ResearchDart
Talk to us

Reverse bad completes in a language suppliers already use

After cleaning, paste the session IDs you are rejecting and choose a removal reason from the Global Data Quality code frame. The supplier sees exactly why, quota reopens, and the pattern shows up in quality reports by source.

Talk to us Reversal API

Why a shared code frame matters

Reconciliation used to mean spreadsheets of IDs with free-text reasons that every supplier interpreted differently. A shared set of codes turns each reversal into feedback a supplier can act on at the source.

The Global Data Quality initiative introduced its feedback loop code frame to the industry in January 2026. ResearchDart stores every reversal against one of its 18 codes and maps its own pre-survey decisions to the same codes.

Code definitions summarised from the GDQ feedback loop.

  1. Clean the data

    Run your usual checks on speeding, straight-lining, trap questions and open ends.

  2. Paste session IDs

    Reverse up to thousands of completes in one action from the project page, or call the reversal API per session.

  3. Choose a GDQ code

    Add a note if it helps the supplier understand the pattern.

  4. Suppliers see it immediately

    Payout drops to zero, quota reopens and the reason appears in their reconciliation feed.

The 18 GDQ removal reasons

Every reversal on ResearchDart uses one of these codes.

CodeReasonDefinition
1 Bot detection Failure at checks designed to ensure transactions are with humans rather than bots.
2 Third party fraud tool failure Flagged by third party fraud detection tools.
3 Geo-location check Failure at checks used to determine participant location is correct.
4 Participant duplication Indication of duplicate participants based on an identifier such as IP address or cookies.
5 Suspicious survey entry time Time of entry into the survey is suspicious.
6 Speeding / racing Completing a questionnaire faster than reasonably expected.
7 Excessive interview length Length of interview above what could reasonably be expected.
8 Straight lining / flat lining The same answer to the majority of grid or scale questions.
9 Inconsistent / contradictory answers Data does not align within one question or across questions.
10 Red herring / explicit trap question Failure at a question designed to check whether participants are paying attention.
11 Knowledge question Failing questions the participant should be able to answer.
12 Over-qualification / over-claiming Qualifying for an excessive range of categories.
13 Open end: poor quality Open end is too short, irrelevant, vulgar or nonsensical.
14 Open end: duplicate The same open-end answers repeated across questions or participants.
15 Open end: AI completed Open end appears to be copied from AI-generated text.
16 Duplicate responses in closed questions Closed-question answers suggest multiple instances of the same participant.
17 Ghost complete Complete recorded in the participant system but not in the survey data.
18 Unspecified issue / other Another reason not covered by the other codes.

Quality by source

The console groups pre-survey blocks and post-survey reversals by supplier and GDQ code, so you can see which sources cost you completes and why.

Quality by sourceLast 30 days
SupplierRemoval rateTop reasons
10424.1%610
126317.8%2415
11888.3%136
Console view with illustrative data

Reconciliation questions

What is the GDQ feedback loop?
A code frame published by the Global Data Quality initiative, a collaboration of research associations including the Insights Association, ESOMAR and MRS. It defines 18 standard reasons a respondent can be removed for quality, so buyers and suppliers describe reversals the same way.
Which completes can be reversed?
Any recorded complete on a project that has not already been reversed. Terminates, over quotas and quality terminates were never billed as completes, so they cannot be reversed.
Does a reversal reopen quota?
Yes. Reversed completes stop counting toward the project target and the supplier allocation, so replacement respondents can enter.
How do suppliers find out?
The session record changes immediately. Suppliers see the GDQ code, label, note and date in the transactions API and exports, and the payout for that session drops to zero.
Can router decisions be reported with GDQ codes too?
Yes. Sessions stopped by ResearchDart carry a router reason that maps to a GDQ code, for example duplicate devices to code 4 and speeders to code 6, so pre-survey and post-survey removals appear in one report.

Further reading

Data Quality

The GDQ removal codes, and why free-text reversal reasons have to go

Reconciliation has run on spreadsheets of IDs and reasons like "bad data" for as long as online sample has existed. The Global Data Quality code frame...

 Data Quality

Ghost completes: the reconciliation problem nobody budgets for

A ghost complete is a complete your sample platform recorded and your survey data does not contain. They are invisible until someone matches IDs, and...

 Research Operations

What the data quality benchmark says about your feasibility estimates

The Insights Association and GDQ benchmark pooled close to two million survey records from 51 companies. Most coverage quoted the removal rates. The m...

Put a quality layer in front of your next study

Tell us whether you buy sample or supply it. We set up your account, share sandbox links and walk through the first integration with your team.

Talk to us Read the docs