What are Omnibus Surveys?
On this page
An omnibus survey shares one fielded questionnaire across several clients, helping distribute setup and sampling costs. It can provide descriptive opinion or awareness estimates, and some providers support randomized questions or methodological experiments. Before using it for a pricing decision, check the design, target population and response measure: a hypothetical choice does not directly measure actual demand.
- An omnibus survey is a shared questionnaire where multiple clients each buy a few questions on one fielded instrument, run on a fixed cadence by the provider.
- Compare dated quotes for the same market, currency, respondent count, question complexity, extras and turnaround; billing units differ by provider.
- Versta's April 2025 guidance suggests comparing custom quotes when a brief exceeds about five or six questions, but this is vendor guidance rather than a universal break-even point.
- Omnibus and custom surveys describe fielding formats. Either may support randomized stimuli when the provider and study design allow them.
- An intervention claim needs an identified comparison and explicit endpoint; actual demand requires relevant behavioral evidence.
What is an omnibus survey, and how does it work?
SSRS's April 2025 description, checked in October 2026, specifies twice-monthly Friday-to-Monday fielding and a standard sample of 1,000 US adults. It randomizes placement of client topics and offers methodological experiments. Confirm whether the provider can support the proposed stimulus assignment, eligible audience, instrument and analysis; shared scheduling and other clients' topics remain operating constraints.
How much do omnibus surveys cost, and how fast do they run?
The SSRS page lists USD $1,000 per question for the first three questions, then $750 per question, with a three-question minimum and 15-question maximum. Three standard questions therefore cost $3,000; five cost $4,500 before extras. YouGov's serviced-survey page advertises coverage across 70+ markets, but it does not substantiate the $1.50 self-service unit price used in the older comparison. Obtain a matching dated quote rather than compare per-response and per-question prices as if equivalent.
When does a custom survey beat omnibus?
Versta's April 2025 guidance suggests that a custom survey can become competitive above roughly five or six questions or for a narrower audience. Actual economics depend on incidence, targeting, complexity, sample, schedule and included analysis. Request both routes against the same brief; custom design control may also matter when the study needs specific experimental or measurement features.
| Omnibus | Custom Survey | Randomized Experiment | |
|---|---|---|---|
| Cost | Shared fielding, often priced per question; include minimums and extras | Quote the same target, sample and instrument | May be implemented within either fielding format; quote the required design and endpoint |
| Timing | Provider fielding calendar and submission cutoff | Recruitment, design and review schedule | Assignment, measurement and validation schedule |
| Evidence | Description or a supported randomized task, depending on design | Description or experiment, depending on design | Identified contrast on the measured response, with design-specific uncertainty |
| Fit | Brief compatible with the shared audience, slots and instrument | Specialized coverage, depth or control requirements | Intervention questions with clear assignment, baseline and response source |
Can an omnibus survey tell you what people will actually do?
A question can report opinion, preference, or recalled behavior, with its measurement limits. A randomized survey task can estimate an effect on the shown response. Neither format automatically proves what people will purchase or do later; validate the actual endpoint when the decision requires it.
Does sitting next to unrelated questions affect your data?
Question order and surrounding topics can influence responses. SSRS explicitly randomizes client-topic placement to reduce order effects. Ask any provider how placement, stimulus order, respondent burden, duplicate topics and language versions are handled. Randomizing topics is different from randomizing the intervention, and no single ordering procedure removes every context effect.
What proves a causal effect instead of a stated opinion?
For an intervention test, define baseline and alternatives, eligible assignment units, randomization or other identification assumptions, and the response measured. Discrete-choice models, Mixed Logit and ICLV are estimators, not sources of identification on their own. A multinomial logit's IIA restrictions may be unsuitable for some substitution questions; select and check the model against the design and data rather than assume Mixed Logit is mandatory.
For a Subconscious comparison, agree on the generated-choice endpoint and relevant independent human evidence. The July 2026 working paper reports mean Spearman correlation on estimated choice-parameter ranks of .73 across 43 design-filtered studies. This is aggregate replication evidence, not sales accuracy or validation of the proposed omnibus question; public per-study replication data is not available.
What should a buyer do next?
Bring the target audience, question count, intended endpoint, assignment needs, deadline and acceptance criteria to procurement. Compare omnibus and custom quotes on the same scope, and confirm the provider's support for the required design. For a generated-choice comparison or validation plan, discuss the decision with Subconscious.