The strongest product concept testing platforms are Zappi and Kantar for norm-based stage gates, Qualtrics for enterprises already on it, and panel-led platforms such as Vase.ai, Milieu Insight and Toluna for fast, affordable reads. Teams outside FMCG should prioritise flexible design, feature trade-off tools and real-respondent quality over templated FMCG norms.
Most concept testing advice is written for packaged goods: flavours, packs and shelf context. Product teams in consumer tech, financial services, durables, telco and direct-to-consumer brands have a different problem. Their concepts are bundles of features, a price and a promise, and the question is usually which combination to build.
We build Vase.ai, so weigh our view accordingly. If you work in FMCG, our comparison of concept testing platforms for FMCG launches is the better read. This piece is for everyone else.
What changed in product concept testing?
Three shifts matter when choosing a platform.
- Synthetic respondents arrived. Several vendors now offer AI-generated "respondents" that predict how people would answer. They can help early brainstorming, but they are not a substitute for real consumers when money is on the line. Ask any vendor to state clearly whether results come from people or models.
- Quality control moved into fieldwork. Better platforms now check responses as they arrive, rather than cleaning data afterwards. That matters because fraud and inattentive answers are a bigger risk for online concept tests than for most study types.
- Turnaround expectations collapsed. Product teams working in sprints now expect a concept read in days. A six-week study often lands after the build decision has already been made.
How do the main platforms compare?
| Platform | Best for | Feature trade-off tools | Consider instead if |
|---|---|---|---|
| Zappi | Templated tests with norms | Some, within templates | You need bespoke designs |
| Kantar | High-stakes gates needing benchmarks | Strong, with analysts | Speed and budget are tight |
| Qualtrics | Enterprises already using it | Strong conjoint and MaxDiff | You lack in-house research skills |
| Toluna | Always-on self-serve testing | Available | You need deep SEA quotas |
| Milieu Insight | SEA consumer reach | Standard question types | You need advanced trade-off modelling |
| Vase.ai | Fast SEA concept reads with expert support | Standard concept measures | You need global norms |
| SurveyMonkey | Very small budgets, DIY | Basic | The decision is expensive |
Ipsos and NielsenIQ also run product concept work, typically as full-service projects. Sample-only suppliers such as Cint and Dynata suit teams with their own scripting and analysis.
What should non-FMCG product teams prioritise?
Design flexibility over templates. A banking app concept or an appliance upgrade does not fit a pack test template. You need control over how features, price and benefits are shown.
Trade-off methods. Asking people to rate each feature produces a list where everything is important. MaxDiff forces choices between features, and conjoint estimates how features and price combine. If the decision is which features to build, you need at least one of these.
The right audience. Screen for people actually in the market, such as those planning to change phone, broadband provider or car within a set period. General population samples overstate interest in almost every new product.
Real people, verified. Given the rise of synthetic data, confirm that respondents are real, verified and checked during fieldwork. It is the single question most likely to separate a trustworthy read from an expensive guess.
When is a competitor the better choice?
If your product must clear a global innovation gate that scores concepts against a category norm, Kantar or Zappi is the better choice. We do not hold those norms.
If you need full choice-based conjoint with market simulators, and have analysts to run it, Qualtrics is very capable, and consolidating onto a tool your company already licenses often beats switching.
If you want to run hundreds of small tests a year across many countries self-serve, Toluna is built for that volume.
What mistakes should product teams avoid?
The most common mistake is testing a concept without a price. Interest in a new product almost always falls once people see what it costs, and the size of that fall is often the most useful number in the whole study.
The second is testing too late. If engineering has already committed to a feature set, the test can only confirm or embarrass the decision. Test while there are still options on the table.
The third is relying on one market. A concept that works in Singapore may not work in Indonesia or Vietnam, where price sensitivity and category maturity differ. If you plan a regional launch, include your largest market in the first round.
Where does Vase.ai fit?
We fit product and innovation teams launching in Southeast Asia who need a real-consumer read quickly and want research expertise available, not compulsory. Our verified panel of SEA consumers, AI-validated responses and fast turnaround keep studies quick and reliable. We use AI to check real people, not to replace them.
Send us your brief or your research question and we will suggest a study design and sample. You can run it DIY or with our research experts, and results land in a real-time dashboard.
How should you run a product concept test?
Write the decision down first: build A or B, launch at this price or not. Then choose the measures that answer it, typically purchase intent, uniqueness, relevance and a feature trade-off exercise.
Test concepts monadically where budget allows, show a realistic price, and set a pass mark before fieldwork. A concept that only wins without a price attached has not really won.
Frequently asked questions
Are synthetic respondents good enough for concept testing?
They can be useful for generating hypotheses or pre-screening very early ideas. For decisions involving real investment, results should come from verified human respondents, because models tend to reproduce what is already known rather than detect genuinely new reactions.
What is the difference between MaxDiff and conjoint in concept testing?
MaxDiff ranks a list of features or benefits by forcing respondents to choose the most and least important. Conjoint shows complete product profiles with different features and prices, and estimates how much each attribute drives choice. Conjoint is more powerful but needs larger samples and more analysis.
How long does a product concept test take?
Platform-based tests can return results within one to three days once the stimulus is final. Full-service agency projects typically take three to six weeks. The usual delay is finalising concept boards and price points, not fieldwork.
