agileA manufacturing platform startup running Scrum with a team of six to fifteen people has a definition of done problem. The company operates a platform that connects small manufacturers with raw material suppliers. It handles supplier discovery, price comparison, order management, logistics tracking, and quality certification. The company has been around for two years with thirty one employees. Nine people work in product development on a single Scrum team. (1/34)
The definition of done is weak. It is a checklist with five items. Code is written. Code is reviewed. Unit tests pass. Feature is deployed to staging. Product owner accepts the feature. The five items get checked and the feature is declared done. But the feature is not really done. (2/34)
Last sprint, the team built a new supplier rating feature. It let manufacturers rate suppliers on quality, delivery speed, and communication. The feature was deployed to staging. The product owner accepted it. All five items were checked. The feature was released. (3/34)
The release caused a problem. The rating data was not backed up. The backup process was not part of the definition of done. The data was stored in a database on a single server. The server failed. The failure caused data loss. Four hundred and twelve manufacturers lost six months of rating history. (4/34)
Those manufacturers were angry. Support tickets increased by three hundred percent. The support team was overwhelmed and could not respond. The lack of response caused churn. Eight percent of affected manufacturers left. That was twenty six manufacturers generating forty one thousand dollars per month. That revenue is gone.
The definition of done must be meaningful. (5/34)
Li Ka-shing built CK Hutchison on the diversified investment strategy. The model was simple. Li realized that the biggest risk in business is concentration. If all your money is in one investment, one failure destroys everything. He attacked the concentration by diversifying across industries, geographies, and asset classes. The diversification reduced risk and protected the portfolio, which enabled growth. (6/34)
Li applied the same thinking to every investment decision. He did not ask one question. He asked five. Is the revenue stream diversified. Is the customer base diversified. Is the supply chain diversified. Is the geography diversified. Is the asset class diversified. The five questions ensured that no single point of failure could destroy the investment. (7/34)
For a manufacturing platform startup, the definition of done problem is the same. The definition of done is concentrated on five items that do not cover everything. The lack of coverage creates single points of failure. Those failures cause data loss. Data loss causes churn. Churn destroys revenue.
Li's diversified investment strategy says: diversify the definition of done. That eliminates single points of failure. That protects the product. That protects the revenue. (8/34)
The Core Principle
Li's diversified investment strategy was built on a simple insight. The best way to create a meaningful definition of done is to cover multiple dimensions of completeness instead of concentrating on a single dimension like code quality. Li did not evaluate an investment by asking one question. He asked five. The five questions covered five dimensions. Those dimensions ensured that no single point of failure could destroy the investment. (9/34)
For a manufacturing platform startup, the definition of done covers one dimension. That dimension is code quality. It does not cover data integrity, security, performance, documentation, or support readiness. The lack of coverage creates single points of failure.
Li's diversified investment strategy says: cover multiple dimensions. Multiple dimensions eliminate single points of failure. That protects the product.
Four Steps to Apply the Diversified Investment Strategy (10/34)
1. Audit the Current Definition of Done and Identify the Dimensions It Misses
Li audited every investment at CK Hutchison. The audit identified missing dimensions. That created a gap analysis. The gap analysis drove improvement. Improvement reduced risk. Reduced risk protected the portfolio. (11/34)
You should audit your current definition of done the same way. For a manufacturing platform startup, the audit might look like this. The scrum master leads a definition of done audit with all nine team members. The audit takes one hour. It reviews the current five item checklist against six dimensions. (12/34)
Dimension one is code quality. Covered. Code is written, reviewed, and unit tests pass. Dimension two is data integrity. Not covered. No data backup, validation, or migration items exist. Dimension three is security. Not covered. No security testing, vulnerability scanning, or access control verification. Dimension four is performance. Not covered. No load testing, response time measurement, or resource utilization checks. Dimension five is documentation. Not covered (13/34)
. No user documentation, API documentation, or runbook updates. Dimension six is support readiness. Not covered. No support team notification, FAQ updates, or monitoring dashboard configuration.
The audit reveals that the current definition of done covers one of six dimensions. The five missing dimensions are single points of failure. Those failures caused the data loss. The data loss caused the churn. The churn destroyed the revenue. (14/34)
For a Scrum team of six to fifteen, the audit should happen in one session of no more than one hour. It should evaluate the current definition of done against at least six dimensions. The audit should be part of the sprint retrospective.
2. Add One Checklist Item for Each Missing Dimension Based on Past Failures (15/34)
Li added one safeguard for each missing dimension in every investment. Each safeguard was specific and based on past failures. The past failures informed the safeguards. The safeguards prevented future failures. That protected the portfolio.
You should add one checklist item for each missing dimension based on past failures. For a manufacturing platform startup, the additions might look like this. (16/34)
Past failure one was data loss due to missing backup. The added item is: Data backup is configured and tested for all new data stores. Past failure two was a security gap due to missing vulnerability scans. The added item is: Vulnerability scan is completed and all critical findings are resolved. Past failure three was performance degradation due to missing load tests. The added item is: Load test is completed and response time is under two hundred milliseconds at peak load (17/34)
. Past failure four was user confusion due to missing documentation. The added item is: User documentation is updated and published to the help center. Past failure five was support team overload due to missing notification. The added item is: Support team is notified and FAQ is updated for all user facing changes. (18/34)
The new definition of done has ten items covering six dimensions. Code quality has three items. Data integrity, security, performance, documentation, and support readiness each have one. Product owner acceptance is one. The definition of done is now diversified. (19/34)
Last sprint, the team built a new order tracking feature. The evaluation against the new definition of done revealed that data backup was not configured. That was caught before release. The backup was configured. Data loss was prevented. Churn was prevented. Revenue was protected. (20/34)
For a Scrum team of six to fifteen, add one checklist item for each missing dimension based on past failures. The new definition of done should cover at least six dimensions. The addition should be part of the sprint retrospective.
3. Weight Each Dimension by Business Impact
Li weighted each investment dimension by business impact. The weighting ensured that the most critical dimensions received the most attention. That reduced the most risk and protected the portfolio. (21/34)
You should weight each dimension by business impact so the team prioritizes the most critical items. For a manufacturing platform startup, the weighting might look like this. The product manager weights each dimension on a scale of one to five. Five is highest impact. (22/34)
Code quality is a three. It affects maintainability and long term cost. Data integrity is a five. It affects customer trust, churn, and revenue. Last sprint's data loss caused eight percent churn and cost forty one thousand dollars per month. Security is a five. A security breach would cause more churn than data loss. Performance is a four. It affects user experience, retention, and revenue. Documentation is a two. It affects onboarding and activation. Support readiness is a three (23/34)