Conjoint Analysis for Pricing: Methods, Outputs & When to Use It

Conjoint analysis is a research technique that reveals consumer preferences by forcing trade-offs between product features. Instead of asking "How important is X?" (which gets vague answers), conjoint asks "Would you prefer Product A with features X, Y at price P1, or Product B with features X', Y' at price P2?"

By analyzing patterns across many such choices, conjoint decomposes overall preference into part-worth utilities for individual features -- including price. This allows you to calculate the WTP for specific features: how much more will consumers pay for real chocolate vs. compound? For a resealable pack vs. a standard one? For organic ingredients vs. conventional?

Key advantages over direct WTP questions:
- Realistic: consumers make trade-offs in conjoint just as they do at the shelf
- Less susceptible to hypothetical bias than direct price questions
- Reveals relative importance of features -- not just whether they matter, but how much
- Can simulate market share for product configurations that do not yet exist

Types of conjoint:
- Choice-Based Conjoint (CBC): Most common. Respondents choose from sets of alternatives.
- Adaptive Conjoint (ACA): Adapts questions based on earlier responses. More efficient.
- Menu-Based Conjoint: Respondents build their own product. Good for customizable products.

In FMCG pricing, CBC is the standard because it mimics the shelf choice decision.

Conjoint model (multinomial logit):
Utility of product j = Sum(Beta_k x Feature_Level_jk) + Beta_price x Price_j

Where Beta_k = part-worth utility for feature level k

WTP for a specific feature improvement:
WTP_feature = -(Beta_feature_upgrade - Beta_feature_base) / Beta_price

Example from biscuit conjoint:
- Beta(real chocolate) = 1.8, Beta(compound chocolate) = 0.6
- Beta(price) = -0.45 per dollar
- WTP for real chocolate = -(1.8 - 0.6) / (-0.45) = $2.67

Interpretation: consumers would pay up to $2.67 more for real chocolate vs. compound.

Market share simulation (logit model):
Share_j = exp(U_j) / Sum(exp(U_k)) for all products k

Predicted revenue = Share_j x Market_Size x Price_j

This allows "what-if" simulation: "If we upgrade to real chocolate and raise price by $1.00, what happens to market share?"

CrunchField conducted a CBC conjoint among 800 premium biscuit buyers. Features tested: chocolate type, pack format, pack size, brand, and price.

Part-worth utilities (relative to baseline):
- Real chocolate vs. compound: +1.20 utility units (WTP = $2.67)
- Resealable pack vs. standard: +0.54 utility units (WTP = $1.20)
- 300g vs. 200g pack: +0.72 utility units (WTP = $1.60)
- CrunchField brand vs. private label: +0.90 utility units (WTP = $2.00)

Current product: Real chocolate, standard pack, 300g, branded, $4.29
Conjoint-predicted WTP: $3.29 (PL baseline) + $2.67 + $0.00 + $1.60 + $2.00 = $9.56

Wait -- $9.56? That seems absurdly high. And this is exactly where conjoint requires careful interpretation. The raw WTP sum assumes each feature is independently valued and there is no diminishing returns. In practice:

Adjusted WTP (applying 40% conjoint-to-reality discount): $9.56 x 0.60 = $5.74

This adjusted figure aligns much better with revealed WTP from scanner data ($5.10 for Premium Loyalists, $4.35 for Mainstream Regulars).

Actionable insight: The conjoint reveals that the resealable pack upgrade (WTP = $1.20, cost to implement = $0.08/pack) has the highest ROI of any feature improvement. CrunchField added resealable packs, took a $0.30 price increase, and saw zero volume decline because the feature WTP far exceeded the price increase.

Conjoint is the gold standard for pricing research in FMCG, but it must be used correctly:

1. Design matters enormously: The features included, the levels tested, and the price range all influence results. Include only features that consumers actually notice and care about. Test prices within a realistic range -- not $1 to $20 for a biscuit pack.

2. Sample must match your target: A conjoint on "all grocery shoppers" is less useful than one segmented by your actual target consumer. Run separate analyses for premium seekers and value seekers -- their part-worths will differ dramatically.

3. Combine with other methods: Conjoint excels at relative valuation (which features matter most, what is the premium for each). Combine with a Van Westendorp PSM for absolute price range and a Gabor-Granger direct price-point test for demand curve shape.

4. Beware of feature overload: Including too many features (more than 6-8) creates cognitive burden and degrades data quality. Focus on the features that are actual decision drivers, not the full product specification.

5. Update regularly: Part-worth utilities shift as consumer preferences evolve, competitors launch new products, and your brand equity changes. A conjoint study older than 18-24 months is directional at best.

6. Hypothetical bias still exists: Conjoint reduces but does not eliminate hypothetical bias. Discount WTP estimates by 10-20% for real-world pricing decisions.