Golden Ratio Analysis
Golden Ratio Analysis is a diagnostic technique. When a situation seems out of balance, the team compares some of its proportions with the golden ratio, which is about 1.618. For example, it examines the makeup of the team, the mix of test cases or the spread of risks in the register. It names the proportion that departs from the ratio and decides whether to correct it. George W. Anderson describes the technique in Design Thinking for Tech (2023).
Purpose
Golden Ratio Analysis compares the proportions of a situation with the golden ratio to find the one that is out of line. Anderson uses it to explain why a product, an interface or a situation looks badly proportioned. A team senses that a project is out of balance without knowing where. The technique gives it a list of proportions to check and a numeric benchmark for each.
The technique settles one question: which proportion does the team correct? It produces a gap list: for each dimension examined, the two quantities compared, their ratio, the explanation of the gap and the decision.
Usage
When to use it
- Imbalance perceived after a first analysis: re-examine the situation in terms of proportions after a cause analysis or a brainstorming session.
- Scope or quality review: look for the part of the project that weighs too much or too little.
- Test plan under review: compare happy-path cases and error cases.
- Team whose makeup changes: compare delivery and coordination, experienced and new people.
- Screen that users find unbalanced: compare the widths of the main and secondary areas.
When not to use it
- The team must justify a decision to a committee: compare the options with a decision matrix.
- The team must establish the cause of a known problem: the ratio flags a proportion, not a cause; trace the chain with the five whys.
Description
The golden ratio
In the Fibonacci sequence, each term is the sum of the two before it: 1, 1, 2, 3, 5, 8, 13, 21, 34. The ratio between two neighbouring terms tends towards the golden ratio. Its value is (1 + √5) / 2 = 1.6180339… Anderson rounds this ratio to 1.6 to 1.
Anderson sees the Fibonacci sequence in the growth of flowers, seashells and galaxies. He concludes that the golden ratio can be a benchmark for the proportions of a team or a project. The calculation is exact. Ron Knott (University of Surrey) points out that not all plants show the Fibonacci numbers.
What the book gives
Anderson's scenario is leading a community from point A to point B. A first review leaves an impression of imbalance. Anderson names nine dimensions in which to look for the imbalance. He proposes checking whether each one follows the golden ratio or departs from it. If something seems wrong, the fix may lie in rebalancing to this ratio. Elsewhere in the book, Anderson also cites the technique for judging the fit of a project's scope and quality.
The pairs of quantities, the tolerance band, the steps and the example that follow are our proposal.
The dimensions
Each dimension compares two quantities measured in the same unit. The team may replace a pair with another if it writes the new pair down before counting.
| Dimension (Anderson) | Quantities compared | Where to count them |
|---|---|---|
| Dimensions of the journey | Durations of preparation and deployment | Project schedule |
| Makeup of the staffing model (internal, external) | Internal and external people | Resource plan |
| Sizing of the sprint work items | Small and large items, by the team's threshold | Backlog tracking tool |
| Structure of the team | People who deliver and people who coordinate | Project organisation chart |
| Look and feel of the prototype or interface | Widths of the main and secondary areas of a screen | Mock-up |
| Communications cadence | Recurring meetings and written communications per month | Communication plan |
| Profiles of the team (diversity or experience) | People experienced in the domain and people new to it | Organisation chart, without names |
| Mix of test cases | Happy-path cases and error or boundary cases | Test plan |
| Progression or growth in the risk register | Risks with a response under way and risks with no response decided | Risk register at a review |
For profiles, the proposed pair concerns experience. A pair on diversity can touch on ethnic origin or health. These are sensitive personal data (art. 5 let. c no. 2 FADP): the team does not count them.
The steps
- Describe the perceived imbalance. The team states in one sentence what seems out of balance and where it observes it. "User acceptance testing (UAT) finds input errors that no test case covers" lends itself to the analysis. "The project is not running smoothly" does not yet.
- Select the dimensions. The team keeps the dimensions that this imbalance affects.
- Name the quantities. For each dimension kept, the team writes down the two quantities compared and their unit. It notes which of the two it expects to be the larger.
- Count. The figures come from the project's tools. A ratio estimated from memory only puts a number on the initial impression.
- Calculate the ratio. The team divides the larger quantity by the smaller. The ratio is therefore always 1 or more.
- Read the ratio. The team compares it with the band from 1.5 to 2.
- Explain each gap. The team looks for a known reason that justifies the proportion: a constraint, a choice, a project phase.
- Find the cause. For a gap with no known reason, the team looks for the cause with the five whys.
- Correct the cause. The team decides how to correct the cause it found. It names an owner and sets the date of the next count.
- Journey
- Staffing model
- Sprint work item size
- Team structure
- Prototype or interface
- Communications cadence
- Team profiles
- Test cases
- Risk register
Reading the ratio
From 2/1 on, the ratios between neighbouring terms of the sequence alternate around 1.618: 2, 1.5, 1.667, 1.6, 1.625, 1.615. They all stay between 1.5 and 2. The tolerance band covers this interval (bounds included). This band is a convention.
A ratio inside the band gives no signal. As the first pitfall explains, it does not prove that the proportion is right. A ratio outside the band is a gap.
The ratio does not say which of the two quantities should be the larger. Eight internal people to five external ones give 1.6. Five internal to eight external also give 1.6. The team therefore writes down the expected order before counting. A ratio inside the band whose order is reversed is also a gap.
A bolt-on technique
Anderson classes Golden Ratio Analysis among the bolt-on techniques that complement another method. The team runs it after a root cause analysis or a brainstorming session.
The pitfalls
Treating the ratio as a target. The ratio asks a question. The project's data answer it.
Correcting to hit the number. A team that adds test cases to reach 1.6 chooses its cases by a number. The correction targets the problem that the gap shows. The ratio follows.
Comparing quantities in different units. People divided by hours give a meaningless number.
Trusting a ratio drawn from small numbers. In a team of thirteen people, only the 8/5 split falls inside the band. The splits 7/6 and 9/4 give 1.17 and 2.25: a single person tips the reading. The team moves one unit from one quantity to the other and recalculates. If the reading changes, the ratio says nothing.
Confusing it with the relative estimation scale. Story points follow a modified Fibonacci sequence that spaces out the values. This scale compares no proportion with the golden ratio.
AI considerations
A language model reads the exports of the backlog, the test plan and the risk register. It classifies test cases as happy-path or error cases from their titles. It classifies backlog items as small or large by the team's threshold. The spreadsheet counts and divides. The team checks the classification on a sample. For a dimension with no obvious pair, the model proposes several pairs and justifies each one. The team chooses. For each gap, the model lists explanations that the team verifies.
Asked about the golden ratio, a language model often repeats the legends of the Parthenon or of harmonious faces. A justification of this kind is not proof. The model also cannot tell whether a gap is intended. Only the team knows the constraints and the choices that explain it.
Even when limited to experience, the team's profiles are personal data (art. 5 let. a FADP). The team sends an external AI service only totals, with no names or roles that identify a single person.
Examples
A team of thirteen people is migrating the billing system of a health insurer in the canton of Vaud. It perceives an imbalance: user acceptance testing (UAT) finds input errors that no test case covers.
| Dimension | Quantity expected to be larger | Quantity expected to be smaller | Ratio | Reading | Decision |
|---|---|---|---|---|---|
| Structure of the team | 8 people who deliver | 5 people who coordinate | 1.6 | Inside the band. 7/6 and 9/4 fall outside it: no signal | None |
| Profiles of the team | 10 people with more than five years on the old system | 3 people who joined this year | 3.3 | Gap justified: the migration requires knowledge of the old system | Kept |
| Mix of test cases | 120 happy-path cases | 15 input-error cases | 8.0 | Gap with no known reason. Cause: UAT found 52 types of input error with no test case | Write one case per type. Ratio after the addition: 120 / 67 = 1.8. Owner: the test lead. Next count in May |
| Progression of the risk register | 24 risks with a response under way | 18 risks with no response decided | 1.3 | Gap justified: the load test has just opened these 18 risks | Kept. Next count in May |
Visualisations
The diagram in the description follows one dimension from the perceived imbalance to the decision. The dimensions table gives a pair of quantities for each of the nine. The band diagram shows where 1.618 comes from and why the tolerance runs from 1.5 to 2. The example figure places the migration team's four ratios on this band.
Cost
| Dimension | Level | Rationale |
|---|---|---|
| Preparation | Low | On the first pass, the team spends more time writing the pairs and the expected order than extracting the counts from the project's tools. |
| Execution | Low | A one-hour session to read the ratios and explain the gaps. |
| Documentation | Low | One line per dimension in the gap list. |
Tools
A shared spreadsheet is enough for a team on site or remote. Each row carries a dimension, its two quantities and a formula that divides one by the other. Conditional formatting colours the ratios outside the band. The figures come from the tracking tools (Jira, Azure DevOps), the test management tool and the risk register. For the look of an interface, the mock-up tool gives the width of the areas. Anderson files the technique among the visual-collaboration exercises. The team uses a physical or virtual whiteboard (Miro, Mural) in the session where it explains the gaps.
Sources
- George W. Anderson, Design Thinking for Tech: Solving Problems and Realizing Value in 24 Hours, Addison-Wesley, 2023, ISBN 978-0-13-793303-7. Hour 12 "Exercises for Increasing Creativity", section "Design Thinking in Action: Golden Ratio Analysis", pp. 177-178: the ratio, the scenario from point A to point B, the nine dimensions and the rebalancing. Same hour, introduction: the technique as a bolt-on to other methods. Hour 5 "Visible and Visual Teamwork", section "Design Thinking in Action: Exercises for Visual Collaboration": the technique among the visual-collaboration exercises. Hour 24 "Design Thinking for Project Velocity": the technique for judging the fit of scope and quality. Appendix A, answers for Hour 12: the ratio as a measuring stick. Appendix B "Summary of Design Thinking Techniques and Exercises": the one-sentence definition.
- Ron Knott, "Fibonacci and Golden Ratio Formulae", University of Surrey: the value of the golden ratio and the convergence of the ratio between neighbouring terms.
- Ron Knott, "Fibonacci Numbers and Nature", University of Surrey: the plants that do not show the Fibonacci numbers.
- Ron Knott, "The Golden section ratio: Phi", University of Surrey: the absence of plans of the Parthenon and the studies that dispute the use of the golden ratio in art and architecture.
The pairs of quantities, the tolerance band, the steps, the pitfalls and the example are our proposals.

