Omnibus surveys: when renting three questions beats a study

An omnibus survey is a shared survey: many clients each pay to add a few questions, and the provider fields the combined questionnaire on a fixed schedule to one general-population sample, usually about 1,000 adults. Omnibus surveys are the cheapest way to put a handful of closed questions in front of a nationally representative sample, with results in one business day to about a week.

The catch is in the word "general". An omnibus is excellent when your question is about everyone, and close to useless when it is about a small slice of everyone.

An omnibus survey is one questionnaire with many paying passengers

NORC at the University of Chicago, which runs the AmeriSpeak Omnibus, describes it plainly: multiple clients contribute questions to a shared survey, and each pays only for its own questions. The provider recruits, fields and weights the sample once and spreads that cost across everyone on board, so you get a national sample without paying to recruit it. Your three questions sit between someone else's questions about snack brands and someone else's questions about streaming services. (The respondents do not get a say in the seating plan.)

An omnibus is a cousin of syndicated research, where one study is sold to many buyers, except that you write your own questions and only you see the answers.

Published prices run $500 to $1,000 per question for 1,000 US adults

Three well-known US providers publish their rates, each for about 1,000 adults aged 18 and over.

"Question" has a specific meaning in these price lists. NORC counts a unit as a simple closed question, such as a yes or no item, a checklist of up to 10 answers or a grid of up to four statements. YouGov counts a question as up to 10 single or multiple choice answers, and charges a scale question per three statements on the same scale. A question with 14 answer options or eight grid rows can quietly become two or three billable units.

Extras cost extra. NORC lists $350 for a custom banner table (a cross tabulation by groups you choose), $300 for each additional profile variable and $1,500 for coding open-ended answers. SSRS charges $100 per question to add 25 Spanish-language interviews. These are list prices from the providers' own pages, so confirm the current rate card before you budget.

Short version: a four-question omnibus wave for 1,000 US adults costs roughly $2,300 to $4,000 at published prices, depending on the provider.

Ask how the panel was recruited before you compare prices

Pew Research Center found that online opt-in polls, where people volunteer through ads or social media, attract "bogus respondents" who rush through surveys for the reward. In one Pew experiment, 12% of opt-in respondents under 30 claimed to be licensed to operate a class SSGN nuclear submarine. The real share of Americans with that license rounds to 0%.

A probability panel recruits people by random selection, for example from residential addresses, so volunteers cannot sign themselves up. SSRS says of its panel that no one can volunteer and members are randomly selected and invited.

Pick a probability panel when the number itself will be quoted, for example in a press release or an investor deck. An opt-in sample is fine when you only need a direction, such as which of two names people prefer. Either way, ask every provider how its respondents got onto the panel.

1. Use an omnibus when the question is about all adults

An omnibus works best for short, closed questions where the right population is the general public:

• Awareness. Which of these brands have you heard of? A brand awareness survey covers how to order unaided and aided questions.
• Incidence. What share of adults own a dog, bought a used car this year, or use a meal kit? This is often the most valuable omnibus number, because it sizes the market before you pay to survey it.
• Simple attitudes. Do you trust online reviews?
• Tracking. The same three questions every quarter, to see whether a campaign moved anything.

Here is a made-up example. A small company sells a refillable cleaning spray under the brand name Dropwell. To measure awareness before and after a TV campaign, it puts four questions on an omnibus: unaided awareness, aided awareness with eight brands, brands bought in the past year, and where people buy cleaning products.

At YouGov's published rates for 1,000 adults, that is $300 plus 4 times $500, or $2,300 a wave. At SSRS, it is 3 times $1,000 plus $750, or $3,750. At NORC, it is 4 times $1,000, or $4,000. Two waves, one before and one after the campaign, cost $4,600 to $8,000. Compare that with the published custom survey costs in market research pricing.

2. Skip the omnibus when your buyers are a small slice of the population

Every omnibus respondent is a random adult, so if few of them are your buyers, few of the answers are useful.

Pew's explanation of margin of error puts numbers on it: a sample of 1,067 gives about plus or minus 3 points, while a subgroup of about 160 gives about plus or minus 8.

Back to the made-up Dropwell example. Say the company also wants the views of professional house cleaners, and assume 2% of adults clean homes for a living. Out of 1,000 respondents, that is about 20 people. The usual formula for a 95% margin of error on a 50% answer is 1.96 times the square root of 0.25 divided by the sample size. For 20 people that comes to about plus or minus 22 points, which means "somewhere between 28% and 72% of cleaners like refill pouches". That is a coin toss with a decimal point.

B2B audiences are the extreme case. Hospital procurement managers and software company CFOs barely appear in a random sample of adults, so start with conversations with people who do the job.

3. Skip the omnibus when you have more than about ten questions

Per-question pricing adds up, and SSRS caps clients at 15 questions per wave.

Run the numbers for 12 questions at the published rates. NORC comes to 5 times $1,000 plus 5 times $850 plus 2 times $750, or $10,750. SSRS comes to 3 times $1,000 plus 9 times $750, or $9,750. YouGov comes to $300 plus 12 times $500, or $6,300. At that size you are paying custom-survey prices for a survey you do not control.

A dedicated market research survey gives you a screener that admits only your target buyers and follow-up questions that depend on earlier answers.

Short version: use an omnibus for up to about ten closed questions about the general public. Commission a custom survey for anything longer or narrower.

4. Write questions that survive being someone else's question 14

Your questions arrive mid-questionnaire with no introduction, so write each one to stand on its own.

1. Put the context inside the question. "Before today, had you heard of Dropwell, a refillable cleaning spray?" works anywhere. "Had you heard of it?" works only if the previous question was yours.
2. Ask one thing per question. Pew's guide to writing survey questions warns against double-barreled questions, which ask about two things at once and produce answers nobody can interpret. "How satisfied are you with the price and quality?" is two questions. On an omnibus it is also two invoices.
3. List every reasonable answer. In one Pew poll, 58% chose "the economy" as their top voting issue when it was offered as an option, versus 35% who named it unprompted. Make the list complete and keep the categories from overlapping.
4. Ask for your questions to be placed together. Pew documents order effects, where an earlier question changes answers to a later one. You cannot control what other clients ask, so ask the provider to keep your block together and rotate your answer options.
5. Use the demographics you already get. NORC includes age, sex, education, race and ethnicity, and household income in its standard banner table. Do not spend a paid question asking for a person's age.

Your next step

Write down the one number you need and who it is about. If it is about all adults and fits in four closed questions, book the next omnibus wave. If it is about a small professional group, talk to some of them first: Instant Expert finds people matching a description (or takes the people you name) and sends them paid invitations for a short call or a written answer, and you pay only if someone books or answers. For help designing the study itself, start with the directory of consumer insights experts.