Disclosure first: ResearchDart is a routing layer between many suppliers and buyer surveys. We earn more when you route more. Read this with that in mind.
There are two broad ways to source respondents for an online survey. Buy from one panel that recruits, profiles and rewards its own members. Or connect your survey once to an exchange or router, and let several suppliers send respondents through it. Neither is better in general. Several kinds of study are better served by a single panel, and pretending otherwise would be selling.
How they differ
| Single panel | Exchange or router | |
|---|---|---|
| Reach for rare audiences | One panel's members | Several suppliers combined |
| Speed | One panel's responsiveness | Several sources in parallel |
| Consistency wave to wave | High | Depends on keeping the supplier mix fixed |
| Duplicate risk | Low inside the panel | Higher, because people join several panels |
| Integration | One per panel | One for every supplier |
| Quality visibility | What the panel chooses to report | Outcomes and reasons comparable across suppliers |
When a single panel is the better choice
Trackers and longitudinal work. Changing the source between waves can move your results more than the market moved. A stable panel removes that variable. You can hold a supplier mix fixed on a router, but a single panel does it by default.
Recontact studies. If you need the same people again, you need the panel that holds their identities.
Studies that depend on deep profiling. When the panel's own long-term data about members matters to targeting or analysis, a router cannot supply it.
Simple, high-incidence studies one good panel fills comfortably. Adding suppliers adds duplicate risk and reconciliation work for no gain.
When an exchange or router is the better choice
Low-incidence or niche audiences no single panel can fill in time.
Tight deadlines, where parallel sources shorten field.
Multi-country work, where the best supplier differs by market.
Programmatic buying, where projects are created and managed through an API.
When you want to compare sources. Quality terminates, screening flags and reversals by supplier on the same study are the fairest comparison you will get. That is harder when every source is its own separate project.
The costs of routing across suppliers
Duplicates. People belong to several panels. Deduplicate across the whole project, by respondent ID and by device, and use screening that detects one person behind many accounts.
Uneven quality. Suppliers differ, and some sub-sources are bad. In the 2025 Insights Association and GDQ benchmark, suppliers removed 13.7% of respondents before or during the survey globally. You need per-supplier reporting, not a blended average.
More reconciliation. More suppliers means more parties to reconcile with. Shared GDQ removal codes make that manageable; free-text reasons do not.
What most teams end up doing
A trusted panel as the core of a study, with additional suppliers routed in to fill hard cells or speed up the end of fieldwork. A router makes that practical because the core panel is just one allocation among several, with the same endlinks and the same session records.
Questions to settle before choosing
- How rare is the audience, and can one panel realistically deliver it on time?
- Does this study have to be comparable with earlier waves?
- How will duplicates across sources be detected?
- Who will look at quality by source, and how often?
- Do you need to recontact anyone?
If the honest answers point to one panel, use one panel.