What Is Sample Integrity?
Sample integrity means your survey respondents are real, unique, and attentive enough for decision-grade insights - not just 'completed' rows.
Sample integrity is the confidence that the people in your dataset are who they claim to be, appear once, and respond with enough care for the insight to be trustworthy.
It is broader than “fraud detection.” Fraud is one failure mode. Integrity also covers low-effort humans, professional survey takers, and synthetic text that looks fluent but empty.
If a complete can be automated, farmed, duplicated, or AI-written and still look fine in a flat file, you do not have integrity - you have a row count.
Why the term matters now
Buyers used to ask: “Did we hit n=1,000?” The better question is: “Of those 1,000, how many would survive a serious integrity review?”
When integrity fails, you see:
- Incidence collapse mid-field
- Tracker drift that “doesn’t match brand reality”
- Painful reconciliations with suppliers
- Stakeholders who stop trusting research
Integrity is also a commercial term. Clients increasingly ask how fraud was blocked, not only how many rows were deleted. Teams that can show in-field gates and audit-ready reasons win that conversation. Teams that only show a cleaning log look late.
Integrity vs quality vs cleaning
| Concept | Focus |
|---|---|
| Sample integrity | Are respondents real, unique, and non-abusive at entry and in-session? |
| Data quality | Are answers attentive, consistent, and usable for analysis? |
| Cleaning | Removing bad rows after (or during) field |
Cleaning is necessary. It is not a substitute for a pre-survey gate. Quality without integrity is polishing contaminated sample. Integrity without quality still needs attention and consistency checks. You want both - sequenced correctly. See pre-survey screening vs post-field cleaning.
What ops should measure
- Blocked-at-gate rate by supplier and reject category
- Residual cleaning rate after the gate
- Duplicate attempts across suppliers on the same project
- Open-end / AI-risk flags on studies that depend on verbatims
Those metrics tell a clearer story than “final n” alone. They also make reconciliations shorter because evidence already exists - see how to reduce reconciliations and bad completes.
How Maxna thinks about integrity
Maxna AI is built as sample integrity infrastructure for market research - not a general fake-account product. We block bots, farms, duplicates, geo/device fraud, and AI-written answers before they distort insights. Tiers like Lumen, Cadence, and Aether match depth to study risk without changing the definition of integrity.
Learn more on the platform page or for your seat: panels, agencies, brands. For the full threat map, read the 2026 fraud guide. Contact us or review pricing when you are ready to operationalize the definition.