Decision engine architecture: inputs, scoring, thresholds and outputs
The essentials in 30 seconds
- Decision engine architecture has 6 layers: inputs, scaling, weights, aggregation, thresholds and outputs.
- The EQOS Asia engine scores 9 dimensions on a φ(S) scale from 0 to 1 (EQOS Asia method page, as of October 2026).
- A STOP zone lies below 0.618 and an OK zone above it.
- Outputs include 3 priority risks, 3 trajectories and a 30/60/90 day plan.
- Sentinel re-measures the model monthly.
A decision engine is built from layers that each do one job: inputs, scaling, weights, aggregation, thresholds and outputs. Seeing the layers separately helps a buyer judge a tool and helps a team design its own. This guide describes each layer, shows how the EQOS Asia engine arranges them, and lists checks before relying on the result. It builds on the walk-through of how a situation turns into a score.
On this page
Compare the EQOS Asia offers and request a 30-minute pre-diagnostic.
What is the architecture of a decision engine?
The architecture is the arrangement of the parts that take a question and return a verdict. A decision engine has 6 layers: inputs, scaling, weights, aggregation, thresholds and outputs. Each layer receives the result of the previous layer and passes its own result on. The design keeps responsibilities separate, so a team can inspect one layer without reading the whole system.
A layered design also supports change. A new data source touches the input layer only. A revised policy zone touches the threshold layer only. Teams can then explain each change in one sentence and test it alone. The figure lists the 6 layers in order, from the first question to the delivered output.
How do inputs enter a decision engine architecture?
The input layer gathers the facts about the situation. In the EQOS Asia process these include internal data from finance, HR, operations, sales and product across entities and countries, external sources such as LinkedIn, ICIS, SimilarWeb and Crunchbase, and qualitative documents such as presentations, emails and meeting notes (EQOS Asia method page, as of October 2026). Several formats are accepted: PDF, video, audio and raw data.
Good input design dates and sources every item. It also protects the data. The EQOS Asia process signs an NDA before any exchange, works in an encrypted isolated environment, and hosts data in Singapore (EQOS Asia method and pricing pages, as of October 2026). Questionnaire answers are encrypted in the browser before they are sent. These controls belong to the architecture as much as the formulas do.
How does the scoring layer work inside a decision engine?
Scaling puts different inputs on one scale, for example 0 to 1, so a revenue figure and a trust rating can meet in one calculation. Weights state how much each dimension counts. Aggregation then combines the weighted dimensions into one score. These 3 steps form the core of the engine, and each one can be reviewed on its own.
The EQOS Asia engine measures 9 dimensions: coherence, integrity, resonance, care, trust, transparency, clarity, stability and energy (EQOS Asia method page, as of October 2026). Its 2008 back-test page explains that each dimension is raised to its weighting in a geometric product, not a simple average, so the lowest score drags the whole system down (EQOS Asia 2008 back-test page, as of October 2026). The expanded coefficients stay proprietary.
How do thresholds turn a decision engine score into an output?
A threshold turns a number into a zone. The EQOS Asia method places a STOP zone below 0.618 and an OK zone above it on the φ(S) scale from 0 to 1 (EQOS Asia method page, as of October 2026). A reader sees at once on which side a situation falls. The figure draws the scale with its cut-off at 0.618.
Thresholds belong in their own layer, because they express policy while the score expresses measurement. A team can keep the same score and move a cut-off when its risk appetite changes. Writing the cut-off, its reason and its owner in the record keeps the change transparent and lets reviewers compare a past verdict with the current rule.
Which outputs does a decision engine architecture deliver?
The output layer packages the result for a decision maker. In the Audit pillar the system returns a φ(S) score, identifies fragility zones and lists 3 priority risks. The Prediction pillar adds 3 trajectory scenarios and the dates of tipping thresholds, and the Modelling pillar delivers a complete map of the system (EQOS Asia method page, as of October 2026).
The deliverable is a personalised interactive HTML report with immediate decisions, deferrable decisions, a recommended sequence and a 30/60/90 day plan, presented to a Comex in person or by video call from Singapore (EQOS Asia method page, as of October 2026). Good outputs separate what to do now from what can wait, so leaders leave a meeting with an order of actions.
How does a decision engine architecture keep a record and stay current?
A record layer stores the inputs, the version of the calculation and the result. With a deterministic core a reviewer can recompute a past verdict from that record. The EQOS Asia method states that any data scientist with access to the Codex, under NDA, can recompute a verdict identically (EQOS Asia method page, as of October 2026).
A refresh layer keeps the reading current. Sentinel re-measures the model monthly, sends alerts when a threshold is approached, and runs quarterly review sessions (EQOS Asia method and pricing pages, as of October 2026). The architecture treats a decision as a living measurement. The record shows what changed between 2 readings, and the refresh shows when the next reading is due.
How do you check a decision engine architecture before relying on it?
Ask 5 questions of any engine. Which inputs does it take, and are they dated? How does it scale them? Where are the weights stated? Which thresholds apply, and who owns them? Can a second person recompute the result? An engine that answers all 5 gives a reviewer something to verify, and one that hides a layer deserves more questions.
The public simulator lets a reader rate the 9 dimensions and watch the result, and the documented method describes the layers behind it. A free 30-minute pre-diagnostic by video, scheduled in SGT, applies the EQOS Asia equation to a real case (EQOS Asia contact page, as of October 2026). This page is general information, not legal, tax or investment advice.
Frequently asked questions
What are the layers of a decision engine?
A decision engine has inputs, scaling, weights, aggregation, thresholds and outputs. Each layer takes the previous layer’s result and passes on its own, so a team can inspect or change one layer without rewriting the others. The EQOS Asia engine applies the same logic across 9 dimensions.
Where do thresholds sit in the architecture?
Thresholds sit after aggregation and turn the score into a zone. In the EQOS Asia method a STOP zone lies below 0.618 and an OK zone above it on the φ(S) scale from 0 to 1 (EQOS Asia method page, as of October 2026). A separate layer makes policy changes transparent.
How does the EQOS Asia engine protect inputs?
The process signs an NDA before any exchange, works in an encrypted isolated environment, hosts data in Singapore, and encrypts questionnaire answers in the browser before sending (EQOS Asia method and pricing pages, as of October 2026). The Mapping and Mission tiers add a reinforced NDA.
What does the EQOS Asia engine return?
It returns a φ(S) score between 0 and 1, fragility zones, 3 priority risks, 3 trajectory scenarios and a 30/60/90 day plan in an interactive HTML report (EQOS Asia method page, as of October 2026). A deterministic core allows a reviewer to recompute each verdict.
Test your knowledge
1. How many layers does the architecture have in this guide?
- 4
- 6
- 9
Show the answer
6.
2. Below which φ(S) value does the STOP zone begin?
- 0.618
- 0.300
- 0.900
Show the answer
0.618.
3. How does the 2008 back-test page describe the aggregation?
- A simple average
- A geometric product
- A random draw
Show the answer
A geometric product.
4. Which service re-measures the model monthly?
- Reading
- Mapping
- Sentinel
Show the answer
Sentinel.
5. Where does the EQOS Asia process host data?
- Singapore
- Paris
- Dubai
Show the answer
Singapore.
About the author
Stéphane Jambu
President, EQOS Forecast Asia
Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile
Sources: EQOS Asia method page (eqos-asia.com/method.html), pricing page (eqos-asia.com/pricing.html), 2008 back-test page (eqos-asia.com/case-studies/crise-2008.html), simulator page (eqos-asia.com/simulator.html) and contact page (eqos-asia.com/contact.html), consulted on 22 October 2026. Page updated on 22 October 2026.