Sampling and Respondent Recruitment: Getting the Right People into Your Research

Why sampling decides the value of research

A beautifully written questionnaire and sophisticated analysis cannot rescue research answered by the wrong people. If a survey meant to represent rural households reaches mostly urban ones, or a study of IT decision-makers includes people who have never influenced a purchase, the results will mislead, however precise they look.

Sampling is the process of choosing who will take part; recruitment is the work of finding and engaging them. Together they decide whether research findings can be trusted and generalised. This guide explains how sampling & recruitment works in practice, from the first feasibility check to the final weighted sample.

Start with a precise target population

Every study should begin with a written definition of who it represents. “Consumers” or “businesses” is not enough. A useful definition specifies characteristics such as age, location, behaviour, role and timeframe: “women aged 25–45 in metro and tier-1 cities who bought baby food in the last three months”, or “finance heads at manufacturing companies with 100–1,000 employees in India”.

The definition drives everything else: screening questions, sample sources, quotas, feasibility and how results are described. It also prevents a subtle but common problem, where a sample quietly drifts towards whoever is easiest to reach.

Probability and non-probability sampling

Most commercial research uses non-probability samples, particularly online panels, with quotas and weighting to keep them balanced. Large public and social research, such as our government and opinion research, more often uses probability-based designs such as multi-stage household sampling. Both are valid when chosen for the right purpose and described honestly.

Sample size: how many is enough?

Sample size depends on the precision required and on the subgroups to be analysed. For a simple percentage from a large population, about 385 completes give a margin of error of around ±5% at 95% confidence, and about 1,067 give around ±3%. These figures assume random sampling, and they apply to each subgroup you want to report.

In practice, the number of subgroups often drives the total. Comparing four regions and two age groups means eight cells, each needing an adequate base. Qualitative research works differently: depth and coverage of key segments matter more than statistics. Our quantitative research guide covers sample sizes in more detail.

Feasibility and incidence

Feasibility asks a simple question: can we realistically reach this many qualifying people, in these markets, within this timeline and budget? The key variable is incidence, the share of people contacted who qualify. A study of smartphone users has very high incidence; a study of people who bought a specific industrial machine in the past year may have incidence below one per cent.

Low incidence increases cost and time, because many more people must be screened to find each qualifying respondent. Honest feasibility uses real panel counts, past response rates and realistic incidence estimates, and flags risks before fieldwork starts.

• Estimate the size of the qualifying population in each market
• Estimate incidence from profile data, past studies or a short incidence check
• Apply expected response and completion rates by source
• Compare the result with required completes and timeline
• Agree adjustments early if numbers are tight

Quotas: balancing the sample in field

Quotas set target numbers for key characteristics, such as age, gender, region or company size, so that the achieved sample reflects the population or meets analysis needs. Quotas can be natural, mirroring the population, or boosted, over-sampling small groups so they can be analysed, then down-weighted at analysis.

Quotas must be monitored daily. Easy groups fill quickly; hard groups lag. Effective fieldwork management shifts effort, sources and invitations towards lagging cells long before the deadline. Our Sampledge platform shows quota progress across sources in real time.

Sample sources

Different audiences need different sources, and many studies blend several:

• Online research panels for consumers, business professionals and healthcare audiences
• Client customer lists, where consent allows, for satisfaction and loyalty research
• Telephone samples for audiences with limited internet use
• Face-to-face recruitment in households, shops, markets and institutions
• Professional networks, associations and directories for niche B2B roles
• Community and snowball recruitment for small, connected groups

Frequent Research recruits from its own panels, Turn2Opinion, MyFrequentRewards and MyHealthOpinions, plus vetted partners and offline data collection teams. Blending sources improves reach but requires consistent screening and deduplication across them.

Screening: letting the right people in

Screening questions decide who qualifies. Good screeners are short, neutral and hard to game. They avoid revealing which answers qualify, include several plausible options, and check consistency with profile data. For B2B and healthcare studies, they confirm real involvement and expertise rather than relying on job titles.

Screeners also exclude people who should not take part: those working in research, advertising or the client’s industry, and people who have taken part in similar research recently. These exclusions protect the integrity of results.

Verification and fraud prevention

Recruitment attracts some people who misrepresent themselves to earn incentives. Verification counters this: identity and contact checks at panel registration, deduplication across sources, digital fingerprinting for online surveys, telephone re-screening for qualitative participants and high-value B2B respondents, and consistency checks between screener answers and profile data.

Professional and healthcare audiences may be verified against registers or directories. Our guides to business panels and healthcare panels explain these checks in more detail.

Hard-to-reach audiences

Some groups are harder to reach: rural households, older citizens, senior executives, rare medical specialists, people with specific conditions, or communities wary of outsiders. Reaching them well usually means combining methods, extending fieldwork windows, working with trusted local partners, adapting language and timing, and sometimes accepting smaller but robust samples supplemented by qualitative depth.

Honesty matters most here. Over-promising quotas for hard-to-reach groups leads either to missed deadlines or, worse, quietly relaxed criteria. A good partner explains the trade-offs upfront.

Weighting: the final adjustment

Even well-managed samples differ slightly from the population. Weighting adjusts the influence of each respondent so the sample matches known population profiles on key variables, and it restores natural proportions after boosted samples.

Weights should be modest and documented. Large weights mean some groups were under-represented and reduce effective sample size, which should be reported alongside results. Weighting cannot fix a sample that missed entire groups; it only corrects imbalances among people who were actually reached.

Recruitment for qualitative research

Qualitative recruitment is more personal and detailed. Participants for interviews and groups are screened on behaviour and attitudes, re-contacted to confirm, briefed on the session and reminded beforehand. Mixing participants within groups, avoiding frequent research participants and confirming articulacy are all part of good practice.

Because each qualitative participant carries so much weight, a single wrong recruit can distort findings. Our qualitative research guide explains how recruitment fits into the wider study.

An example: a hard-to-reach B2B audience

A client needs 200 procurement heads at hospitals with at least 100 beds, across six states, in four weeks. Feasibility shows the online panel can deliver about half. The plan combines panel invitations with telephone recruitment from hospital directories, screening for real involvement in purchasing, and quotas by state and hospital type.

During fieldwork, quotas are tracked daily; telephone effort shifts to states that lag. Every telephone-recruited respondent is verified by role and hospital before the survey link is sent. The final sample meets the design, is weighted lightly by hospital type, and is documented with source mix and verification outcomes.

Common sampling mistakes

Most of these are prevented by careful design and honest feasibility before fieldwork begins.

• Vague target definitions that let the sample drift
• Ignoring incidence when setting timelines and budgets
• Too few completes in subgroups that matter for decisions
• Screeners that reveal the qualifying answers
• Blending sources without deduplication
• Heavy weighting hiding poor coverage
• Describing results as representing groups the sample did not reach

Sampling across multiple countries

Multi-country studies add complexity. Population structures, internet penetration, panel coverage and incidence differ by market, so the same quotas and sources rarely work everywhere. Each country needs its own feasibility check, locally sensible quotas and an appropriate source mix, while keeping the target definition consistent enough for results to be compared.

Coordination matters: one point of contact, a shared quota dashboard across countries and consistent screening and quality checks for every source. Our Sampledge platform supports this, giving the client a single view of progress across markets.

Respondent experience and response rates

Response rates depend heavily on how people are approached and treated. Clear invitations that explain who is asking and why, honest length estimates, mobile-friendly surveys, fair incentives and early screening all improve participation and data quality.

Over-surveying the same people reduces engagement over time, so panels manage invitation frequency carefully. For telephone and face-to-face recruitment, courteous introductions, convenient appointment times and follow-up reminders make a measurable difference.

Documenting the sample

Every study should end with a clear sample description: the target definition, sources used and their share, quotas set and achieved, incidence, removals and reasons, and any weighting. This documentation lets readers interpret results correctly and lets future studies be compared or repeated.

Working with a sampling partner

Ask how feasibility is calculated, which sources will be used and in what proportions, how screening and verification work, how quotas are monitored, and how sampling and weighting will be documented. A transparent partner shares source mix, removals and weighting effects as standard.

Read our complete market research guide, or request feasibility for your next study.

Frequently asked questions

• Can you boost small groups for analysis?

  Yes. We over-sample small groups so they can be analysed reliably, then weight them back to their natural proportion for total results.

• How do you avoid professional respondents?

  We limit how often panel members take part, exclude recent participants in similar studies, check consistency across surveys and remove members with poor quality scores.

• What is a feasibility check?

  An assessment of whether the required number of qualifying respondents can be reached within the timeline and budget, based on real panel counts, incidence and response rates.

• What is incidence in market research?

  The share of people contacted who qualify for a study. Low incidence means more screening is needed, which increases cost and fieldwork time.

• What is the difference between quotas and weighting?

  Quotas control who is recruited during fieldwork; weighting adjusts the data afterwards so the sample matches the population on key characteristics.

• Can you reach rural and offline audiences?

  Yes, through telephone and face-to-face fieldwork with local-language interviewers, often combined with online sample for connected groups.

• How do you stop the wrong people qualifying?

  Neutral screeners, profile consistency checks, deduplication, technical checks and, for high-value audiences, telephone verification.

• Do you provide sample only, without the full study?

  Yes. Many agencies use us for sample and recruitment only, supplying respondents to their own surveys or qualitative sessions.

• Can you recruit for focus groups and interviews?

  Yes. We recruit, re-screen, confirm and remind participants for qualitative research, online and in person.

• How do you document the sample?

  We report the source mix, quotas achieved, removals and reasons, and any weighting applied, so results can be interpreted correctly.

Still curious? Ask the assistant Talk to a specialist →

Explore our Market Research services