How many survey respondents do you need to validate a startup idea, and what will they cost?

By Chorus Research, operated by Collective Measure, Inc. · 1 October 2026

Founders sizing a validation survey have two questions: how many respondents, and what they cost. Of the four guides we read on 30 September 2026, each answers one question or the other; none prices the small sample it recommends for validation. This article sets out the published sample-size ranges for idea and concept validation, explains the trade-off behind them, and then prices that sample at Chorus Research's public rate.

The short answer

Published guides put early validation surveys in a range of roughly 50 to 200 respondents, with a decision-grade benchmark of about 384 when you need a ±5% margin of error at 95% confidence. At Chorus Research's published US general-population price of $1.00 per completed response, that range costs $50 to $200, and the decision-grade sample costs $384.

The rest of this article shows where those numbers come from and when they do not apply.

What the published guides say about sample size

Two vendor guides we read on 30 September 2026 give ranges by purpose rather than one magic number.

A sample-size guide from Koji lists 50 to 150 responses for concept validation, 384 for decision-grade metrics (±5% at 95% confidence for populations above roughly 20,000), and 300 to 500 for pricing research.

A guide from Vase.ai separates directional reads from confident ones: 100 to 200 respondents for a pitch or directional validation; 100 to 200 per concept for a monadic concept test, or 100 to 300 in total for a sequential monadic design; and 300 to 400 or more when a business decision rides on the result.

A third source, an article from the research consultancy Relevant Insights, warns that there is no universal formula. Five factors decide sample size: the analytical plan, population variability, confidence level, margin of error and budget. The ranges above are rules of thumb; the right number for your study depends on those five factors.

Why 384 keeps showing up

The standard large-population formula is n = z² · p · (1 − p) / e². With z = 1.96 for 95% confidence, p = 0.5 as the most conservative variance assumption and e = 0.05 for a ±5% margin, that is 3.8416 × 0.25 / 0.0025, which rounds to 384. Relevant Insights makes the same point about using 50% as the conservative assumption when you know nothing about the population.

The trade-off: margin of error shrinks slowly

Vase.ai's guide gives this ladder at 95% confidence:

At a flat per-complete price, going from 100 to 400 respondents quadruples the cost and halves the margin. Going from 400 to 1,000 costs two and a half times more and trims the margin by less than two points. A small sample is defensible when a directional read is enough, which is Vase.ai's own framing; it is not enough when the gap you need to detect is smaller than the margin of error. Relevant Insights' larger-sample example makes the same point: 950 respondents at about ±3.2%.

Three sizing mistakes

Sizing the study, not the segment. If you want to compare two customer segments, each needs the full sample. Koji's guide lists this as a common mistake: at decision grade that is 384 per segment, 768 in total.

Sizing per study, not per concept. In a monadic concept test each concept is shown to its own group, so each concept needs its own 100 to 200 respondents.

Starting from a number rather than a decision. The ForEntrepreneurs guide to surveys recommends working backward from the business decision the survey must support and keeping objectives narrow. Its case study reached more than 1,000 respondents through a targeted audience, but the method, not the number, is the lesson.

What a validation sample costs

This is where the four pages we read stop. None of them prices an order below 300 respondents. The lowest priced points are an illustrative comparison in the consultancy article, about $3,400 for 400 respondents against about $8,000 for 1,000, which is an illustration rather than a rate card, and Vase.ai's own do-it-yourself tiers, as read on 30 September 2026, of RM 5,000 to RM 12,000 (Malaysian ringgit, excluding SST) for 300 to 400 respondents. Vase.ai recommends 100 to 200 respondents for directional validation, so its cheapest priced tier sits above the sample it recommends for that purpose. The ForEntrepreneurs guide names SurveyMonkey Audience, Qualtrics, SurveyGizmo and Typeform as tools, but gives no prices.

So a founder who has settled on 100 to 200 respondents gets no price for that order from any of those four pages. What moves a quote, such as survey length, audience incidence and turnaround, is covered in our companion article on how much survey responses cost.

The same sample at Chorus Research's public price

Chorus Research publishes its prices on the homepage. A US general-population audience is $1.00 per completed response with a $5 minimum per launch; custom audiences start at $1.00 per completed response, and the exact figure for a custom audience comes from the estimate for that audience. Prices as published on 30 September 2026; the estimate in the product is the source of truth.

The price is quoted per completed response. The $5 minimum only matters for launches of fewer than five completes, which none of these are. For a custom audience, run the estimate first: through the hosted MCP server, the estimate_cost and estimate_launch tools return the figure, and that endpoint never charges on its own; a launch returns a browser checkout link.

How to decide

1. Write down the decision the survey must support.
2. Pick the smallest range that fits: directional or concept validation at 50 to 200; decision grade at 384; per segment or per concept if you are comparing.
3. Check the margin of error you are accepting against the ladder above.
4. Price it. At the published US general-population rate, the arithmetic is the sample size in dollars; for other audiences, estimate before you launch.

Run the estimate

Chorus Research offers a REST API and an MCP server for AI assistants. The hosted endpoint is mcp.chorusresearch.io/mcp, and its tools include list_audiences, create_survey, estimate_cost, estimate_launch and get_survey_status. Documentation: the getting-started guide, the REST API, MCP, the FAQ, and refund terms. Chorus Research is operated by Collective Measure, Inc.

Sources

Third-party figures are paraphrased from pages read on 30 September 2026; rates and recommendations change, so check the publisher before you budget.

• Vase.ai, survey sample-size guide (read 30 September 2026): validation ranges, margin-of-error ladder, DIY tier pricing in MYR.
• Koji, survey sample-size guide (read 30 September 2026): concept-validation range, the 384 benchmark and formula, per-segment sizing.
• Relevant Insights, sample-size article (read 30 September 2026): five-factor caveat, 50% conservative assumption, illustrative cost comparison.
• ForEntrepreneurs, guide to surveys (read 30 September 2026): decision-first method, case study, tools named.
• Chorus Research public prices, as published on the homepage on 30 September 2026.