Contact turnover risk score (0–12)
Score it at kickoff:
- Owner is an employee 0 / contractor or consultant 2
- Holds decision authority 0 / must escalate 2
- Under 3 months 0 / 3–6 months 1 / over 6 months 2
- Backup contact 0 / single point 2
- Client industry turnover low 0 / high 2
- Independent of one executive's preference 0 / dependent 2
0–3: normal minutes. 4–7: decision log and dual contacts. 8+: a contact-change clause in the contract and a 3–5% communication buffer.
Three companies, three conclusions
- 20-person trading company, owner is the contact: decisions arrive verbally over LINE, so write them back as text and get a confirmation.
- 120-person manufacturer, IT manager is the contact: typically 6–8, expertise but no budget authority, so map the approval levels.
- 500-person service chain, contact is an external consultant: highest risk, so add a parallel internal contact to the meetings.
Hidden cost breakdown
Six months, NT$1.8M, one contact change:
- Re-explanation and context transfer: 16–24 hours
- Requirements re-confirmation and disputes: 20–40 hours
- Rework on completed items: 30–80 hours
- Idle resources 2–6 weeks, a real loss if nobody is redeployed
- Client side: the new contact needs 40–80 hours
- Trust reset: one or two extra review rounds
- Total: roughly 8–15% of project value, mostly absorbed by the vendor
Vendor scorecard: turnover resilience
Ten dimensions, 0–3 each:
- A searchable decision log, not just minutes
- Estimates cite decision or requirement IDs
- Documents stored where the client can reach them
- A second engineer knows this project
- Build instructions that actually run
- Status reports legible without narration
- Requirement changes go through written orders
- Proactive flags: "this conflicts with D-017"
- Handover package defined in the contract
- Free re-onboarding when the contact changes
24+ is healthy; 15–23, close the gaps next phase; below 15, one turnover costs 8–15%.
How ScriptWalker works, and who we are wrong for
Three engagement models:
- Project: the log and requirement IDs are standard deliverables; one two-hour re-onboarding when the contact changes.
- Retainer: monthly reports legible without narration, plus a rolling current-state document.
- Advisory: in-house developers but no process discipline — we install the process, development stays internal.
We are the wrong fit when:
- The client refuses any written confirmation
- Nobody will put a name on a decision, so disputes land on the vendor
- The budget covers development hours only — protection gets cut, time gets spent
Kickoff and handover playbook
- Month 1: week one, the decision log and ID convention; week two, primary and backup contacts plus the approval matrix; week four, the one-page current state.
- Turnover week (days 0–7): pause new requirements, send the log and the one-pager, hold a two-hour walkthrough — goals, ten key decisions, progress, open items.
- Days 8–21: reopening is allowed, but only through a change order with cost and schedule impact.
- Days 22–30: re-confirm acceptance criteria and sign a current-state agreement.
- Day 90 review: put the delay and added hours into the retrospective and the next quote.
Decision checklist
- ☐ Project longer than four months?
- ☐ Only one point of contact?
- ☐ Contact is a contractor or consultant?
- ☐ Major decisions escalate?
- ☐ Ever had a "we agreed not to" dispute?
- ☐ Do minutes record why?
- ☐ Estimates traceable to decision IDs?
- ☐ Documents reachable by the client?
- ☐ A second person on each side?
- ☐ Contract addresses contact changes?
- ☐ Current-state doc readable in half a day?
- ☐ New requirements via written change orders?
- ☐ Communication buffer in the quote?
Eight or more checks: build the log at the next meeting. Two hours to set up; one turnover costs 8–15% of project value.
FAQ
How is a decision log different from meeting minutes?
Minutes are chronological and record process; a log is indexed by ID and records only conclusions and reasons. If a reader still asks "why," the minutes are not enough.
What if the client will not help maintain documents?
The vendor writes, the client confirms. Send a decision summary within 24 hours; no objection within three business days is confirmation.
Should we charge when a new contact reopens a decision?
If they never knew the reason, re-explaining is free. If they know it and still change direction, that is a requirements change: change order, priced.
What about turnover on the vendor's side?
Same logic in reverse: a second engineer picks the project up in half a day, with build instructions that run and environment variable templates.
Do small projects need all this?
Not the full apparatus. Under three months and NT$200,000, one shared document recording "decision plus reason," ten minutes a week.
Next step
If your contact just changed mid-project, we will spend 30 minutes sorting which decisions have no recorded reason, which deserve re-discussion, and which should be met by citing the original agreement. No fee.
- Email: [email protected]
- Phone: 0916-224-047
- LINE: @ufv9089p