The Short Answer
Professional respondents are people who take surveys frequently, often for income. Panel conditioning is what happens to anyone — professional or not — when repeated survey exposure changes their attitudes, knowledge, or behaviour.
The instinct is to treat both as fraud. The evidence does not support that framing. Studies have found that professionals do not systematically differ demographically or politically from non-professionals and do not show more response instability — and in some analyses novices were the most likely to give low-quality answers.
The real problem is concentration and change. A comScore study found that fewer than 1% of members across the ten largest online panels were responsible for 34% of all completed questionnaires. When a third of your market's "voice" comes from a fraction of a percent of panelists, you are not measuring your market. You are measuring a small, unusual, highly practised subpopulation — and one whose views may have been shaped by the very act of being surveyed.
Two Distinct Problems, Often Confused
Problem 1: Concentration
Panels are steeply skewed. A small core answers constantly; most members answer rarely or never. Estimates of professional prevalence vary widely by platform — roughly 1.7% on one source and 7.7% on another — but prevalence understates the effect, because what matters is their share of completes, not their share of membership.
The consequence is representativeness, not honesty. Frequent takers differ in ways that matter commercially: more time available, more reward-motivated, more practised at recognising screener logic and at guessing what a study wants.
Problem 2: Conditioning
Conditioning is subtler and applies to everyone. Panel conditioning harms data when respondents change their original attitudes and behaviours because of the survey stimulus — Pew's methodological work on this is the standard reference.
Concretely: ask someone monthly whether they have considered switching banks, and you have prompted them to consider switching banks. Ask about brand awareness repeatedly, and you teach the brand list. In trackers, this is a live threat to the wave-over-wave comparison the tracker exists to produce.
This is the uncomfortable part: conditioning is worst exactly where panels are most useful. Longitudinal designs need re-contactable members, and re-contact is what causes conditioning.
How It Shows Up in Your Data
- Suspiciously smooth trackers. Practised respondents answer consistently, which can look like stability but is partly an artifact of practice.
- Screener over-qualification. Frequent takers learn which answers keep them in. Ask about category usage and you get "yes" more often than the population supports.
- Compressed variance. Long exposure produces habitual scale use — an underappreciated cousin of central tendency bias and extreme response bias.
- Awareness inflation. Aided awareness climbs across waves because the survey itself taught the list.
- Attitude drift without market change. The classic conditioning signature: your metric moves, the market did not.
What Buyers Can Actually Do
1. Ask about frequency caps. Vendors that limit how many surveys any one respondent can take are more likely to protect data quality. Ask for the cap and the enforcement mechanism across blended supply, not just the owned panel.
2. Ask for tenure and frequency distribution of your delivered sample — median surveys completed in the last 30 days, and share of completes from members above the cap. Good providers have this. It is a better quality signal than panel size.
3. Refresh the panel between tracker waves, or run a rotating design where a portion of each wave is new. This trades some precision for protection against conditioning.
4. Do not over-screen. Every extra qualifying question teaches respondents what you want. Keep screeners behavioural and hard to game — see writing screener questions.
5. Distinguish quality from conditioning in your cleaning. Attention checks and fraud detection catch inattentive and fake respondents. They do not catch conditioning, because a conditioned respondent is attentive, real, and sincere. Cleaning cannot fix a frame problem.
6. Understand the sourcing model you bought. River sample has near-zero conditioning risk and near-zero profile depth; opt-in panels invert both. Our sample sourcing guide covers the trade-off in full.
The Structural Fix: Stop Sampling Strangers
Every mitigation above manages a symptom of one root cause — you are renting access to a pool of people whose defining shared trait is that they answer surveys for rewards. Caps and refreshes reduce the damage. They do not change the frame.
For most questions product, marketing, and CX teams ask, the correct frame is not a panel at all. It is your own customers: people whose relationship with you is the actual subject of the research, who have no reward-seeking survey career, and whose identity you already know.
Koji is built for that:
- AI-moderated voice and text interviews with your own users. No panel intermediary, so no professional respondents and no cross-study conditioning — each participant is a real customer with real stakes.
- Import from CSV or your CRM and send personalized interview links, so every response ties to a known account and duplicates are impossible.
- Conversation resists gaming in a way checkboxes cannot. A practised respondent can breeze through a grid; an AI interviewer that asks why, then follows up on the answer, surfaces whether the experience is real. That is the structural advantage of interviews over surveys against this failure mode.
- Six structured question types — open_ended, scale, single_choice, multiple_choice, ranking, yes_no — give you quantifiable answers alongside the reasoning, without a separate survey instrument.
- No moderator bias, identical framing across every session, and automatic thematic analysis with one-click reports.
- Published pricing — €29/month Insights, €79/month Interviews, credits included, and only conversations scoring 3+ consume credits.
Panels remain the right tool for population norming, category entry, and competitor customers. But if your tracker is drifting and your market is not, the problem may not be your questionnaire.
Start free at koji.so — from question to insight in hours, not weeks, no research expertise required.
Frequently Asked Questions
What is a professional respondent? Someone who takes surveys frequently, often for income. Prevalence estimates range from roughly 1.7% to 7.7% of members depending on the platform, but their share of completed surveys is far higher than their share of membership.
Do professional respondents give bad data? Not necessarily. Research has found professionals do not systematically differ demographically from other respondents or show more response instability, and novices are sometimes the most likely to give low-quality answers. The bigger issue is representativeness and concentration, not dishonesty.
What is panel conditioning? When repeated survey participation changes respondents attitudes, knowledge, or behaviour, so you end up measuring the effect of being surveyed rather than the market. It affects all repeat participants, not just frequent takers.
How do I detect panel conditioning in my data? Look for awareness inflation across waves, compressed variance, unusually smooth trends, and metric movement with no corresponding market change. Attention checks will not catch it, because conditioned respondents are attentive and sincere.
How do I reduce conditioning risk in a tracker? Ask your provider for frequency caps and tenure distributions, refresh part of the sample each wave, avoid over-screening, and hold source composition constant between waves so shifts are not mistaken for market change.
Is there a way to avoid professional respondents entirely? Yes — research your own customers instead of buying panel sample. People invited by personalized link from your CRM or product have no survey-taking career and no cross-study conditioning, and their identity is already verified.
