The added capabilities, called Research Defender’s Second Shield, allows users to find quality problems across sample sources and automatically stop problematic sources while research is still in field.

Second Shield uses 101 identity, device and behavioural signals to assign entrants to a research study with a risk score to decide whether they should be admitted.

Areas of focus for the tool include open-end quality, device fingerprint, supplier track record, respondent behaviour and project context.

These scores are then combined with research outcomes to identify patterns associated with a sample source, with those deemed to be delivering poor-quality traffic automatically shut off.

Vignesh Krishnan, co-founder and executive vice-president of product at Rep Data, said: “Fraud prevention across our industry has largely centred on determining whether you’ve got a good respondent.

“We can now zoom out and ask whether the source supplying those respondents is right for that specific project. That gives us another level of protection that simply isn’t possible when you’re looking at each respondent in isolation.”