{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-06-30T12:49:49.206Z"},"content":[{"type":"documentation","id":"fa6583da-3702-414e-8589-f60691970fc5","slug":"quota-sampling-guide","title":"Quota Sampling: A Practical Guide to Getting a Representative Sample","url":"https://www.koji.so/docs/quota-sampling-guide","summary":"A practical guide to quota sampling: defining cells, setting proportional vs non-proportional quotas, building screeners that route respondents, and monitoring cells to fill. Contrasts quota sampling with stratified, convenience, and random sampling, and shows how Koji automates quota enforcement with single_choice screeners and live composition tracking.","content":"## The short answer\n\n**Quota sampling is a non-probability method where you decide in advance how many participants you need from each subgroup — your \"quotas\" — and then recruit until each cell is full.** If your customer base is 60% small business and 40% enterprise, you set quotas so your sample mirrors that split, then stop recruiting each group once its quota is met. It gives you a sample that *looks* representative on the dimensions you care about, fast and cheaply, without the cost and complexity of true random sampling.\n\nUse quota sampling when you need a balanced sample on a deadline and a small, known margin of error is acceptable — which describes most product and customer research. Avoid it when you need to generalize to a population with statistical confidence (election polling, regulated market sizing); for that, use probability sampling. The honest framing: quota sampling controls *composition* but not *selection bias within each cell*, so it trades a little rigor for a lot of speed.\n\nWith an AI-native platform like Koji, quotas become almost self-managing: a `single_choice` screener question routes each respondent into the right cell, and you can watch cells fill in real time instead of tallying spreadsheets by hand.\n\n## What quota sampling actually is\n\nEvery sampling method answers one question: *who gets into the study?* Quota sampling answers it in two moves.\n\n1. **Define the cells.** Pick the characteristics that matter for your research — say, company size, role, and region — and divide the population into mutually exclusive groups (cells). For example: enterprise vs. SMB × admin vs. end-user = four cells.\n2. **Assign a quota to each cell.** Decide how many participants you want in each, usually to match the real-world proportions of your population. Recruit until each cell hits its number, then close it.\n\nOnce a cell is full, additional volunteers who'd fall into it are turned away — even if they're eager. That's the mechanism that keeps your sample from skewing toward whoever happens to respond first (typically your most engaged, most vocal users).\n\n## Quota sampling vs. other methods\n\nQuota sampling is frequently confused with stratified sampling. They look similar — both divide the population into subgroups — but the selection step is completely different.\n\n| Method | Subgroups? | Selection within subgroup | Can generalize statistically? |\n|---|---|---|---|\n| **Quota sampling** | Yes | Non-random (recruit until full) | No |\n| **Stratified sampling** | Yes | Random within each stratum | Yes |\n| **Convenience sampling** | No | Whoever is easiest to reach | No |\n| **Simple random sampling** | No | Random from whole population | Yes |\n\nThe takeaway: **quota sampling is the non-probability cousin of stratified sampling.** It gives you the same balanced composition, but because participants inside each cell aren't selected randomly, you can't attach a true confidence interval to the result. For qualitative and product research — where you want a *range* of perspectives rather than a population estimate — that's usually a fair trade. See [probability vs. non-probability sampling](/docs/probability-vs-non-probability-sampling) for the deeper distinction.\n\n## When to use quota sampling\n\nQuota sampling is the right call when:\n\n- **You're doing qualitative or mixed-methods research.** You want to hear from every important segment, not estimate a percentage to the decimal.\n- **You're on a deadline or a budget.** Quota sampling is dramatically faster and cheaper than building a random sample frame.\n- **You know your population's composition.** You can only set good quotas if you know the real proportions (e.g., from your CRM or analytics).\n- **You need to guarantee coverage of small but critical segments.** A pure random sample might miss your 5% enterprise tier entirely; a quota guarantees you hear from them.\n\nAvoid it when the deliverable is a statistically defensible population estimate, or when the dimensions that matter are unknown — in that case start with exploratory probability sampling.\n\n## How to run a quota sample, step by step\n\n### 1. Choose your quota dimensions\nPick the 1–3 characteristics most likely to change the answer to your research question. Common choices: customer segment, role/persona, lifecycle stage, region, plan tier. Resist the urge to add more than three — every extra dimension multiplies your cells and shrinks each cell's size.\n\n### 2. Set the quota for each cell\nDecide between **proportional quotas** (cells mirror the population — 60/40 if your base is 60/40) and **non-proportional quotas** (deliberately over-sampling a small segment so you have enough responses to analyze it, e.g., a 50/50 split even though enterprise is only 20% of the base). Non-proportional quotas are common in product research when a small segment is strategically important. If you over-sample, weight the data back to true proportions during analysis.\n\n### 3. Build a screener that routes respondents into cells\nA [screener question](/docs/screener-questions-guide) at the start of the study identifies which cell each respondent belongs to. In Koji, this is a `single_choice` structured question — \"Which best describes your company size?\" — whose answer assigns the respondent to a cell automatically. Strong screeners are the backbone of quota sampling; weak ones let people self-misclassify and corrupt your cells.\n\n### 4. Recruit and monitor cells in real time\nAs responses arrive, track each cell's fill rate. The classic failure mode is over-recruiting easy cells (your most engaged segment) while a hard cell sits empty. Because Koji reports respondent breakdowns live, you can see a cell stalling and redirect outreach before you waste interviews. Once a cell is full, close it so further volunteers in that group are screened out.\n\n### 5. Close cells and analyze\nWhen every cell hits its quota, stop. During analysis, compare cells against each other — quota sampling is built for exactly this kind of segment-by-segment comparison. If you used non-proportional quotas, apply weights so aggregate figures reflect the true population.\n\n## Where quota sampling goes wrong (and how to avoid it)\n\n- **Selection bias inside the cell.** Even with the right number of enterprise admins, if they're all from your happiest accounts, the cell is biased. Recruit from a broad, neutral source rather than only your fan base. See [sampling bias](/docs/sampling-bias-research).\n- **Too many dimensions.** Three dimensions of three levels each is 27 cells. Keep it simple.\n- **Misclassification.** A vague screener lets respondents land in the wrong cell. Use precise, mutually exclusive `single_choice` options.\n- **Treating it as probability sampling.** Don't report a margin of error you can't defend. Quota samples describe; they don't statistically generalize.\n- **Stopping too early on the hard cells.** Resist the temptation to \"call it\" when the easy cells are full but a strategic cell is half-empty — that defeats the purpose.\n\n## How Koji makes quota sampling almost automatic\n\nTraditional quota sampling means a researcher babysitting a spreadsheet, manually tallying who's come in and who to turn away. Koji removes that overhead:\n\n- **Structured screeners route automatically.** A `single_choice` screening question assigns each respondent to a cell without manual tagging. (Koji supports six [structured question types](/docs/structured-questions-guide): open_ended, scale, single_choice, multiple_choice, ranking, and yes_no.)\n- **Live composition tracking.** You watch cells fill in real time rather than reconciling counts after the fact.\n- **Import a known panel.** If you already have a segmented list, [import respondents via CSV](/docs/importing-participants-csv) with their segment attached.\n- **AI interviews scale the recruit.** Because each interview is conducted by an AI moderator over voice or text — no scheduling, no moderator — you can fill even hard cells in days, not weeks, while the analysis writes itself the moment the last cell closes.\n\nThe result is a balanced, segment-rich sample with a fraction of the operational drag — and the same conversational depth you'd get from a moderated interview, across every cell.\n\n## Related Resources\n\n- [Structured Questions Guide](/docs/structured-questions-guide) — the six question types, including the single_choice screeners that power quotas\n- [Probability vs. Non-Probability Sampling](/docs/probability-vs-non-probability-sampling) — where quota sampling fits in the sampling family\n- [Screener Questions Guide](/docs/screener-questions-guide) — writing screeners that route respondents cleanly into cells\n- [Sampling Bias](/docs/sampling-bias-research) — the within-cell bias quota sampling doesn't fix on its own\n- [Screening Participants Effectively](/docs/screening-participants-effectively) — keeping the wrong people out of your cells\n- [Importing Participants via CSV](/docs/importing-participants-csv) — loading a pre-segmented panel into a Koji study","category":"Research Methods","lastModified":"2026-06-30T03:15:14.993846+00:00","metaTitle":"Quota Sampling: A Practical Guide (Definition, Steps, Examples)","metaDescription":"Quota sampling explained: what it is, how it differs from stratified and convenience sampling, when to use it, and a step-by-step workflow for setting and enforcing quotas with screeners.","keywords":["quota sampling","quota sampling definition","quota vs stratified sampling","non-probability sampling","how to set quotas research","quota sampling example"],"aiSummary":"A practical guide to quota sampling: defining cells, setting proportional vs non-proportional quotas, building screeners that route respondents, and monitoring cells to fill. Contrasts quota sampling with stratified, convenience, and random sampling, and shows how Koji automates quota enforcement with single_choice screeners and live composition tracking.","aiPrerequisites":["probability-vs-non-probability-sampling","screener-questions-guide"],"aiLearningOutcomes":["Define quota cells from the dimensions that matter to your research question","Set proportional and non-proportional quotas and know when each applies","Distinguish quota sampling from stratified, convenience, and random sampling","Build screeners that route respondents into the correct cell","Avoid within-cell bias, misclassification, and over-recruiting easy cells"],"aiDifficulty":"beginner","aiEstimatedTime":"11 min read"}],"pagination":{"total":1,"returned":1,"offset":0}}