Zurück

Best Concept Testing Research Companies: 6 Ranked (2026)

Matthieu Saussaye

A concept test is only worth paying for if it tells you what to change, not just which idea scored highest.

If you are looking for concept testing research companies, you are probably weighing two options: hand the study to a large agency, or run it yourself on a platform. The big names have decades of benchmarks. The platforms are faster and cheaper. SmartInterview sits in between: the AI does the work an agency researcher would do, from questionnaire to coded open answers, at the price and speed of a DIY tool. Here are six providers compared on deliverables, cost and speed.

The Short Answer

  • SmartInterview: the best choice for most teams. The AI drafts the concept test from your objective, recruits respondents in Switzerland and Europe or uses your own customers, probes vague answers with a follow-up question, and returns coded themes, sentiment and cross-tabs, typically in about 48 hours. First 50 respondents free.

  • Kantar: the reference for large CPG pipelines, with ConceptEvaluate AI screening 10 to 100 concepts from $5,500, trained on nearly 40,000 past innovation tests.

  • Ipsos: the deepest innovation forecasting offer, from fast idea screening to InnoCPT concept-product tests that model trial, repeat and incrementality.

  • Zappi: automated concept tests scored against category norms, with results in as little as 4 hours, built for enterprise teams testing many ideas.

  • Toluna: a global panel of more than 81 million consumers, with automated Concept Test Express and custom studies on top.

  • Pollfish: cheap DIY reach from $0.95 per response, or full-service projects from $5,000 per market.

Concept Testing Companies Compared

Provider

Price entry point

How respondents work

Analysis

Best for

SmartInterview

First 50 respondents free; plans from CHF 45/month for 150 respondents

Recruited audience in Switzerland and Europe, plus your own customer list in the same study

AI follow-ups on open answers, automatic theme coding, sentiment, cross-tabs, Excel/SPSS

Brands and agencies that need the reasons behind the scores, fast

Kantar

ConceptEvaluate AI from $5,500; custom studies quote-based

AI prediction from past tests (no panel) for screening; panel for full tests

Predicted trial, uniqueness and relevance against its database

Large CPG teams screening many early ideas

Ipsos

Quote-based

Representative consumer samples; InnoCPT on Ipsos.Digital (US, UK first)

Trial and repeat forecasts, source of volume, incrementality

Launch decisions that need a volume forecast

Zappi

Quote-based platform

Panel sample from ESOMAR and ISO certified suppliers; synthetic respondents for pre-screening

Trial and breakthrough potential against real-time norms

Enterprise innovation teams with a steady pipeline

Toluna

Quote-based

Own global panel of 81M+ consumers

Automated scorecards, AI mass-qual (Concept Explorer), custom analysis

Multi-country studies on one panel

Pollfish

DIY from $0.95/response; full service from $5,000 per market

Respondents invited inside mobile apps and websites (250M+ consumers)

DIY dashboard; presentation-ready report in full service

Quick, cheap consumer pulses

Prices are public entry points at the time of writing; where a vendor publishes none, I say quote-based. If you are choosing software rather than a company, read our guide to the best concept testing platforms.

How Concept Testing Research Actually Works

A concept test shows a description of a product, pack, claim or price to people in your target and measures how they react: appeal, purchase intent, uniqueness, credibility, and the reasons behind each. It is one of the core market research methods, and it is easy to do badly.

Full-service vs DIY

With a full-service company, a research team designs the questionnaire, buys the sample, runs fieldwork and writes the report. You pay for their judgment and their benchmarks, and you wait for the project manager. With a DIY platform, you do all of that yourself, which is cheap only if someone on the team knows how to write a clean concept test and code a few hundred open answers. The hidden cost of DIY is not the sample. It is the analyst time afterwards.

AI-native platforms change that split. When the software drafts the questionnaire, checks it, probes thin answers and codes the verbatims, you get most of what a full-service team delivers without the agency calendar.

Monadic or sequential monadic?

In a monadic test, each respondent sees one concept. Scores are clean and comparable to norms, but you need a separate cell of respondents per concept. In sequential monadic, each respondent sees several concepts in rotated order, which cuts sample cost but lets the first concept colour the others. We cover the trade-offs, and proto-monadic designs, in monadic vs sequential monadic testing. Whatever the design, the sampling and bias rules still apply.

What a good deliverable contains

  • Core KPIs per concept: purchase intent (usually top-two box on a 5-point scale), appeal, uniqueness, credibility, and price perception.

  • A clear winner, with the margin: is the gap between concepts real, or within noise for your sample size?

  • The reasons: likes and dislikes coded into themes, with real quotes behind each theme.

  • Segment cuts: customers vs non-customers, age, region, current brand.

  • Concrete fixes: which claim confuses, which benefit lands, what to rewrite before launch.

Most reports are strong on the first two and thin on the last three, because the open answers are short and nobody codes them properly. That is why open-ended questions deserve as much care as the scales.

Costs and timelines

A traditional agency concept test usually runs several weeks from briefing to debrief, and the large firms rarely publish prices. Public entry points give a sense of scale: Kantar lists ConceptEvaluate AI from $5,500 and Pollfish lists full-service projects from $5,000 per market. Automated platforms are faster: Zappi advertises results in as little as 4 hours, Kantar's AI screening in as little as 24 hours. Hard-to-reach B2B or medical audiences cost more everywhere.

The Best Concept Testing Research Companies in Detail

1. SmartInterview — AI-native concept testing with the reasons built in

SmartInterview starts from your objective ("which of these four concepts should we launch, and at what price?") and drafts the full questionnaire, with a quality score before launch. You choose respondents recruited in Switzerland and Europe, your own customer list through a reusable link, or both, with every answer tagged by source. When someone writes "not for me", the AI asks why. Respondents can answer by text or voice, in 99+ languages. In our studies, AI follow-ups deliver up to 4x longer answers and 2.2x more topics. The open answers are coded automatically into themes linked to verbatims, with sentiment and cross-tabs, exportable to Excel or SPSS. Built and hosted in Switzerland, GDPR compliant, plans from CHF 45 per month, results typically in about 48 hours.

Choose it if: you need to know why a concept wins or loses, in Europe, without an agency budget or timeline.

2. Kantar — benchmarks at scale for big pipelines

ConceptEvaluate AI, sold on Kantar Marketplace, predicts concept performance from nearly 40,000 real innovation tests and more than 6 million consumer evaluations, screens 10 to 100 concepts in as little as 24 hours, and starts at $5,500. Kantar says it is close to 90% accurate against past survey results. The AI screening uses no live respondents: a filter, not a final validation.

Choose it if: you are a large CPG brand with dozens of early ideas and want them ranked against category history before spending on fieldwork.

3. Ipsos — forecasting when volume is the question

Ipsos has the most developed innovation forecasting practice of the group. Its InnoCPT solution on Ipsos.Digital combines concept and product testing to predict trial and repeat purchase, with modules for sales forecasts, source of volume and incrementality; Ipsos says it delivers results three times faster than traditional methods. It launched in the US and UK first. Expect quote-based pricing and an agency-style engagement.

Choose it if: the decision needs a year-one volume estimate, not just a preference ranking.

4. Zappi — automated tests against category norms

Zappi automates standard concept tests so that every idea is scored the same way and compared with real-time norms. It covers early stage (idea screening, concept testing) and late stage (pack, pricing), reports trial and breakthrough potential, and advertises results in as little as 4 hours. It also uses synthetic respondents to pre-screen large idea pools. Pricing is not public.

Choose it if: you run a high-volume pipeline and want consistent scores across teams and years.

5. Toluna — one big panel, automated or custom

Toluna owns a global panel of more than 81 million consumers and sells concept testing in several forms: Concept Test Express, an automated solution on Toluna Start; Concept Explorer, an AI-powered mass-qual tool; and fully custom concept tests. Pricing is quote-based.

Choose it if: you need many countries on a single panel with a vendor that can also run the study for you.

6. Pollfish — cheap reach, optional full service

Pollfish invites respondents inside mobile apps and websites and claims access to 250M+ consumers. DIY surveys start at $0.95 per response; full-service projects start at $5,000 per market and include a presentation-ready analysis. Fine for simple concepts, weaker on rich open answers. See our Pollfish alternatives comparison.

Choose it if: budget is tight, the concepts are simple, and you mostly need numbers.

Where I draw the line

If the board wants a volume forecast for a national launch, I would still call Ipsos or Kantar: their benchmarks are the product. But most concept tests are not that. They are a team choosing between three or four directions and needing to know which one to back and what to fix. Paying agency rates and waiting weeks for that is a habit, not a requirement.

The failure I see most often is a report with clean purchase-intent bars and a slide of ten random quotes. The scores tell you who won; nobody can say why the runner-up lost, so the winning concept ships with the same weak claim it had in testing. The reasons are the deliverable.

Matthieu Saussaye, Founder of SmartInterview.

How to Brief a Concept Test (Whoever Runs It)

  1. Write the decision first. "Launch A or B", "keep or drop the premium tier", "which claim leads the pack". One decision per study.

  2. Make the concepts comparable. Same length, same format, same level of finish. A polished render beside a text card will win for the wrong reason. Our guide to validating a product idea shows how to write them.

  3. Pick the design. Monadic if you need clean scores per concept, sequential monadic if sample is limited and you have three or more concepts.

  4. Fix the KPIs before fieldwork. Purchase intent, appeal, uniqueness, credibility, price, on consistent scales (see our Likert scale guide).

  5. Ask why, and make sure someone probes. At least one open question per concept, with a follow-up when the answer is vague.

  6. Plan the cuts. Decide the segments you will compare and size each cell so the comparison holds.

  7. Demand fixes, not only scores. The report should end with what to change. Pharmalp, a Swiss natural health brand, used this approach to validate pricing and positioning of four new supplement concepts; read the Pharmalp case study.

When another company is the better call

  • You need a validated sales forecast for a national launch: Ipsos or Kantar.

  • You must screen 50+ early ideas against decades of category norms: Kantar ConceptEvaluate AI or Zappi.

  • You need ten or more countries outside Europe in one wave: Toluna or Pollfish.

For the wider landscape, see our reviews of market research tools and survey analysis tools.

Frequently Asked Questions

What do concept testing research companies do?

They show your product, pack, claim or price ideas to people in your target market and measure reactions such as appeal, purchase intent and uniqueness. Full-service firms also design the study, buy the sample and write the report; platforms let you run it yourself.

How much does a concept test cost?

It ranges widely. Agency studies are usually quote-based and run into thousands per market: Pollfish lists full service from $5,000 per market and Kantar ConceptEvaluate AI starts at $5,500. On SmartInterview, the first 50 respondents are free and plans start at CHF 45 per month for 150 respondents.

How long does concept testing take?

Traditional agency projects often take several weeks from brief to debrief. Automated platforms are much faster: Zappi advertises results in as little as 4 hours, and SmartInterview typically returns analysed results in about 48 hours.

What is the difference between monadic and sequential monadic testing?

In monadic testing, each respondent sees one concept, which gives clean scores but needs more sample. In sequential monadic, each respondent sees several concepts in rotated order, which is cheaper but adds order effects.

Should I use a full-service agency or a DIY platform?

Use an agency when you need a volume forecast or category benchmarks that only it holds. For most choices between a few concepts, an AI-native platform such as SmartInterview gives you the questionnaire design and analysis an agency would, faster and for far less.

Have four concepts and one decision to make? Test them on SmartInterview: your first 50 respondents are free.

Table of Contents

Verwandte Artikel

Kostenlos anmelden

Kostenlos anmelden

Kostenlos anmelden