Thought Leadership Series | Brief 3 of 7
Co-authored by Dr. Brian Harris · Shan Kumar · Dennis Sobotka · David Ciancio | September 2026
Occasion Architecture: let consumers reveal the demand before we build the decision tree.
If Occasion Management is going to organize strategy around consumer demand, the retailer cannot begin by deciding what the occasions are. The architecture has to be discovered from how consumers actually live, shop, and consume.
The Foundational Question
Breakfast. Dinner Tonight. Game Day. Health & Wellness. These labels are intuitive – which is precisely the risk. Retailers and CPG companies have decades of experience describing demand through categories, brands, and familiar consumption occasions. If we begin an Occasion Management process with a list of occasions already named, the research can easily validate what the industry already believes rather than reveal how consumers actually organize demand.
Occasion Architecture starts one step earlier. It asks: what recurring consumption purposes exist in consumers’ lives, how do consumers describe them, what context triggers them, and what needs must be satisfied to complete them?
The governing principle is simple: the occasion comes from the consumer, not from us.
Why Purchase Data Alone Cannot Define an Occasion
A transaction tells us what was bought. It does not reliably tell us why it was bought, who it was for, when or where it would be consumed, what the shopper wanted but could not find, or whether part of the same consumption need was fulfilled at another retailer.
The ambiguity is obvious at product level. Eggs can serve Breakfast, baking, Dinner Tonight, or household replenishment. A single basket can contain products intended for several different consumption purposes. Assigning the entire basket to one occasion therefore forces a precision the transaction does not contain.
Occasion discovery must combine observed behavior with stated purpose and consumption context. Purchase data becomes behavioral evidence – the ‘what’ – while the consumer’s own account supplies the ‘why.’ Neither is sufficient alone.
The Unit of Learning Is the Household Over Time
The architecture should not be inferred one basket at a time. Repeated observation of a household over approximately 4–6 weeks allows ambiguous purchases to resolve into recurring patterns. What appears random in one trip can become meaningful when the same household repeatedly solves a similar consumption need across stores, channels, and weeks.
That longitudinal view also exposes fragmentation. A household may buy the fresh components of an occasion at one retailer, pantry components elsewhere, and a missing item digitally. The occasion exists in the consumer’s life even when no single retailer sees the complete basket.
The household over time reveals the recurring demand. The individual basket is evidence – not the unit that defines the occasion.
The Discovery Sequence: Occasion First, Decision Tree Second
Occasion Architecture solves two different questions in sequence. First, Occasion Discovery identifies recurring consumption purposes and contexts. Only after an occasion has been identified and validated do we ask what needs must be satisfied within it and build its Shopper Decision Tree.
Consumer behavior → purpose & context → Consumption Occasion → Shopper Decision Tree → Occasion Need States → products
This sequencing matters because two different behaviors are involved. Complementarity helps reveal what belongs together to complete a consumption purpose. Substitution helps reveal the alternatives consumers consider within a particular need. Conflating the two risks forcing category logic onto an occasion that is inherently cross-category.
A Consumer-Led Occasion Architecture Study
A practical discovery study would recruit a behaviorally diverse consumer panel rather than screen people into occasions that have already been named. Household composition, life stage, shopping frequency, retailer and channel mix, digital usage, work and school routines, cooking behavior, and other contextual differences help ensure the study can reveal mainstream as well as emerging demand.
For roughly 4–6 weeks, participating households would capture purchases and intent close to the event. The exact technology can vary, but the research requirement is consistent: purchases and diary entries are joined at the trip level – not just captured over the same weeks – so the ‘what’ and the ‘why’ can be analyzed together rather than reconciled after the fact.
| Evidence | What It Captures | Why It Matters |
|---|---|---|
| Structured receipts / purchases | Product, category, price, quantity, retailer, and channel | Provides the behavioral record of what was actually bought across the household’s shopping ecosystem. |
| In-the-moment diary | Purpose, intended consumer, context, constraints, alternatives, and unmet needs | Captures the consumer’s ‘why’ without relying only on retrospective recall. |
| Consumption / usage | What was ultimately consumed, when, where, with whom, and with what substitutions | Connects shopping behavior to the consumption purpose the purchase was intended to serve. |
| Digital behavior | Search, browse, abandonment, substitution, recommendation response, and basket-building behavior | Adds increasingly granular evidence of intent, changing channels, and emerging needs. |
Let the Consumer Supply the Vocabulary
The questions should remain open enough for consumers to describe the demand in their own terms: What were you trying to accomplish? Why then? Who was it for? When and where would it be consumed? How much time or effort were you willing to put into it? What else belonged to the same purpose? What did you want but fail to find, and how did you solve the need instead?
From that language, the research can derive three layers of attributes rather than impose them in advance:
| Attribute Layer | What It Describes | Illustrative – Not Prescriptive |
|---|---|---|
| Occasion context | When, where, with whom, and under what circumstances consumption occurs | weekday/weekend; morning; work/school; solo/family/guests; preparation time |
| Functional need | What the consumer needs the occasion to accomplish | healthy; convenient; indulgent; portable; fresh; budget; premium; protein |
| Product expression | The features through which products can satisfy the need | fresh/frozen; organic; ready-to-eat; pack size; flavor; format; nutritional claim |
Product descriptions, claims, and package information can enrich and scale the attribute library later. They should not be the source of consumer need. Products enter downstream as expressions of demand, not as the starting point for defining it.
From Behavior to a Validated Occasion
The analysis joins two streams. Unsupervised analysis of structured purchases can surface products that repeatedly travel together, recurring combinations, and patterns of complementarity or substitution. Separately, AI-assisted analysis of diary language can extract purpose, context, unmet needs, and the vocabulary consumers use to describe them.
AI should function as a hypothesis engine, not a black-box authority. Purchase patterns show what travels together; consumer language helps explain why. Expert interpretation then joins those signals, and consumer validation tests whether the resulting occasion labels and boundaries actually reflect consumer meaning.
The output of the first discovery layer is a validated set of recurring Consumption Occasions – not yet a merchandising taxonomy and not yet a measurement scorecard.
Then Build the Shopper Decision Tree
Once an occasion is validated, the analysis moves inside it. The Shopper Decision Tree describes the needs a consumer must solve to complete that occasion across categories and channels. Under the SDT, we will call these branches Occasion Need States to distinguish them from need-state terminology within a category CDT.
The SDT is therefore structurally different from the traditional Consumer Decision Tree. A CDT primarily describes how shoppers choose and substitute among products within a category. The SDT describes how consumers assemble complementary needs across categories to fulfill an occasion. The two connect, but they answer different questions.
| Architecture | Question Answered | Primary Behavior |
|---|---|---|
| Shopper Decision Tree (SDT) | What complementary needs must be fulfilled to complete this Consumption Occasion? | Complementarity across needs and categories |
| Occasion Need State | What specific consumer requirement within the occasion are we trying to satisfy? | Need fulfillment across products, categories, and channels |
| Consumer Decision Tree (CDT) | Within a category or choice set, what alternatives will the shopper substitute among? | Substitutability / variety choice |
Category Architecture produces the Consumer Decision Tree – how a shopper navigates choice within a category. Occasion Architecture produces the Shopper Decision Tree – how a shopper assembles a cross-category basket to fulfill an occasion. Same rigor, one level up.
Core, Emerging, and Niche Occasion Need States
Not every Occasion Need State has the same strategic meaning. Core Occasion Need States define the baseline requirements a retailer must fulfill credibly to compete for the occasion. Emerging Occasion Need States may be smaller today but show changing consumer behavior and future growth. Niche Occasion Need States may matter deeply to specific households or customer segments and can become sources of loyalty or differentiation.
This is where digital engagement becomes especially important. Search language, AI-assistant requests, substitutions, abandoned baskets, and recommendation response can surface changing needs faster than traditional POS alone. The Consumption Occasion itself may remain stable while the way consumers solve it – channels, missions, products, and Occasion Need States – changes materially.
What This Study Still Needs to Prove
A methodology this depends on consumer behavior has to earn its conclusions honestly. Before an Occasion Architecture can be treated as settled, four questions need real answers:
- Receipt-extraction accuracy – can abbreviated, inconsistent receipt descriptions be matched reliably enough to identifiable products to support the analysis?
- Panelist compliance – can participants sustain accurate receipt and diary capture over 4–6 weeks without meaningful drop-off or fatigue?
- Architecture stability – do recurring occasions and Shopper Decision Tree branches emerge with enough consistency across different households to support validation, rather than each household producing an idiosyncratic pattern?
- Consumer-language quality – does the research capture language rich and specific enough to define real attributes, rather than collapsing back into labels the researchers already had in mind?
These aren’t reasons to defer the work. They’re the tests a rigorous study has to pass before its output becomes the foundation the rest of Occasion Management builds on.
What the Occasion Architecture Produces
A robust Occasion Architecture study should produce more than a list of occasion names. Its deliverables are the foundation on which the rest of Occasion Management depends:
- Occasion Architecture – a validated set discovered from consumer behavior rather than assumed in advance.
- Shopper Decision Tree – for each occasion, a validated tree and its Occasion Need States.
- Attribute library – a consumer-derived set spanning occasion context, functional needs, and product expressions.
- Documented linkage – from household behavior and stated purpose through occasions, Occasion Need States, attributes, and products.
- Validation protocol – a repeatable process that can periodically test whether the architecture has materially changed.
Foundational Architecture, Then Continuous Learning
The full Occasion Architecture should not need to be rebuilt every year. Once established and validated, it becomes foundational infrastructure for Occasion Management. The more frequent task is to monitor what is changing underneath it: channel blurring, digital shopping behavior, emerging Occasion Need States, new product expressions, and shifts in how households fulfill the same recurring need.
That suggests a two-speed cadence. Conduct deeper consumer revalidation periodically when evidence suggests structural change. In between, use digital behavior, consumer panels, retailer data, and CPG insight as continuous signals that may trigger targeted updates to an SDT or Occasion Need State without reopening the entire architecture.
CPG partners can contribute valuable consumer research, innovation pipelines, and category expertise to that learning process. But the retailer’s Occasion Architecture must remain consumer-led and common across suppliers; otherwise, each partner risks describing the occasion through the boundaries of its own portfolio.
Where Architecture Ends – and Measurement Begins
Occasion Architecture defines the common consumer language. It does not by itself determine Share of Occasion, competitive gaps, or retailer performance. Those are downstream measurement questions.
Once the occasions and Occasion Need States are validated, the same architecture can be applied to external panel data and retailer loyalty, POS, and digital data. That creates a common unit through which performance can be benchmarked against named retailers, diagnosed across internal competitive clusters, and ultimately rolled up to occasion performance.
Brief 3 defines the demand. Brief 4 asks the next question: once we know what the occasion and its Occasion Need States are, how do we measure whether a retailer is actually winning them?
The Principle to Protect
Occasion Management is intended to move retail planning closer to consumer demand. Occasion Architecture is what keeps that promise honest.
If we begin with retailer categories, predefined occasion labels, or manufacturer product attributes, we risk rebuilding today’s structure with new terminology. If we begin with repeated consumer behavior, stated purpose, and the consumer’s own language, we create the possibility of seeing demand the way the shopper sees it – including the emerging and unmet needs that today’s category structures may not yet reveal.
Discover the Demand Before You Decide How to Win It
The biggest risk in Occasion Management is starting with the answer already in mind. If retailers begin with today’s categories, familiar occasion labels, or manufacturer-defined attributes, they may simply recreate today’s view of demand under new terminology.
Start with the consumer. Discover the recurring needs, unmet demand and emerging Occasion Need States that today’s category structure may not yet reveal. The growth opportunity is not only in what shoppers buy from you today – it is in understanding the demand they are already solving somewhere else.
See the demand first. Then decide how to win it.
Next: Brief 4 – how to measure whether a retailer is actually winning the occasions this architecture reveals.← Brief 2: Occasion Management: The Framework in a NutshellSeries guideBrief 4: Occasion Measurement: Performance and Share →
Occasion Management is a proposed management discipline for strategically managing recurring consumer demand across traditional categories and functional boundaries. Like Category Management before it, Occasion Management is expected to be a journey – refined over time through practical application and learning. Intent AI welcomes collaboration with forward-thinking retailers and CPG partners to advance the discipline.
© 2026 Intent AI, Inc. All rights reserved.
This publication is provided for educational and informational purposes. It may be shared in its complete and unmodified form with attribution to Intent AI and the authors. No portion may be reproduced, modified, incorporated into another publication, or used for commercial purposes without prior written permission from Intent AI, Inc.
The Occasion Management framework, terminology, methodologies, measurement approaches, exhibits, and related intellectual property described in this publication are proprietary to Intent AI, Inc. unless otherwise noted.
For permissions, collaboration, or further information, contact [email protected].