Conjoint studies

See how customers trade off features, price, offers, and messages, and model the market before you commit.

Talk to us about conjoint See how it works
What a respondent sees: one choice task of fourteen Illustrative example

Which of these cards would you choose?

$0 annual fee 1.5% cash back No lounge access
$95 annual fee 3x on travel 2 lounge visits
$550 annual fee 5x on travel Unlimited lounge

Sample choice task, illustrative: three credit cards at $0, $95 and $550 annual fee with different rewards and lounge access. The respondent has chosen the $95 card.

Questions this answers

  • What is the right price, and how sensitive is demand to it?
  • Which features are worth building, and what is each worth to customers?
  • Which bundle or tier structure maximizes uptake or revenue?
  • Which offer, message, and channel combination drives the best response for each segment?
  • How will we perform against specific competitor products?

Use it when

You're setting a price, designing a product or bundle, or choosing between offers and creative, and need to know what each element is worth.

Look elsewhere when

You have a list of 20 or more items to narrow down. Start with MaxDiff, then bring the leaders into a conjoint.

Where we apply it

Pricing
Willingness to pay, price elasticity, and revenue-optimal price points.
Product optimization
Feature bundling, tiering, and good-better-best design.
Offer optimization
Incentives, fees, and rewards, balancing uptake against cost.
Creative testing
Headlines, images, and calls to action as attributes, so you learn what drives response.
Channel design
The best mix of channel, timing, and message.

How we run it

Typically 2 to 3 weeks from kickoff to simulator.

One team, end to end. The same Sibilance researchers design the attributes, field the study, estimate the model, and build the simulator you keep. One vendor, one point of accountability, and no hand-offs between design and analysis.

  1. Frame the decisionAgree the choice to be made and the realistic market, including competitors.
  2. Build attributes and levelsGrounded in qual, so they reflect how customers actually think.
  3. Design the experimentStatistically efficient, with a task respondents find realistic.
  4. Field with quality controlsScreening for speeders, straight-liners, and fraudulent responses.
  5. Estimate and validateIndividual-level preferences, checked against holdout tasks.
  6. Simulate and recommendA market simulator, scenario plans, and the winning configuration.

What you get

  • Attribute importance and part-worth values
  • Willingness to pay by feature and segment
  • A market simulator you keep (Excel or web)
  • Recommended configurations by segment
  • Scenario plans for pricing and competitive moves

Our techniques

Choice-based, adaptive, and menu-based conjoint; hierarchical Bayes estimation; price-sensitivity modeling; value-maximizing optimization that weighs uptake against the cost of each feature.

Market simulator Illustrative data
Preference share simulator Four card configurations with different features and prices. Configuration 3 wins with 38 percent preference share, followed by configuration 4 at 27 percent, configuration 1 at 21 percent, and configuration 2 at 14 percent. Feature Config 1 Config 2 Config 3 Config 4 Travel rewards 1x2x3x5x Lounge access NoneNone2 visitsUnlimited Foreign transaction fee 3%NoneNoneNone Annual fee $0$39$95$550 Preference share 21% 14% 38% 27% Preference share simulator Four card configurations with different features and prices. Configuration 3 wins with 38 percent preference share, followed by configuration 4 at 27 percent, configuration 1 at 21 percent, and configuration 2 at 14 percent. Config 1 · $0 annual fee 1x travel · no lounge · 3% foreign fee 21% Config 2 · $39 annual fee 2x travel · no lounge · no foreign fee 14% Config 3 · $95 annual fee 3x travel · 2 lounge visits · no foreign fee 38% Config 4 · $550 annual fee 5x travel · unlimited lounge · no foreign fee 27%

Text summary of the simulator: four credit card configurations are compared on travel rewards, lounge access, foreign transaction fee and annual fee. Config 3 ($95 fee, 3x travel rewards, 2 lounge visits, no foreign transaction fee) wins with 38% preference share, ahead of config 4 at 27%, config 1 at 21% and config 2 at 14%. Figures are invented for illustration.

Case example: from one mailer to targeted offers

Challenge

A financial services client sent the same direct mail offer to every prospect.

Approach

A conjoint across offers, messages, and channels, with a simulator to plan strategy by segment and across the portfolio.

Result

Segment-specific offers and creative, with lower offer costs and higher uptake.

10–20% better than historical offer benchmarks

Questions we hear

Works well with

Segmentation & personas Know who you're designing for.
MaxDiff & TURF Narrow long feature lists first.
How many features can a conjoint include?

Typically six to ten attributes. For longer lists we use MaxDiff first, or an adaptive design.

Is stated preference reliable for pricing?

Conjoint forces trade-offs, which makes it far more predictive than asking what people would pay. We validate with holdout tasks and, where possible, in-market data.

Can we test creative with conjoint?

Yes. Breaking creative into elements shows which headline, image, and offer drive response, and which combinations work together.

What sample size do we need?

Usually 300 to 1,000 or more respondents, depending on the design and the segments you want to read.

Do we keep the simulator?

Yes. It's yours to run new scenarios after the project ends.

How long does it take?

Typically 2 to 3 weeks from kickoff to simulator.

Talk to us about conjoint

Tell us what you're pricing or designing. We'll come back with a proposed design, sample, and timeline.

No spam. Unsubscribe at any time.