The Short Answer
Survey sample is priced per complete — a finished, qualified response — and the price is driven by three variables: incidence rate (IR), length of interview (LOI), and how specific your audience is.
In 2026, online survey completes run roughly $2–$15 each for general population work, while a 250-respondent study with real screening criteria lands closer to $40–$80 per interview. The spread is not vendor greed. A B2B study at 70% incidence might carry a CPI of $7.00, while one targeting high-level executives at 3% incidence runs $50.00 — the same questionnaire, a 7x price difference.
Understanding the formula is the difference between accepting a quote and negotiating one. And once you see it, an uncomfortable question follows: you are paying most of that money to find the right people, not to learn anything from them.
The Three Variables That Set Your Price
1. Incidence Rate (IR)
IR is the percentage of people the provider contacts who actually qualify for your study. If you needleft-handed espresso drinkers who switched brands in the last 90 days, the provider may screen 20 people to find one. You pay for all 20.
This is why cost rises non-linearly as IR falls. The increases become more pronounced once incidence drops below 20%. Roughly:
| Incidence rate | Typical audience | Relative cost |
|---|---|---|
| 70%+ | General population, broad consumer | Baseline |
| 30–50% | Category buyers, common job roles | 1.5–2x |
| 10–20% | Specific software users, niche B2B | 3–5x |
| Under 5% | C-suite, physicians, rare conditions | 8–15x |
The practical lever: every additional screening criterion multiplies cost. Before adding one, ask whether it changes the decision you will make with the results. Our guide to writing screener questions covers how to qualify tightly without over-screening.
2. Length of Interview (LOI)
Longer surveys cost more per complete because respondents need larger incentives to finish them — and because drop-off rises with every minute. A 5-minute survey and a 25-minute survey are different products at different prices.
LOI is also where cost and quality collide. Long instruments produce straightlining and speeding, so you pay more per complete and get worse data. If your survey has crept past 15 minutes, the fix is usually cutting questions, not raising the incentive.
3. Audience Specificity
Beyond the IR math, some audiences carry a structural premium because the panel supply is thin: physicians, IT decision-makers with budget authority, C-suite executives, owners of specific medical devices. Honoraria for physicians alone run $3–$8 per minute, before the provider's margin.
How to Read a Sample Quote
A quote that says "$12 CPI, n=400" is incomplete. Ask for these five things in writing:
- Assumed incidence rate. Quotes are conditional. If actual IR comes in lower than assumed, the price gets revised mid-field — this is the single most common budget surprise in market research.
- Assumed LOI, and what happens if the questionnaire runs long in soft launch.
- What counts as a complete — and who pays for terminates, over-quota respondents, and completes you reject for quality.
- Feasibility: can they actually deliver n=400 in your target window, or will field time stretch?
- Whether replacement completes are free if you scrub responses for fraud or inattention.
The soft launch matters more than people expect. Field 10% of your sample first, check actual IR and LOI against the assumptions, and re-price before the full launch. Skipping it is how a $5,000 study becomes an $11,000 study.
The Costs the CPI Doesn't Include
CPI covers the respondent. It does not cover the work around them, which is usually the larger number:
- Questionnaire design — and rework when soft launch reveals a broken screener
- Programming and testing
- Data cleaning — removing speeders, straightliners, and fraudulent completes, then buying replacements
- Analysis and reporting — the dominant hidden cost, since a data table is not an insight
- Follow-up qualitative — because when the survey says 40% are dissatisfied, you still do not know why, and that answer requires a second study
Overall, most organizations spend $5,000 to $75,000 on a single market research project in 2026, and traditional in-depth interviews run $500–$1,500 per conversation once moderator prep, execution, and transcription are counted.
The Question Worth Asking: Why Are You Buying Sample At All?
Sample pricing exists to solve one problem — you do not have access to the people you need. That is genuinely true for category-entry research, competitor customers, or general population norming.
It is often not true for the questions product and marketing teams actually ask. Churn drivers, onboarding friction, pricing sensitivity, feature demand, win/loss — the right respondents for those are already in your CRM, your product, and your support queue. For those studies, you are paying a 3–15x incidence premium to reach strangers who resemble your customers instead of talking to your customers.
This is where the cost model breaks in your favour. With Koji, you interview your own users:
- AI-moderated voice and text interviews run in parallel, around the clock — 40 interviews in the time one scheduled call takes.
- Incidence rate is 100% when you invite your own customers. No screening premium, because you already know who they are — import them from CSV or sync from your CRM.
- Six structured question types — open_ended, scale, single_choice, multiple_choice, ranking, yes_no — mean one study returns both the percentage and the reason behind it. No survey-then-interview double spend.
- Automatic thematic analysis removes the analysis line item entirely.
- Published pricing: €29/month for Insights, €79/month for Interviews, credits included. A text interview costs 1 credit, a voice interview 3. Only conversations scoring 3+ consume credits, so junk sessions do not bill you.
Compare that to a 5% incidence study at $50 CPI: 100 completes is $5,000 in sample alone, before analysis. The same budget buys years of Koji — and produces conversations rather than crosstabs.
When you do need external reach, compare the recruitment platforms on IR handling and replacement policy, not headline CPI.
Start free at koji.so — from question to insight in hours, not weeks, with no research expertise required.
Frequently Asked Questions
What does CPI mean in market research? CPI is cost per interview or cost per complete — the price you pay for one finished, qualified response. It is calculated from the incidence rate, length of interview, and how hard your target audience is to reach.
How much does survey sample cost per respondent in 2026? General population online completes typically run $2–$15 each. Studies with real screening criteria land around $40–$80 per interview at n=250. Low-incidence audiences such as executives or physicians can reach $50+ CPI on their own.
What is a good incidence rate for a survey? Above 50% is comfortable and keeps costs near baseline. Between 20% and 50% is normal for targeted B2B or category research. Below 20% costs rise sharply, and below 5% you should expect an 8–15x premium over general population.
Why did my sample provider raise the price mid-field? Almost always because actual incidence rate or length of interview came in worse than the quote assumed. Run a soft launch with 10% of your sample to validate both before committing to full field.
Does a longer survey cost more per complete? Yes. Longer instruments require larger incentives and suffer higher drop-off, so the per-complete price rises — and data quality falls through speeding and straightlining. Cutting questions is usually cheaper than raising incentives.
Can I avoid paying for sample entirely? For any question about your own customers — churn, onboarding, pricing, feature demand — yes. Interviewing your existing users has a 100% incidence rate and no screening premium. External sample is genuinely necessary for category entry, competitor customers, and general population benchmarks.
