Copperleaf, a fictional snack brand, is planning a new protein bar line. Two questions: which bar to build, and which flavors to launch it in. Asking people what they want gets "everything, cheaper." So we used two methods built for trade-offs. A choice-based conjoint shows buyers sets of bars that differ in protein, sugar, texture, protein source and price, and asks which one they would buy. Their choices reveal what each feature is worth. A TURF analysis then finds the few flavors that together reach the most buyers.
Bar buyers
800
bought a protein bar in the past month
Choice tasks
12 each
3 bars plus "none" per task
Features tested
5
15 levels in all, 4 test prices
Flavors tested
10
rated one at a time, same buyers
Try a choice task Illustrative
Task 1 of 12. If these were the only bars available, which would you buy?
Click a bar to choose it.
Features and levels
| Feature | Levels shown |
|---|---|
| Protein | 10 g · 15 g · 20 g |
| Sugar | 12 g · 6 g · 2 g |
| Texture | Chewy baked · Crispy layered · Soft nut butter |
| Protein source | Whey · Plant |
| Price per bar | $1.79 · $2.29 · $2.79 · $3.29 |
Prices are fictional test levels chosen to span the shelf. None is the real price of any product.
Who answered
How the two methods fit
Each buyer's twelve choices are turned into a value, or part-worth, for every level of every feature. The spread between a feature's best and worst level shows how much that feature drives the choice. Price matters most, then protein. Protein source, the topic of the longest internal debate, matters least.
Feature importance · n=800 Measured
Share of the total spread in part-worths. Adds to 100.
By user type
| Feature | Daily n=310 | Occasional n=490 |
|---|
Daily users trade price for protein. Occasional users are the price-sensitive half of the market.
Part-worths, average utility Measured
Values within each feature are centered on zero. Compare levels within a feature, or compare spreads across features. A single value on its own has no meaning.
What each upgrade is worth, in cents per bar Measured
Each upgrade's part-worth gain converted to price using the slope between the $2.29 and $2.79 test levels. Use it to weigh cost against value when setting the build.
Build a Copperleaf bar and see how it would do against two competitor bars and the option to buy nothing. The number is share of preference: of 100 buyers choosing among these four options, how many pick each. It shows direction and size of a change. It is not a sales forecast.
Build the Copperleaf bar
"Current bar" is Copperleaf's existing product as described in the test.
Copperleaf share of preference
0%
Competitor bars in the test Fictional
Ironpeak 20 g protein, 6 g sugar, chewy, whey, $2.79
Meadowbar 15 g protein, 6 g sugar, soft nut butter, plant, $2.29
Share of preference, 4 options Simulated from measured utilities
Same bar at each test price
This sample uses average utilities to keep it light. A live simulator runs on each buyer's own utilities and can be cut by user type, region or channel.
Each buyer rated all 10 flavors on a five-point purchase scale. A flavor reaches a buyer if they would definitely or probably buy it. TURF (total unduplicated reach and frequency) counts how many buyers would find at least one flavor they want in a line-up. Tick flavors to build your own line-up, or use the buttons.
Build a line-up · n=800 Measured
Bar and figure: share of all 800 buyers who would definitely or probably buy that flavor on its own.
Reach of your line-up
0%
Best line-up at each size Measured
Each bar is the best possible line-up of that size. The fifth flavor adds less than 2 points.
Why the most popular three are not the best three
Cookie Dough is the third most popular flavor, liked by 30% of buyers. But 92% of its fans (217 of 236) already like Chocolate Peanut Butter or Salted Caramel. Adding it reaches few new people.
Blueberry Crumble ranks fourth on its own. But 86 of its 213 fans like neither of the top two, so it opens the line to a different group of buyers.
Swapping one for the other lifts reach from 70.1% to 78.5% with the same number of flavors.
Line-ups compared Measured
| Line-up | Reach | Buyers |
|---|
12.1% of buyers (97 of 800) would buy none of the 10 flavors, so no line-up can pass 87.9%.
The readout answers the two questions Copperleaf asked, using the same 800 buyers for both, so the bar and its flavors are built for the same people.
Recommendation
A 20 g crispy bar,
in four flavors
Build a crispy layered bar with 20 g of whey protein and 6 g of sugar, tested at the $2.79 level. Against the same two competitors, its share of preference is 40%, against 16% for the current bar. Launch it in Chocolate Peanut Butter, Salted Caramel, Blueberry Crumble and Mint Chocolate, which together reach 82% of buyers. Hold Cookie Dough for later: it is popular, but its fans are already covered.
Decisions, in order
Scorecard Measured
| Measure | Current plan | Recommended |
|---|---|---|
| Share of preference vs 2 competitors | 16% | 40% |
| Protein, sugar | 15 g, 12 g | 20 g, 6 g |
| Texture | Chewy | Crispy |
| Flavors, reach | 3 most popular, 70% | 4 by TURF, 82% |
The four flavors
Chocolate Peanut ButterSalted CaramelBlueberry CrumbleMint ChocolateNext step: an in-home use test of the 20 g crispy bar in these four flavors, to confirm the formula delivers what the concept promised.
Where this study could mislead