Consumer Decision Tree (CDT): The Order Shoppers Actually Choose

The purchase decision hierarchy is the fixed order in which a shopper filters a category, narrowing from everything on the shelf down to the one pack that goes in the basket. It settles a question you need answered before you set a single price: "When my shopper reaches the shelf, what do they decide first, and what do they decide last?"

A shopper standing in front of a supermarket fixture does not weigh every product at once. They run a short series of filters, each one narrowing the field, until a single choice is left.

The hierarchy is a sequence of gates

In biscuits, a shopper might move through four gates in order:
1. Type: a sweet biscuit or a savoury cracker.
2. Brand: a leading brand, a challenger, or private label.
3. Format: a single pack, a multipack, or a sharing bag.
4. Price: standard shelf price, or whatever is on promotion.

The order is not universal. In some categories price comes first. In others format or the occasion leads. The hierarchy is specific to the category, often to the channel, and sometimes to the shopper segment.

Why the order decides who competes

Attributes near the top act as gates: they decide which products even get considered. If brand is the first gate and a shopper settles on one brand, every other brand is gone before price, format, or flavour is weighed at all. The higher your brand sits, the more it is shielded from a shopper switching away on a deal or a different pack.

Three calculations turn a decision hierarchy from a hunch into something you can measure and defend in a range review.

Decision-level importance

Estimate it from conjoint analysis, a discrete choice model, or a Shapley value decomposition of purchase drivers. The attribute with the largest share is the first gate.

Brand vulnerability

When brand explains only 15% of the decision, as it does in commoditised categories like bottled water, the brand is wide open to a shopper trading away on price or format.

Switching probability at a gate

Attributes higher in the tree carry lower switching probability. Those are the sticky ones, the gates worth defending.

An illustrative biscuit split

• Type: 32% of the decision variance
• Brand: 28%
• Pack format: 18%
• Price: 14%
• Flavour: 8%

That reads Type, then Brand, then Pack, then Price, then Flavour. The top two gates settle 60% of the choice before price is even reached.

Research in a frozen pizza category turned up two different hierarchies sitting inside one shelf, one for each of its two largest shopper groups, with a smaller third group rounding it out.

Scenario: one category, two decision trees

A panel study split the category's shoppers into three groups by how they navigate the shelf.

The brand-loyal group (45% of shoppers)

They run the gates in this order:
- Brand first: which maker?
- Type: thin crust, deep dish, or stuffed.
- Size: single, twin, or family.
- Price: regular or on promotion.
- Topping: pepperoni, cheese, or margherita.

The job here is to be seen first, so brand awareness and shelf visibility do the heavy lifting.

The price-driven group (35% of shoppers)

Their order flips:
- Price first: what is on promotion, what is under $5?
- Size: how much food for the money?
- Type: thin crust or deep dish.
- Brand: reputable enough?
- Topping: the preferred flavour.

The job here is to be the best deal in view, so a competitive price and a prominent display matter most.

The remaining 20% are occasion-driven: their first decision is type or format, set by what they are planning, a quick weeknight dinner or a weekend gathering. This group responds most to pack-format innovation.

A single averaged tree would put neither group's real first gate at the top, and would misread both. Splitting the tree by segment is what makes the shelf, the price, and the promotional plan line up with how people actually choose.

Pinning down the real hierarchy takes research, not a workshop guess. A few methods do most of the work, and the people who use them well treat the result as segment-specific rather than one tree for everyone.

Read actual purchase sequences

Decision-tree analysis on panel data infers the order from what shoppers actually do over time. If they hold the segment steady but switch brands freely, segment sits above brand.

Test the trade-offs directly

Conjoint analysis asks shoppers to choose between product profiles that vary on brand, price, size, and flavour. The attribute with the highest part-worth utility is the top gate.

Watch the shelf, then ask

Eye-tracking, in real or virtual stores, shows what shoppers look at first, which usually maps to the top of the tree. A short intercept ("what did you decide first?") and an out-of-stock substitution check round it out. If a shopper swaps to the same brand in a different format, brand sits above format; if they swap format within the brand, format sits higher.