Which membership should we launch, and at what price?
A conjoint study built to answer a product decision rather than to demonstrate a method. Consumers said price mattered. Their trade-offs said something else entirely.
- Role
- Consumer insights analyst
- Context
- Membership businesses routinely set features and price by intuition, then discover the error through churn.
- Business question
- Which product and price combination captures the most demand - and what are buyers willing to give up to get it?
- What I owned
- Survey and concept design
- Conjoint estimation
- Market simulation
- Pricing and portfolio recommendation
- The written report
- Methods
- Ratings-based conjoint, 20 concepts
- Part-worth utility estimation
- Attribute importance
- First-choice market simulation
Academic study, simulated results. A graduate research project using a graduate-student convenience sample. All shares below are simulated preference shares within this study's sample - not market share, revenue, or the result of a real product launch. The value of the work is the decision logic, not the magnitude of the numbers.
Ask people what they care about and they will tell you price. Ask them to choose and they will show you something else.
Direct importance questions reward socially acceptable answers. Conjoint forces respondents to give something up, which is the only way to observe a trade-off. I built 20 realistic membership concepts across four attributes and collected appeal ratings on a seven-point scale, then decomposed those ratings into the value of each individual feature.
| Attribute | Levels tested | Business question it answers |
|---|---|---|
| Class type | Reformer Pilates · Cycling · Yoga | Which modality should the product be built around? |
| Monthly price | $20 · $40 · $60 | What will the market bear, and how steeply does demand fall? |
| Booking window | 48 hours · 1 week · 2 weeks · 4 weeks | How much scheduling flexibility is worth paying for? |
| Classes included | 4 · 8 · 12 · Unlimited | Does more volume increase perceived value, or just cost? |
What people were buying mattered roughly four times more than what it cost.
Fig. 01 - Relative importance of each attribute in driving stated preference. See note 1.
Direction of preference, level by level.
Importance says which lever moves demand. Part-worth utilities say which setting of that lever to choose.
| Attribute | Preferred | Weaker | What it implies |
|---|---|---|---|
| Class type | Reformer Pilates | Cycling, then yoga | Build the flagship product around Pilates. Yoga is not the lead offer. |
| Monthly price | $20 | Utility fell consistently as price rose | No price threshold effect observed - demand is smoothly price-sensitive across the tested range. |
| Booking window | One week | Two and four weeks both reduced preference | Longer horizons read as restriction, not planning. One week is the ceiling, not the floor. |
| Classes included | Unlimited | Lower counts, but the gap was small | Unlimited wins on paper and costs the most to deliver, for the least preference gain of any attribute. |
What the evidence says
Consumers prioritized the type of wellness experience over both price and the number of classes included.
Class volume had the highest-utility level (unlimited) but the smallest overall influence on choice - the classic signature of an attribute that costs the operator a great deal and buys very little preference.
What I would do about it
- Invest in the right modality before competing on price. The cheapest route to demand here is the product decision, not the discount.
- Do not lead with unlimited. Fund studio capacity and instructor quality instead, where the preference actually sits.
- Cap the booking window at one week and treat it as a feature, not a constraint.
- Treat $20 as an acquisition price, not a permanent one - and test the fall-off above the $60 ceiling before assuming it continues.
Get the product right and you can charge less, not more - and still take the market.
The simulation below tests that claim by introducing the recommended concept into a hypothetical competitive set and recalculating first-choice preference.
What happens when the recommended membership enters the market.
Baseline: a Pilates membership priced at $60 competing against cycling at $40. The cycling product dominates. Introduce reformer Pilates at $20 with a one-week booking window and four classes a month, and the preference structure inverts.
Fig. 02 - First-choice simulated preference share, study sample. See note 2.
What the evidence says
The existing Pilates product was not failing because Pilates was unwanted. It was failing because it was priced at $60 in a set where the alternative cost $40.
The recommended concept takes 71% of simulated first-choice preference and eliminates the incumbent Pilates product entirely.
What I would do about it
Launch reformer Pilates at $20/month, one-week booking, four classes. Retire the $60 Pilates product rather than repositioning it - the simulation shows it retains no first-choice preference once the new concept exists.
Then validate before committing capital: this is a small student sample rating hypothetical concepts. The direction is strong enough to justify a live pricing test; the magnitude is not strong enough to justify a build.
One product to launch. One to consider second.
Reformer Pilates, $20/month
| Booking | One week |
| Volume | 4 classes per month |
| Simulated share | 71% |
Built on the highest -importance attribute at its most preferred level, priced at the point of strongest demand.
Yoga, $40/month
| Booking | One week |
| Volume | 12 classes per month |
| Simulated share | ≈14% |
A second product broadens the portfolio into the least-preferred modality at a higher price - which is defensible only if it reaches buyers the flagship does not. That is a segmentation question this study was not designed to answer.
Open question I would resolve before recommending the second product
The source study reports a combined two-product share of 78.6%, but 71% and 14% do not reconcile to that figure, and 14.29% is exactly 1/7 - which suggests a very small effective base. I have therefore left the combined portfolio figure out of the headline and reported only the numbers that reconcile. A second product is a decision I would want the raw simulation output and a confirmed respondent count behind before recommending it.
Reading trade-offs, not reporting averages.
The analytical work here is standard. The judgment is not. The study produces a finding that contradicts how the category actually behaves - operators discount, while the evidence says the product decision carries four times the weight - and the useful output is a pricing and portfolio recommendation with an explicit validation step attached.
I would rather present a $20 recommendation with its limits stated plainly than a confident number nobody can defend in the room.
Notes & sources
- Fig. 01. Class type at 63% is stated in the source study. The remaining three values are read from the study's own importance chart and are approximate.
- Fig. 02. Baseline shares (7% / 93%) and post-introduction shares (71% / 29% / 0%) are reported in the source study. Shares are first-choice simulated preference within the study sample only.
- Sample. Graduate students, convenience sample. Size not recorded in the source document. This constrains every number on this page and is the first thing a reader should be told.
- Method. Ratings-based conjoint, seven-point appeal scale, 20 profiles. Analysis and simulation in Excel.