How widespread is survey fraud in market research?

Quick Summary: Survey fraud in market research affects 15% to 45% of data, according to Kantar, NORC, and the Insights Association. To close the gap, start with purchase verification.

Ask your panel vendor what percentage of their sample is fraudulent, and you'll probably get a shrug or a number pulled from a webinar two years ago. Ask the organizations that actually measure it, and the shrug stops being an option.

• Kantar put the prevalence of survey fraud at 38% in 2023.
• The Insights Association's 2025 findings landed on 45%.
• NORC, reviewing the literature across the industry, puts the range at 15% to 30%, climbing as high as 45% on some platforms.

Three credible organizations, three separate studies; none of them anywhere close to zero. At the low end, that’s roughly one in six; at the high end, it’s nearly one in two. In a typical, self-reported 500-person survey, this means between 75 and 225 respondents aren’t real.

Where does survey fraud in market research come from?

Some survey fraud comes from bots. As early as 2020, commercially available survey bots advertised an "undetectable mode" built on humanlike behavior. Today, AI makes bots harder still to detect.

However, a meaningful share of survey fraud comes from professional respondents who've simply learned the system. They know which screener answers unlock eligibility, and they take enough studies per week that optimizing for speed and plausibility is basically their job.

When Pew tested the detection methods that researchers traditionally rely on, they didn’t work especially well. Two standard checks, one for speeding and one for attention, failed to catch most of the respondents later confirmed as bogus.

Where does survey fraud start?

Survey fraud starts in your panel recruitment.

Pew's comparison found bogus-respondent rates of roughly 4% to 7% among opt-in online sources, versus close to 1% among address-recruited panels. Duplicate IP addresses told a similar story: up to 3% among opt-in panels versus 1% on address-recruited ones, with duplicates nearly six times more common on foreign IPs than domestic ones.

A separate literature review found duplicate devices frequently spanning multiple panels at once. In one flagged batch, 90% had accessed the same study through two different panels, suggesting the same person, or bot, was working more than one vendor at a time.

Opt-in panels aren't worthless. However, self-reporting in recruitment sets a ceiling on how clean your data can possibly get, which no amount of cleaning can raise after the fact.

What drives survey fraud?

Every study of survey fraud in market research cited here points at money as the root cause.

One industry estimate puts fraud-related losses at roughly $350 million in 2024, close to 10% of total incentive spend across market research.

How can insights professionals minimize survey fraud?

Checking self-reported data with more self-reporting isn’t the answer. Speeding traps, attention checks, a "please confirm you're eligible" screener; they all ask respondents to prove they're real by answering more questions correctly. Professional survey takers are good at this, especially with AI on their side.

The one thing fraud can't easily fake is a transaction record that already existed before the study began.

That's the logic behind purchase verification: instead of asking someone to describe their behavior accurately, you start from behavior that's already recorded — a card charge, a receipt, an account connection — and build the study on top of that.

Pogo's panel takes this approach at scale. More than 3 million people who've connected credit or debit cards, accounts, or receipts have generated over $470 billion in tracked transaction volume, recruited through our opt-in app, rather than a survey-only panel built around people whose job is finishing surveys.

Verification doesn't mean that identity checks, duplicate detection, and response-quality screens aren’t necessary. However, it does ensure that your sample is built from people who did the thing, rather than say they did.

Want to talk with us about the difference verification can make to your market research outcomes? Book a quick conversation with Pogo here.