---
title: Concept testing: does the idea land before you build it
description: Test a concept with synthetic personas before you commit a build: comprehension, appeal, credibility, intent and the objection that stops it. What it settles and what it cannot.
---

# Concept testing

A concept is a promise written on one page: what it is, who it is for, what it replaces. Concept testing reads that page the way your audience will read it, before anyone builds anything. It tells you whether the idea is understood, whether it is wanted, and which sentence is doing the damage when it is not.

## What a concept test settles

Most concepts fail on one of five points, and they fail in this order. Naming them before the build is cheaper than finding them in a launch.

### Comprehension

Can the reader say what it is, in one sentence, without the deck? If the restatement comes back as a different product, the concept is not ready to be judged on appeal.

### Relevance

Does it answer a problem the reader recognizes in their own week? A concept that solves a real problem nobody names reads as a nice idea and goes nowhere.

### Credibility

Does the promise sound achievable by the company making it? The same claim from a known brand and from an unknown one does not land the same way, and the test shows which reading you are getting.

## How a test runs here

| Step | What you do | What you get |
| --- | --- | --- |
| 1. Write the concept | The promise, who it is for, and what it replaces. Two or three variants if you are choosing between framings, written to the same length so the comparison is fair. | One survey where every persona reads the same text in the same order. |
| 2. Set the audience | The population the concept targets: age, income, region, or the whole panel. | Answers from a population calibrated on public US survey data. |
| 3. Ask the follow-ups | Comprehension, appeal, credibility, intent, and the open question that explains the score. | A report by question and by segment, with the verbatims that carry the reason. |

## The five questions worth asking

- **What is it?** An open restatement. This is the first filter, and the most informative single answer in the report.

- **Who is it for?** A second restatement. When the audience named by the personas is not the one you wrote for, the concept is aimed at the wrong room.

- **How appealing is it?** A five-point scale, not a yes or no. The midpoint carries as much information as the top box.

- **Would you try it?** Intent, with a time frame. "Someday" is not a signal.

- **What would stop you?** The objection. This is where the next version of the concept is written.

## How many concepts, how many responses

Three concepts in one survey is a practical ceiling: past that, the reader compares instead of reacting, and the first concept anchors the rest. For the sample, a direction needs around 25 responses per segment, and a comparison between two concepts needs around 100.

The [sample size calculator](/us/sample-size-calculator) gives the number for a given margin. If the concept carries a price, test the price separately: [pricing research](/us/pricing-research) has its own rules, because what people say they would pay and what they pay are two different measurements.

## What it does not do

A concept test reads a description, not a product. It does not measure what happens when someone actually uses the thing, it does not forecast market share, and it does not replace a pilot on real traffic. The population is synthetic, calibrated on public statistics: it gives a direction on how the concept is read, not a margin of error as if it were a probability sample. The [reliability page](/us/fiabilite) states what the engine guarantees.

## Test the concept you are about to build

Take the one-pager you already have, put it in front of the audience it targets, and read the restatement before you read the score. It takes minutes, and the free plan is enough for a first pass.

[Run a concept test](/us/login)

Related: [message testing](/us/message-testing), [pricing research](/us/pricing-research), and [how the synthetic population works](/us/personas-synthetiques).