Intent-to-Income™: A Reference Decision Architecture for AI-first Decision Intelligence
This page presents the published research record for Intent-to-Income™, a proprietary Reference Decision Architecture designed by Kittisak Pannutiyarak and developed further within HaNonn for AI-first Decision Intelligence in Customer and Commerce Decisions.
This page provides an English summary and citation record for the published paper. The complete manuscript, references, and versioned repository record are available through the DOI.
Official Abstract
Abstract
AI-mediated search, commerce, and decision-support systems are often evaluated through behavioral or business outcomes such as interaction, engagement, and conversion, which do not necessarily represent the quality of the underlying user decision process.
This paper proposes Intent-to-Income, a proprietary Reference Decision Architecture for AI-first Decision Intelligence focused on Customer Decisions and Commerce Decisions. The architecture defines a canonical core flow—Decision Signals → Decision Need → Decision Support → Decision Quality → Measurable Outcomes—and treats Decision Context as a cross-cutting layer.
It is organized around four core components: Decision Journey Mapping, Decision Signal Matrix, Decision Interface, and Measurement Logic. Decision Quality is explicitly separated from terminal outcomes and is represented through three proposed dimensions: Understanding, Confidence, and Readiness.
These dimensions are conceptual and have not yet been empirically validated. The paper also distinguishes Intent-to-Income from sales funnels, SEO frameworks, conversion-rate optimization frameworks, revenue attribution models, and product recommendation algorithms, while abstracting proprietary implementation details such as internal weighting and evaluation rules.
The contribution is a decision-centered architecture that organizes Decision Signals, Decision Need, Decision Support, Decision Quality, and Measurable Outcomes into a connected structure for AI-mediated Customer and Commerce Decisions. Limitations and directions for empirical validation, measurement development, comparative evaluation, and human oversight are identified.
Decision Intelligence; AI-assisted Decision Support; Human–AI Decision-making; Decision Quality; Customer Decisions; Commerce Decisions; Reference Architecture
Plain-Language Summary
What the Paper Proposes
The paper begins with a simple question: if a user clicks, converts, or completes an action, does that mean the underlying decision was well-informed? Not necessarily.
Research Problem & Gap
The Gap between Signals, Decision Support, and Decision Quality
Decision Support Systems, Decision Intelligence, Consumer Decision Support, and Recommender Systems already have substantial research foundations. However, they often examine specific systems, contexts, or measurement problems separately.
The paper therefore addresses a narrower architectural question: how can Decision Signals, Decision Need, Decision Support, Decision Quality, and Outcomes be explicitly separated and connected within one reference structure for AI-mediated Customer and Commerce Decisions?
Research Context & Supporting Literature ↓
- Decision Intelligence:
Moser, Rengarajan, and Narayanamurthy conceptualize a framework for aligning decision-related intelligence requirements with intelligence-processing capacities.
View source - Decision Intelligence and Data Science:
Pratt, Bisson, and Warin propose an integration framework describing how Data Science can inform components of strategic decision-making.
View source - Consumer Decision Support:
Westerman and colleagues study consumer decision support in internet and in-store settings.
View source - Decision Quality Measures:
Aksoy, Cooil, and Lurie review Decision Quality measures used in recommendation-agent research.
View source
Conceptual Contributions
Four Contributions at the Architecture Level
The paper’s contribution is the organization of established and proposed decision-related constructs within a connected architecture whose layers can be studied, designed, and evaluated separately.
These are architectural propositions, not evidence that the relationships among the layers or the proposed Decision Quality dimensions have been empirically validated.
Architecture Model
Architecture at a Glance
Intent-to-Income™ defines a five-stage canonical Core Flow, treats Decision Context as a cross-cutting layer, and organizes its design around four Core Architecture Components.
The figure represents architectural relationships; it does not establish causality, intervention effectiveness, or measurement validity.
For the complete Core Flow, components, and application scope, see the Intent-to-Income™ Reference Decision Architecture.
Illustrative Application
Knowledge-led Commerce: A Hypothetical Application
The paper uses Knowledge-led Commerce as a domain label for Commerce Decisions that require users to examine information, evaluate evidence, compare alternatives, and understand trade-offs. The following hypothetical scenario applies the architecture to an AI Decision Application for technically complex products with conflicting claims.
This scenario illustrates relationships within the architecture. It is not an End-to-end Case, empirical evaluation, or evidence of causality, and it does not disclose implementation-specific inference or Measurement Logic.
Measurement & Evaluation
Different Evidence Levels Answer Different Questions
Measurement Logic separates evidence about the decision process, candidate indicators of Decision Quality, and downstream outcomes. This prevents interaction volume or business results from being treated as direct measures of Decision Quality.
| Evidence Level | What It Helps Explain | Examples | What It Cannot Establish |
|---|---|---|---|
| Process Signals | How Decision Support is used and where friction occurs | Reviewing evidence, using comparisons, revisiting information, or requesting clarification | More interaction does not establish higher Decision Quality |
| Decision Quality Signals | Candidate evidence relevant to Understanding, Confidence, and Readiness | Assessment responses, evidence-use patterns, calibrated confidence reports, or context-specific indicators | No isolated signal validates a construct or establishes Decision Quality |
| Measurable Outcomes | What occurs downstream of the decision process | Qualified Next Step, Qualified Lead, Assisted Conversion, reduced friction, or a better-fit action | A downstream outcome alone does not establish that Decision Quality was high |
Measurement Guardrail
The paper proposes Measurement Logic for future investigation, not a validated measurement instrument. Construct validity, reliability, operational definitions, and relationships among the evidence levels still require empirical validation.
Research Boundaries
Limitations and Validation Status
Intent-to-Income™ is presented as a conceptual architecture. The following limitations define how its contribution should be interpreted and what remains to be tested.
01
No Empirical Validation Yet
02
Proprietary Implementation Logic Is Abstracted
03
The Architecture Has a Defined Scope
04
Generalizability Remains an Open Question
Discussion & Future Research
Future Research Agenda
Future research should test whether the constructs and relationships proposed by Intent-to-Income™ can be reliably defined, measured, and evaluated across different Decision Contexts.
| Research Track | Key Question | What Must Be Examined |
|---|---|---|
| Decision Quality Validation | Can Understanding, Confidence, and Readiness be defined and measured as distinct but related dimensions? | Construct validity, reliability, operational definitions, and relationships with satisfaction, post-decision regret, and long-term decision utility |
| Measurement Development | How can Understanding and contextual Confidence be assessed without relying only on self-reports or adding excessive cognitive burden? | Behavioral, interaction, assessment, and evidence-use measures combined with validated instruments where appropriate |
| Human Oversight & Decision Authority | How should Decision Interfaces communicate uncertainty, evidence limitations, alternatives, and unresolved trade-offs? | Appropriate reliance, over-reliance, under-reliance, and recognition of insufficient evidence |
| Architecture-Level Evaluation | Does the architecture add value beyond conversion-centric or recommendation-centric approaches? | Comparative evaluation of decision processes, indicator calibration, evidence understanding, readiness, and downstream outcomes |
High-Risk Application Boundary
The Health Decision Model (HDM) is an exploratory application domain within HaNonn. Applying the architecture to health-related decisions would require independent validation, safety evaluation, and stronger human oversight. It should not be interpreted as a diagnostic, treatment, or autonomous decision-making system.
The immediate priority is to establish operational definitions and evaluation protocols before comparing effectiveness or extending claims to other domains.
Conclusion
Research Significance
In AI-mediated environments, systems that support human judgment should be distinguished from systems optimized primarily for interaction, recommendation, or commercial outcomes. Intent-to-Income™ addresses this architectural concern by separating Decision Need and Decision Quality from Signals, Support, and downstream Outcomes.
The paper’s current value is its conceptual foundation and semantic boundaries, which can support critique, measurement development, and empirical research. It does not claim that the architecture or its proposed Decision Quality model has already been validated.
Intent-to-Income™ provides a foundation for studying and designing AI Decision Applications that help people evaluate evidence, alternatives, and appropriate next steps while preserving human Decision Authority.
Publication Record
How to Cite and Access the Paper
Use the Zenodo publication record and DOI when citing the paper, its contribution, or the proposed architecture.
https://doi.org/10.5281/zenodo.22151798
BibTeX Citation ↓
@misc{pannutiyarak2026intent,
author = {Kittisak Pannutiyarak},
title = {Intent-to-Income: A Reference Decision Architecture for AI-first Decision Intelligence},
year = {2026},
publisher = {Zenodo},
version = {1.0},
doi = {10.5281/zenodo.22151798},
url = {https://doi.org/10.5281/zenodo.22151798}
}
Citation Note
Use the DOI when citing the published paper. Use
this page URL
when referring specifically to the web summary or supporting explanations published by HaNonn.
Scholarly Sources
References
The following scholarly sources are cited in the published manuscript.
View full reference list ↓
- Aksoy, L., Cooil, B., & Lurie, N. H. (2011). Decision quality measures in recommendation agents research. Journal of Interactive Marketing, 25(2), 110–122.
https://doi.org/10.1016/j.intmar.2011.01.001 - Broder, A. (2002). A taxonomy of web search. ACM SIGIR Forum, 36(2), 3–10.
https://doi.org/10.1145/792550.792552 - Lee, J. D., & See, K. A. (2004). Trust in automation: Designing for appropriate reliance. Human Factors, 46(1), 50–80.
https://doi.org/10.1518/hfes.46.1.50_30392 - Moser, R., Rengarajan, S., & Narayanamurthy, G. (2021). Decision intelligence: Creating a fit between intelligence requirements and intelligence processing capacities. IIM Kozhikode Society & Management Review, 10(2), 160–177.
https://doi.org/10.1177/22779752211017386 - Pirolli, P., & Card, S. K. (1999). Information foraging. Psychological Review, 106(4), 643–675.
https://doi.org/10.1037/0033-295X.106.4.643 - Power, D. J. (2002). Decision support systems: Concepts and resources for managers. Quorum Books.
- Pratt, L., Bisson, C., & Warin, T. (2023). Bringing advanced technology to strategic decision-making: The Decision Intelligence/Data Science (DI/DS) Integration framework. Futures, 152, 103217.
https://doi.org/10.1016/j.futures.2023.103217 - Rahinel, R., Otto, A. S., Grossman, D. M., & Clarkson, J. J. (2021). Exposure to brands makes preferential decisions easier. Journal of Consumer Research, 48(4), 541–561.
https://doi.org/10.1093/jcr/ucab025 - Schemmer, M., Hemmer, P., Kühl, N., Benz, C., & Satzger, G. (2022). Should I follow AI-based advice? Measuring appropriate reliance in human–AI decision-making. CHI Conference on Human Factors in Computing Systems (CHI ’22), Workshop on Trust and Reliance in AI-Human Teams (trAIt).
https://doi.org/10.48550/arXiv.2204.06916 - Simon, H. A. (1977). The new science of management decision (Rev. ed.). Prentice-Hall.
- Westerman, S. J., Tuck, G. C., Booth, S. A., & Khakzar, K. (2007). Consumer decision support systems: Internet versus in-store application. Computers in Human Behavior, 23(6), 2928–2944.
https://doi.org/10.1016/j.chb.2006.06.006