Sample Quality & Transparency: Our Detailed Answers to ESOMAR's 37 Questions

Global Vox Populi was founded in 2013 and has been conducting market research for over a decade. We field quantitative, qualitative and specialist studies across 40+ markets in 60+ languages, with local teams in 20 countries, and we complete interviews every quarter across consumer, B2B and healthcare audiences in sectors including FMCG, healthcare and pharma, financial services, technology and telecom, automotive, and the public sector.

Our business is market, opinion and social research. We do not run advertising, lead generation or direct-marketing programmes, and participants are never recruited to watch ads or receive promotional messages. This matters because panels that mix research with marketing can produce participants conditioned by heavy promotional exposure, which can bias attitudes and behaviours. Our research operations are certified to ISO 20252:2026, which sets requirements specifically for organisations conducting market, opinion and social research.

Sampling is designed by our research team, not left to defaults in a platform. For every study, a named researcher defines the target population, builds the sampling frame, sets quotas and the weighting plan, and signs off the design before fieldwork. During fieldwork, the same team monitors quota fill, incidence, drop-off and data-quality flags so that they can act when quotas become difficult to fill. This could mean adjusting quota structure, widening the source mix, or recommending post-stratification weighting, always with the client informed.

Frontline project managers are trained in feasibility assessment, quota design, weighting, and quality-control procedures, and training is recorded as part of our ISO 20252 and ISO 9001 management systems, which require documented competence and ongoing training for staff whose work affects the quality of the output. The practical result is that the person you speak to about your sample can explain why it is built the way it is, and what its limits are.

We are a full-service research firm rather than a sample-only supplier. Our five instruments cover quantitative research (trackers, segmentation, pricing and conjoint, concept and product testing, message testing), qualitative research (in-depth interviews, focus groups, ethnography and online communities), consumer behaviour research, B2B research, and healthcare and pharmaceutical research (patient, physician, payer and key-opinion-leader studies).

Beyond sampling, we offer questionnaire and discussion-guide design, survey programming and hosting, translation across 60+ languages, fieldwork management, coding of open ends, weighting, tabulation, analysis and reporting. This means one accountable team handles the work from brief to delivery, which reduces hand-offs, shortens timelines, and keeps quality controls consistent across the whole project. If you only need sample or fieldwork, we can supply that on its own, and we apply the same quality checks.

Sources vary by audience and country, and we are explicit in every proposal about which sources will be used. (a) opt-in research panels that we run or manage; (b) client-supplied customer lists, for satisfaction, loyalty and CX studies; (c) professionally verified databases for B2B and healthcare audiences, such as physicians and specialists; (d) targeted recruitment for qualitative work, communities and online groups.

We do not use a single source to serve every project. A consumer study in a large digital market, a B2B study of industrial buyers, and a physician study in a smaller market each need different sources. The source plan is agreed with you in advance, and any change during fieldwork is disclosed.

We report this by project and by country, because the split differs by market. For each study, our methods appendix lists the share of completes by source, and shows which are sources we own and run directly, which we access exclusively, and which come from third-party partners. Knowing whether a vendor runs a source or merely resells it tells you how close they are to the participants, and how well they can verify recruitment quality, so we do not obscure it.

Recruitment depends on the audience. For professional, B2B and healthcare audiences we use invitation-only recruitment from verified sources, because open recruitment invites fraud among high-incentive groups. For general consumers, we use opt-in recruitment through, with quotas and screening to ensure the sample reflects the target population rather than whoever is quickest to respond.

Where a project needs probability-based sampling, such as address-based or list-based random selection, or face-to-face and telephone approaches in markets with lower internet penetration, we design the project that way, document coverage, and state the limits in the report. Channel mix varies by geography, since online coverage and trust differ significantly between, say, Singapore and Nigeria, and we will tell you what that means for representativeness in your target country.

We apply layered validation, so that no single control carries all the weight:

Technology: digital fingerprinting and de-duplication detect repeat entrants, multiple accounts, and suspicious device patterns.

Screening consistency: screener answers are cross-checked against profile data and against later in-survey answers to catch misrepresentation.

Professional verification: for physicians, specialists and B2B decision-makers, we verify credentials, role and employer against the study's eligibility criteria, and we may use call-backs or documentary checks for high-value audiences.

Human review: analysts review open-ended answers for gibberish, copy-paste text, and AI-generated responses that automated checks miss.

Back-checks: independent re-contact on a sample of every wave verifies that the interview took place and that the respondent was who they said they were.

These steps are documented procedures within our ISO 20252 and ISO/IEC 27001 management systems, and are applied consistently across projects and markets.

Our panel is 100% by email invitation. Where a study uses only partner sources, we identify the partner brand to you and report how participants entered the survey (device type and invitation route). We design surveys mobile-first, since a high proportion of completes in most of our markets come from smartphones, and we test questionnaires on mobile before launch.

We operate a managed-service model. Our researchers and project managers scope the audience, programme the survey, field it, monitor quality and deliver the data and documentation. We do not offer unmanaged self-serve sample purchase, which we believe gives buyers less control over quality than a human team accountable for the result. If you host the survey on your own platform, we can supply samples to your link and apply our quality controls to the traffic.

You can see, and control, where your sample comes from. Our proposals list the planned sources, and our delivery reports show completes by source and country. Buyers can approve or exclude specific sources, ask us to cap the share from any one source, or specify a fixed blend for tracking studies. Third-party sources are introduced only after passing our vetting (see Q15), and are disclosed by name on request.

Because different sources can produce different answers to the same question, consistency matters more than convenience. For studies that will be compared over time, we hold the source mix constant so differences between waves reflect the market, not the sampling.

We match source to application, and we are candid when a source is a poor fit:

Recontact, diary and product-testing studies: need sources where we can reliably go back to the same people, typically our recruited and consented panel members, or client customer lists.

Short vs. long questionnaires: longer instruments need engaged, well-incentivised participants, whereas short polls suit broader opt-in sources; we recommend length limits by source to avoid fatigue.

Mobile-only vs. desktop: surveys are built mobile-first; designs that require large screens (conjoint with complex visuals, for instance) are flagged and tested.

Communities and online qualitative: need recruited, screened and consented participants, typically through targeted recruitment rather than open intercepts.

B2B and healthcare: need verified professional databases and credential checks; consumer panels are not appropriate.

Our process runs in five steps, and it is documented in our quality system:

Define the population. We agree who counts as the target, in which markets, and what a "complete" is.

Design the frame and quotas. We select sources, set quotas, and fix the weighting plan before fieldwork.

Invite in controlled batches. Invitations go out in waves so we can monitor response by group and avoid over-filling the fastest responders, who tend to be unrepresentative.

Screen, field and monitor. Screening, quality checks and quota fill are monitored live.

Validate and weight. After fieldwork we remove failed-quality cases, then apply transparent post-stratification to known benchmarks, reporting the design effect and effective sample size.

On quotas, we recommend only those that matter to the research question. Age, gender and region are standard for general-population studies, and we add variables such as income, urbanicity, education or category usage where they drive the outcome. Over-specifying quotas can slow fieldwork, increase cost and push the sample toward unusual respondents, so we explain the trade-off rather than applying a default template.

Typical profile fields for opt-in panels include age, gender, region, household composition, education, employment and device use, with refresh cycles of 6–12 months. For partner sources, profile depth varies, and we state what is available before you commit. With consent, profiling data can be appended to your dataset, so you do not need to re-ask known questions and can keep surveys shorter. We tell you which profile data is self-reported and which is third-party sourced, because that affects how far you should rely on it.

To estimate feasibility we need the target audience definition, countries, expected incidence, survey length in minutes, required sample size, quotas, any exclusions (for tracker waves or prior participants), and the fieldwork window. For difficult audiences, such as specialist physicians or senior B2B decision-makers, we also need the screening criteria in detail.

We give a range rather than a single figure: a conservative estimate (lower bound), a likely estimate, and the assumptions on which each rests, such as incidence. If real incidence differs from the assumption, we say how that changes cost and timing. Most quantitative studies field in 2–4 weeks and qualitative in 1–3 weeks, depending on incidence and market, and we respond to a brief in under four business hours on median.

We raise the problem as soon as we see it, not at the deadline. The options are presented in writing: relax a quota, extend fieldwork, adjust incentives, add sources, or reduce the sample, with the effect of each on data quality. We never quietly substitute sources.

If a third party is needed, we name them and obtain your approval first. Suppliers are selected and monitored under our ISO 20252 and ISO 9001 supplier-control procedures, which require us to evaluate subcontractors before use, define quality requirements in writing, and review their performance.

We do not use a survey router.

Not applicable, since we do not use a router.

Participants are told who is conducting the research, the general subject area, the approximate time needed, the incentive, how their data will be used and protected, and that participation is voluntary and can be stopped at any time. We avoid details that would prime answers, such as the brand or the hypothesis being tested. Information is consistent across sources, though the invitation format varies (email, app, link), and we tailor wording to local language and norms.

No. Participants receive an invitation to a particular study based on their profile and the study's eligibility criteria, not a menu of surveys to choose from. This avoids self-selection toward higher-paying or more interesting topics, which can skew a sample.

We can adjust incentives during fieldwork, most often upwards for hard-to-reach groups, where low response would otherwise force us to loosen quotas and lose representativeness. Incentive changes are logged, and the incentive level can be recorded at participant level so analysts can check whether it affected who responded. Our approach to incentives is that they should be fair and proportionate to the time and effort required, never so large that they attract fraud or distort participation, and compliant with local rules (see Q32).

We track completion rate, drop-off by question, survey duration and participant comments for every project. Where a questionnaire shows abnormal drop-off, we review and refine it, and these lessons feed into our design guidelines. We can share completion and drop-off benchmarks for comparable studies on request.

Yes, every study closes with a methods appendix and sample debrief. It covers the target definition, sources and share of completes by source, fieldwork dates, quotas set versus achieved, incidence and completion rates, average survey length, exclusions and cleaning rules, number and reasons for removals, weighting method with design effect and effective sample size, and margins of error. We state what the data can and cannot support before decisions are built on it. This documentation also serves as the project record required under ISO 20252, and a sample debrief is available to prospective clients on request.

A person may complete a given study only once, and de-duplication prevents repeat entry across sources. Across studies, we limit participation frequency to prevent professional respondents and survey fatigue, and we apply exclusion rules where a participant has recently taken part in research on the same or a similar category or method, which prevents category bias. We monitor the share of completes from newly recruited versus long-tenured members, as both extremes can affect results.

We hold participation history, date of entry/tenure, source and channel, device type, and quality flags. These can be provided as appended variables in your dataset, and we can analyse results by source, tenure or participation level where you want to check for bias, always subject to participant consent and our privacy obligations. Participant-level data is handled under our ISO/IEC 27001 controls, and shared only with identifiers removed unless the project requires otherwise and consent covers it.

Identity is checked at every point of entry, not only at recruitment. At survey entry, we run device fingerprinting and duplicate checks, verify that the invitation link is unique and was issued to the person using it, and check against blocked or quarantined records. Screener answers are checked for consistency with profile data, and suspicious patterns, such as multiple entries from one device or IP address, implausible locations, or shared accounts, are flagged. For professional audiences, credentials are confirmed before the interview. These checks protect against hacked or shared accounts, duplicate accounts across channels, and automated entries.

For trackers, we lock the source definition, quotas and weighting scheme in the first wave, and treat any change as a documented, approved exception. Each wave we compare the composition of the achieved sample by source, device, tenure and demographics against previous waves, and we investigate drift before reporting. When a source must be replaced, we run overlap checks to estimate the effect on trend lines. Source can be appended to every respondent record, and blend reports by wave and country are provided with delivery.

Every participant carries quality indicators drawn from their survey history: attention-check results, speeding and straight-lining flags, inconsistent or illogical answers, open-end quality, duplicate or fraud flags, and results of back-checks. Depending on severity, we respond in one of three ways:

Remove the individual response from the project dataset and not pay for it;

Quarantine the participant, i.e. suspend them from further invitations while we review; or

Block them permanently for confirmed fraud or repeated poor quality.

We share removal counts and reasons with clients, and we compare profile data with in-survey answers on key variables, such as age, location, and category usage, to catch misrepresentation. Participants can contest removals through our support channel.

We combine design-stage and fieldwork controls:

Random or inattentive responding: embedded attention and trap questions.

Illogical or inconsistent responding: logic checks between related questions, and checks against profile data.

Straight-lining: pattern detection in grids, with questionnaire design that varies scale direction and grid length.

Over-use of "don't know": monitoring of non-response rates per participant and per question.

Incomplete responses: drop-off analysis, and we exclude partials from the final data unless agreed otherwise.

Speeding: completion-time thresholds based on the median duration observed in soft launch. Our speeder rule, e.g. less than 1/3 of median

AI-generated or copy-paste open ends: human review of verbatims.

Fieldwork authenticity: independent back-checks on a sample of every wave.

Automated flags feed human review rather than deleting data blindly, because over-aggressive cleaning can itself introduce bias. We document the rules applied, the number of cases removed, and the reasons, so you know exactly how much of the raw data reached your final dataset.

Our privacy notice is available at globalvoxpopuli.com/privacy. It explains what personal data we collect, our lawful basis for processing, the purposes for which it is used, how long it is kept, who it may be shared with (including processors and sub-contractors), international transfers, security measures, and participants' rights to access, correct, delete, restrict or withdraw consent, and how to contact us or lodge a complaint. Where a study uses a different notice, for example for healthcare research, participants receive the relevant version before they take part.

Our practices are aligned with the GDPR, and we apply local requirements in each country where we operate. Our approach covers:

Legal basis and consent: informed, freely given consent is obtained before collection, with clear information about purpose and data use, and consent is recorded.

Breach response: a documented incident-response procedure, governed by our ISO/IEC 27001 management system, covers detection, containment, assessment, notification of affected parties, clients and regulators within legal deadlines, and post-incident review.

Cross-border transfers: transfers are made only with an appropriate legal mechanism and contractual safeguards.

Retention: personal data is kept only as long as needed for the project and legal requirements, then deleted or anonymised under a defined schedule.

Participants give consent at the point of data collection, and can manage or withdraw it at any time by contacting us at panel_support@globalvoxpopuli.com. They can request access to their data, corrections, deletion, or an end to their panel membership, and we respond within the legal deadlines. For sources owned by partners, we require contractual commitments to equivalent consent, access and withdrawal rights, and a support channel for participants, which is also a participant-support requirement under ISO 20252.

Each country's requirements are checked at project set-up, covering tax treatment, caps and reporting thresholds, limits on gifts or payments to certain professionals (for example, healthcare professionals and public officials), and rules on prize draws. Healthcare incentives are set at fair-market value for time spent, and are structured to comply with pharmaceutical-industry codes and local law. Our legal and compliance review is part of project approval for new markets, and we track regulatory changes as part of the compliance register maintained under our management systems.

We do not collect personal data from children or young people unless a project specifically requires it and it is lawful and ethically justified. When it does, we obtain verifiable consent from a parent or legal guardian, use the age thresholds that apply in the relevant country (and, absent a national definition, the ESOMAR/GRBN definitions of children as 12 and under and young people as 13 to 17), collect only the minimum data required, apply enhanced safeguards in the design of questions and incentives, and follow the ESOMAR/GRBN Guideline on Research and Data Analytics with Children, Young People and Other Vulnerable Individuals, in addition to applicable laws such as the GDPR and COPPA.

Yes. We consider privacy when a study is designed, not after the data has been collected. In practice, that means: a data-minimisation review of every questionnaire so that we ask only what is needed; separation of direct identifiers from response data; pseudonymisation and anonymisation for analysis and delivery; role-based access so that only the project team can see personal data; encryption in storage and transit; defined retention and deletion dates set at project start; and privacy review of new tools and suppliers before use. These controls are embedded in our ISO/IEC 27001 and ISO 20252 processes, so they are audited rather than left to individual discretion.

Global Vox Populi is certified to ISO/IEC 27001:2022 for our Information Security Management System. That framework requires, and our programme includes:

an asset inventory and a formal, documented risk assessment and risk-treatment plan covering the information, systems and suppliers involved in our research;

a Statement of Applicability showing which Annex A controls we apply, covering access control, cryptography, physical and environmental security, operations security, supplier relationships, incident management and business continuity;

internal audits and management reviews, plus periodic independent surveillance and recertification audits by our certification body;

staff security awareness training and confidentiality obligations; and

a published security contact (globalvoxpopuli.com/.well-known/security.txt) for responsible disclosure.

Yes. Global Vox Populi is certified to ISO 20252:2026, the international standard for market, opinion and social research, and to ISO 9001:2026 for Quality Management Systems. ISO 20252 sets requirements specific to research, including documented project design, competence of staff, control of fieldwork and data collection, handling of subcontractors, data protection and security, and complete project records, and independent audits verify that we meet them. ISO 9001 supports this with firm-wide processes for continual improvement, document control, corrective action and management review. In addition, we follow ESOMAR's professional standards and guidelines, although we are not a corporate member of ESOMAR or any other industry body.

We can provide the following on request for the sources used in your study, trended where history exists.

Average qualifying or completion rate, trended by month

Percentage of paid completes rejected per month/project

Percentage of members/accounts removed or quarantined

Percentage of paid completes from members with 0–3 months' tenure

Percentage of paid completes from smartphones

Percentage of completes from owned/branded member relationships versus intercept participants

Average number of dispositions (attempts, screen-outs and completes) per member

Average number of paid completes per member

Active unique participants in the last 30 days

Active unique 18–24 male participants in the last 30 days

Maximum feasibility in a given country with nationally representative quotas, seven days in field, 100% incidence, 10-minute interview

Percentage of quotas fully met at delivery

We prefer to discuss these figures with you rather than hand over a table, because the same number can mean different things in different markets, and unexplained shifts in a metric are a prompt for investigation, not just a statistic.

Have a question that isn't covered here? Email inquiry@globalvoxpopuli.com and we will answer it in writing before you commit to a study