Testing a New Beverage: Taste Preference Is Not the Same as Purchase Intent

A drink can win a taste test and still fail at the shelf. Liking tells us whether someone enjoys a sample. Purchase intent asks whether that person says they might buy it. A paid transaction records what someone chose in a particular shopping situation. These answers are related, but they are not interchangeable.

For beverage buyers, importers and brand teams, a new tropical juice or ready-to-drink tea needs more than a pleasant flavor. It needs a credible route from taste acceptance to a purchase at a workable price, then ideally another purchase. The following staged method is an editorial testing framework, not a sales prediction or a report of tests conducted by Nam Viet.

1. Three questions that require three measures

Hedonic liking describes how much consumers like a tasted product. ISO 11136 provides guidance on measuring consumer liking in controlled areas. Its 2014 edition was confirmed as current in 2025. Its stated scope excludes behavioral tests of how much people actually consume. A liking score alone therefore cannot measure repeat sales.

Stated purchase intent usually comes from a questionnaire about a proposed product or price. Observed purchase requires a transaction. Repeat purchase adds the question of whether an eligible customer comes back within a specified observation period. Teams should preserve these distinctions in dashboards, research presentations and retailer conversations.

2. Why a good-tasting sample may not sell

Shoppers encounter a package, price and alternatives before they taste a bottled drink. The 2026 IFIC Food & Health Survey found that 88% of surveyed US adults named taste and 78% named price as factors in food and beverage decisions. The survey covered 3,005 US adults aged 18 to 80 during March and April 2026. Those figures describe reported general priorities, not sales or demand for a specific beverage.

In a 2025 laboratory study of not-from-concentrate orange juice, researchers varied packaging characteristics while keeping the juice constant. Packaging influenced visual attention and some evaluations. The 81 participants were university students, so the experiment does not represent a national retail market. It nevertheless illustrates the need to separate reactions to the liquid from reactions to its presentation.

A 2015 juice-blend study used sensory assessments alongside an incentive-based willingness-to-pay method with 228 participants. The findings concern one tested blend and population. Their methodological lesson is useful: research can examine monetary trade-offs alongside liking, rather than treating taste as the final buying decision.

3. Stage 1: Test the drink without brand cues

Begin with a specific formulation question. Does a less-sweet mango blend receive different liking scores from the reference recipe? Recruit people who could plausibly purchase the category in the target market. Define recruitment and eligibility before testing, including purchase habits, age range and relevant ingredient-safety screening. Do not recruit only colleagues who already favor the concept.

Present coded samples at consistent temperature, serving volume and preparation conditions. Balance presentation order when feasible. Ask neutral questions about overall liking and relevant characteristics such as sweetness, acidity, aroma and aftertaste. Avoid prompts that suggest the tester ought to find a product premium or healthier.

Report which version was preferred by the people tested, how variable their responses were and what conditions were controlled. A convenience sample of thirty people might reveal formulation issues, but it is not automatically representative of a city or country. Do not call a small pilot a forecast of market preference.

4. Stage 2: Make the shopping choice realistic

Now show the proposed package size, truthful product information, legally checked label, price and relevant alternatives. Ask consumers to choose among the candidate drink, competing options and purchasing nothing. An explicit no-purchase choice helps prevent an artificial result in which someone must select a drink even if none appeals.

Change one main design variable at a time when possible. If testing price sensitivity, keep other features consistent and document the price levels shown. If testing package transparency, avoid changing flavor or price simultaneously. Otherwise, the reason for a better response becomes difficult to interpret. Treat questionnaire-based willingness to pay as stated preference, not a completed sale.

5. Stage 3: Observe paid trial purchases

A limited market trial can answer what a survey cannot. Choose outlets with a defined audience and observation period. Record shelf price, promotion, product availability, units sold and returns. Where possible, compare outlets with similar exposure and stock conditions. A highly discounted test at one busy location is not evidence that the product will sell at full price in every market.

For an importer, landed cost, trade margins and retailer economics matter as much as taste. A product that attracts attention but cannot be offered at a viable shelf price may need redesign. Carry out appropriate safety, claims and local labeling checks before selling. These commercial and compliance checks do not follow automatically from sensory research.

6. Stage 4: Measure what happens after the first sale

First-week demand can reflect novelty, a sampling event or a discount. Repeat purchase provides another signal, although it requires careful measurement. Where privacy permissions allow, define the proportion of identifiable first-time buyers who bought again within a fixed window. Include the observation period and denominator. When customer-level data are unavailable, replenishment trends are only a proxy and should be labeled as such.

Weak repeat numbers could reflect an unsuitable formula, high price, stockouts or a short observation window. A 2021 study of 249 canned-meat testers found that favorable liking and stated willingness to pay did not always coincide. This was not a beverage experiment. It provides a cross-category example of why these questions should be measured separately, not an estimate of beverage conversion.

7. Use a stage-by-stage decision table

Set decision rules before seeing results. A brand team can define minimum acceptable liking, a workable price, trial economics and a repeat signal appropriate to its category. These thresholds depend on business risk and costs. They are not universal scientific cutoffs. When indicators disagree, investigate the reason instead of averaging them into a single success score.

8. How many people should take part?

There is no sample size that suits every beverage test. It depends on the target population, expected product differences, variation in responses, desired precision and subgroup comparisons. A small exploratory group is useful for finding problems. Broader claims call for planned recruitment and statistical design. Report how participants were selected rather than assuming a convenience sample represents the market.

9. Five mistakes that can invalidate a decision

• Equating a blind-test winner with the most commercially attractive offer.
• Reporting only positive purchase-intent answers and ignoring no-purchase responses.
• Comparing stores without recording price, stockouts or promotions.
• Converting a survey percentage into an unsupported national market-size estimate.
• Claiming repeat demand when the observation period is too short or buyers cannot be tracked.