Introduction: Why most “customer understanding” is incomplete
Most CPG organizations believe they understand their consumers.
They have segmentation models, campaign reports, brand trackers, and dashboards filled with behavioral data. On paper, it looks comprehensive.
But when you ask a more operational question—
“Why did this specific shopper buy this product, in this store, at this moment?”
The answers quickly become vague.
Because most existing systems are not designed to answer that level of precision. They are built to describe trends, not to model reality.
- CRM systems focus on identity, but lack transaction depth
- Media platforms optimize for engagement, not purchase truth
- Retail data shows sales, but not intent or causality
- Surveys capture opinions, but not behavior
Each dataset is directionally useful, but fundamentally incomplete.
And when these partial views are combined without a unifying structure, the result is not clarity—it’s distortion.
This is the core limitation:
You don’t have a connected model of how commerce actually works.
To fix this, you don’t need more dashboards.
You need a different way to structure reality itself.
Why graphs—not tables—reflect how commerce behaves
Traditional data systems are built on tables: rows and columns, neatly organized, static, and easy to query.
But commerce is not static.
It is dynamic, relational, and constantly evolving:
- Shoppers move across channels and locations
- Products interact with each other (bundles, substitutes, complements)
- Stores differ in context, constraints, and demand patterns
- Transactions connect everything in real time
Trying to capture this with isolated tables leads to fragmentation.
Graphs solve this problem.
A graph is a structure that represents:
- Entities (users, products, stores, transactions)
- Relationships (who bought what, where, when, and in what context)
More importantly, graphs are continuously updated, allowing the system to reflect changes as they happen.
This is critical for commerce, where timing and context directly influence outcomes.
Commerce SuperIntelligence is built on four foundational graphs that, together, create a unified model of reality.
The Four Graphs: Your commerce nervous system
1. User Graph: Understanding behavior, not just identity
Most organizations define users through static attributes:
age, gender, location, income bracket.
But these attributes don’t explain behavior.
The User Graph focuses on:
- Purchase frequency and recency
- Category preferences and switching patterns
- Responsiveness to promotions or incentives
- Shopping missions (stock-up, impulse, trial, etc.)
- Lifetime dynamics and evolving habits
It continuously updates as new interactions occur.
This allows brands to move beyond broad segmentation and answer:
- Who is likely to try a new product?
- Who is at risk of switching to a competitor?
- Who responds better to bundles vs. discounts?
The shift here is critical:
From “who the user is” → to “how the user behaves over time.”
2. Product Graph: Moving beyond SKU-level thinking
Products are often treated as isolated units—each SKU analyzed independently.
But in reality, products exist in a network:
- Some are substitutes (brand switching)
- Some are complements (bought together)
- Some act as entry points (trial drivers)
- Some are dependent on occasions or timing
The Product Graph captures:
- Hierarchies (brand → category → SKU)
- Attributes (size, flavor, format, pricing tier)
- Relationships (co-purchase patterns, substitution likelihood)
- Promotion sensitivity and elasticity
This allows brands to answer deeper questions:
- If we promote SKU A, which products gain or lose?
- Which product is the best entry point for new users?
- How do bundles influence total basket value?
Without this, optimization often creates hidden losses—
driving sales in one SKU while cannibalizing another.
3. Location Graph: The missing layer in most strategies
Location is often treated as a simple filter:
city, region, or store list.
But physical context plays a much larger role:
- Store formats vary (modern trade vs. general trade)
- Neighborhoods influence purchasing power and preferences
- Foot traffic patterns change by time of day or season
- Store-level constraints affect availability and execution
The Location Graph captures:
- Store-level performance and characteristics
- Micro-market dynamics (not just macro regions)
- Shopper flow and visitation patterns
- Environmental context (urban density, competition, accessibility)
This enables decisions such as:
- Where to launch a new product for maximum adoption
- Which stores are best suited for certain mechanics
- How to localize activation strategies without fragmentation
Without a Location Graph, strategies remain overly generalized—and often inefficient.
4. Transaction Graph: The only source of truth
This is the most critical layer.
Because regardless of how sophisticated targeting or engagement becomes, the only outcome that matters is:
Did a transaction happen—and was it incremental?
The Transaction Graph captures:
- Receipts and POS events
- Basket composition (what was bought together)
- Purchase frequency and timing
- Cross-brand switching behavior
- Verified incremental lift
This layer does two things:
- Validates reality
It confirms whether actions led to actual sales—not just engagement. - Closes the loop
It feeds outcomes back into the system, improving future predictions and actions.
Without a Transaction Graph, everything else becomes speculative.
With it, commerce becomes measurable and optimizable at a fundamental level.
How the graphs work together
Individually, each graph provides value.
But the real power emerges when they are connected.
For example:
- The User Graph identifies a high-potential shopper segment
- The Product Graph determines the best entry SKU
- The Location Graph pinpoints where activation will be most effective
- The Transaction Graph verifies whether the intervention worked
This creates a system where decisions are not based on isolated insights, but on interconnected intelligence.
Instead of asking:
“What performed best last month?”
You can ask:
“What will drive incremental growth next—and where should we act?”
Conclusion: From fragmented views to a unified model
Most CPG organizations operate with multiple disconnected views of the same reality.
Each team sees a different slice:
- Marketing sees engagement
- Sales sees distribution
- Trade sees promotions
- Analytics sees reports
But none see the full system.
The four graphs solve this by creating a shared, continuously updated model of commerce.
This is not just a technical upgrade.
It is a shift in how decisions are made:
- From assumptions → to verified relationships
- From static segmentation → to dynamic behavior
- From isolated optimization → to system-level impact
And most importantly:
From reacting to outcomes → to shaping them.
Summary
To build Commerce SuperIntelligence, brands need more than data—they need structure.
The four foundational graphs provide that structure:
- User Graph → behavior and responsiveness
- Product Graph → relationships and dynamics between SKUs
- Location Graph → physical context and micro-market insights
- Transaction Graph → ground truth and validation
Together, they form a connected system that reflects how commerce actually works—enabling better decisions, faster learning, and measurable growth.
Turning data into a real system
Many organizations already have pieces of these graphs—but they exist in silos, disconnected and underutilized.
The opportunity is not to collect more data, but to connect what already exists into a system that can sense, decide, and act.
Grivy works with brands to structure these layers into a unified model—linking shopper behavior, product dynamics, location intelligence, and transaction truth into one continuous loop.
