Cleria

Methodology

How a response becomes a result you can use.

Invited, screened, quota-matched, and quality-checked before it reaches your dashboard — with the sample size and margin of error on the result.

Sample size and error
Respondent pathEvery response

  1. Invited

    Recruit

    Cleria's panel and vetted sources

  2. Screened

    Screen

    Market, category, and study criteria

  3. Quota-matched

    Quotas

    The population being studied

  4. Passed quality checks

    Quality

    Respondent history and security questions

  5. In your dashboard

    Live

    Aggregated, with sample size and error margin

Steps only. Pass rates differ by study.

Sample size

A margin of error you can see.

Every wave starts at 2,000 respondents per market, which keeps the maximum margin of error at ±2.2 points or less.

Max. margin of error± points · 95% confidence

MoE = 1.96 × √(0.25/n) × 100

Worst case (a 50/50 split), at 95% confidence.

Bar length is the maximum margin of error. Grey is below our 2,000 minimum: at n = 1,000 the maximum is already ±3.1, which is why every wave starts at 2,000 per market (maximum ±2.2).

Dashboards show the sample size and the error margin next to each result. Samples can scale to 50,000 respondents in most markets.

Dashboards show the sample size and error margin next to each result. Most markets can scale to 50,000 respondents when you need finer breakdowns. That sample can be the general population or a target group.

Our own panel

Respondents we know, not strangers from a sample marketplace.

Most research firms buy respondents from third-party sample providers. Each respondent arrives with one survey and no past. Cleria recruits and runs its own global panel, so every answer is checked against everything that person has told us before.

That matters more now that AI can fill in a survey. The result: data from real, attentive people, delivered faster.

  • Faster, with nobody in between.

    Panel, survey and dashboard are one system. No buying sample from vendors and no hand-offs between suppliers, so fieldwork starts as soon as your questionnaire is approved.

  • Fraud checked across surveys, not within one.

    Every response is compared with what that panelist told us before. A third-party provider hands you one survey per respondent, so that is all you can check.

  • Bots and AI-written answers stand out.

    A bot or a language model can fake one convincing survey. Staying consistent with months of earlier answers is much harder, so fakes get caught before they reach your data.

  • Trends you can trust, wave after wave.

    Same panel, same checks and a 12-month category lockout, so a change in your trend is a change in the market, not in where the sample came from.

Panelist #48 213

Member since 2023

34 · Female · Hamburg, Germany · 41 surveys

Survey history

  • 06-2026Sparkling TeaThis survey
  • 03-2026Grocery retailAge 34 · Household of 3
  • 11-2025Mobile networksAge 33 · Household of 3
  • 06-2025StreamingAge 33 · Household of 3
  • + 37 earlier surveys

Quality checks on this response

  • Demographics match earlier surveysPassed
  • Answers consistent with past responsesPassed
  • Open-ended answers compared over timePassed
  • Category lockout respectedPassed
  • Security and attention questionsPassed
Checked against 41 surveys of history5 of 5 checks possible

Illustrative example. Where a study needs vetted recruitment sources, those respondents still go through Cleria's screening and quality checks.

Category lockout

Respondents can't take another survey in your category for 12 months.

To avoid awareness creep, Cleria prevents respondents from answering other surveys in the same category for 12 months inside the Cleria panel.

12-month category lockoutInside the Cleria panel

One respondent's year · example category Sparkling Tea

Every repeat teaches people the brands in the survey, which inflates awareness. That is awareness creep. During the lockout, panelists can still answer surveys in other categories.

Cleria can enforce that lockout because it operates its own panel. Where vetted recruitment sources are used, respondents still go through screening and quality checks.

Fraud and quality

We know every panelist's history, and check every survey.

Because we know each panelist's history and add security questions to every survey, we catch fraud and inconsistent answers that one-off recruitment misses.

Quality pipelineBefore reporting

Where vetted recruitment sources are used, every survey still includes screening and quality checks before a response is accepted.

Representativeness

Representative of the people you are studying.

Samples are designed to be representative of the population being studied — whether that's the general population or a specific target group defined for the client.

Quota matchIllustrative · not a live study

Population being studied · US Sparkling Tea buyers

Target Achieved Share of sample · %
  • 18–2422% target · 21% achieved
    18–24: target 22 percent, achieved 21 percent.
  • 25–3434% target · 35% achieved
    25–34: target 34 percent, achieved 35 percent.
  • 35–4428% target · 27% achieved
    35–44: target 28 percent, achieved 27 percent.
  • 45+16% target · 17% achieved
    45+: target 16 percent, achieved 17 percent.

The 2,000-respondent minimum applies per market, per wave, whether the population is the general population or a defined target group.

Panel and recruitment

A global panel, plus vetted sources where a study needs them.

Cleria uses its global panel and vetted recruitment sources, with quotas, screening, and quality checks on every study.

Cleria global panel

  • Category lockout for 12 months
  • Respondent history on file

Vetted recruitment sources

  • Used where the market or study needs them
  • Screening and quality checks still apply

On every study

Quotas, screening, and quality checks

Your data

Private to your team.

Your data stays private to your team. We never share your data with other clients or use respondent data to train AI models.

  • Private to your team
  • Never shared with other clients
  • Respondent data not used to train AI models
How we handle data

See the dashboard.

Brand Tracker 360™ — 30 metrics, with the sample size and error margin on the result.

Explore features