Primary Survey
how to answer the questions published data can’t
The most useful market questions usually haven’t been answered in a report yet. A well-built primary survey answers them directly. Here is what separates a survey that changes a decision from one that only fills a slide.
The market report on your desk says the segment grew 5% last year. It doesn’t say why buyers are switching suppliers, what a plant manager would pay for a faster coating, or whether mid-sized retailers plan to replace their checkout systems in the next two years. And your competitor bought the same report.
That is the gap a primary survey fills. A primary survey collects new data directly from the people whose decisions you need to understand: customers, buyers, users and industry specialists. You choose the questions, the respondents and the market definition, so the evidence fits the decision rather than the other way round.
Done well, a survey can settle an argument in a steering committee. Done badly, it produces confident charts built on the wrong people. The difference is almost always decided before the first response comes in.
At a glance
- Use a primary survey when published data is too old, too broad or too widely held to support the decision in front of you.
- Who answers matters more than what you ask. A well-written questionnaire can’t rescue a sample of people who don’t make the decision you’re studying.
- Sample size sets precision. At n=100, a 50% result carries a margin of error of about ±10 points. At n=400, it’s about ±5.
- In B2B work, incidence drives cost and timing. If 5% of the people you reach qualify, a sample of 200 means screening around 4,000.
- Start from the decision, then work back to the questions. Surveys designed the other way produce interesting data that nobody acts on.
Why published research runs out
Syndicated reports are built for the average buyer. They size markets in categories someone else chose, usually a year or more after the fact. That is a valuable context and it is where every good project starts.
But strategy questions are rarely average. A pricing team needs to know how procurement managers at mid-tier hospitals would react to a 7% increase. A diligence team needs to know whether a target’s customers plan to renew. No report answers those questions, because nobody has asked them yet.
Published data has a second limit: everyone else can read it too. If the evidence behind your recommendation is available to every competitor, it won’t give you an edge in a pitch, an investment committee or a launch plan.
Four decisions a primary survey can inform
Is the market big enough
Bottom-up sizing starts with buyers: how many there are, how often they buy and how much they spend. A survey supplies the inputs that top-down estimates have to assume.
Will customers pay for it
Concept and pricing tests measure purchase intent, willingness to pay and the trade-offs buyers make between features. Techniques such as Van Westendorp price sensitivity and MaxDiff turn opinions into numbers you can model.
Why do customers choose a competitor
Purchase criteria, supplier ratings, switching intent and share of wallet, measured across the whole market rather than only among your own customers, who already chose you.
Does the investment thesis hold
In commercial due diligence, a survey of a target's customers and its competitors' customers tests retention, satisfaction and pricing power before the deal team commits capital.
B2B and B2C surveys are different jobs
Consumer and business surveys share a method but solve different problems. In a B2C study the people exist in large numbers and the challenge is getting a clean, representative sample. In a B2B study the challenge is finding the people at all.
A note on terms, because they are often confused. CAWI (computer-assisted web interviewing) is an online survey the respondent completes alone. CATI (computer-assisted telephone interviewing) is run by an interviewer over the phone. CAPI (computer-assisted personal interviewing) is face to face, usually with a tablet.
Most survey failures happen before the first response
When a survey disappoints, the analysis usually takes the blame. In our experience the problem almost always started earlier, in one of three places.
Expert interviews make surveys sharper
A survey measures how common something is. It is weaker at explaining why. That is why the strongest projects pair the numbers with a handful of in-depth interviews.
Before fieldwork, three to five expert calls help you use the language your respondents use, list the answer options that actually exist in the market and drop questions nobody can answer accurately.
After fieldwork, a few follow-up calls explain the results that surprised you. If 40% of buyers say they plan to switch suppliers, someone who runs procurement in that sector can tell you whether to believe it.
This is where an expert network earns its place. The same people who can explain a market on a call can help design a survey that measures it properly.
Precision rises slowly with sample size
Note: B2B samples drawn from small, known populations can apply a finite population correction, which narrows the margin.
The practical rule: decide which groups you need to compare before you set the sample, then size each group, not just the total. Quadrupling a sample only halves its margin of error, so it pays to be precise about where the extra responses go.
AI Adoption Survey
The client’s question: How do organizations evaluate and adopt AI, what holds them back, and what builds confidence to invest?
Who we surveyed: Conducted in 2025–2026, the study combined 73 in-depth interviews and 50 written surveys, covering senior professionals across 10+ sector contexts, including banking, healthcare, manufacturing, retail and telecommunications. Interviewees brought perspectives from AI, technology, cybersecurity, governance and marketing.
What the research enabled the client to decide: The findings provided a basis for identifying which decision-makers to engage, which adoption barriers to address, and what evidence and messaging buyers need to move from interest to implementation.
Primary vs secondary research: use both, in order
The efficient sequence is to start with secondary research to establish what is already known, then design the survey to answer only what isn’t. That keeps questionnaires short, which improves completion rates and data quality.
How we run a primary survey
Typical B2B projects run from brief to final report, depending on incidence and sample size.
Sectors we cover, and the questions we're asked
We also run projects in education, financial services, and media and entertainment.
Frequently Asked Questions
A primary survey is a research method that collects new data directly from a defined group of respondents, such as customers, buyers or industry specialists. Unlike secondary research, which reuses existing reports and databases, a primary survey is designed around a specific question, so the results fit the decision they need to support.
Use a primary survey when published sources are too old, too broad or too widely available to answer your question. Common cases are pricing and concept tests, purchase-criteria studies, customer research in due diligence and market sizing for niche B2B segments where reliable published data doesn’t exist.
It depends on the precision you need and the groups you want to compare. At n=100, a 50% result has a margin of error of about ±10 points at 95% confidence. At n=400, it’s about ±5. Size each subgroup you plan to compare, not just the total sample.
B2C surveys ask consumers about their own preferences and usually draw large samples from online panels. B2B surveys ask professionals about decisions they make at work. Their populations are smaller and harder to reach, so recruitment, screening and respondent verification matter more than sample size.
They are survey data collection methods. CAWI (computer-assisted web interviewing) is an online survey completed by the respondent. CATI (computer-assisted telephone interviewing) is conducted by an interviewer by phone. CAPI (computer-assisted personal interviewing) is conducted face to face, usually with a tablet.
Most B2B surveys take 2 to 6 weeks from brief to report, depending on the scope and volume. The biggest factor is incidence, the share of people contacted who qualify. Low-incidence audiences, such as specialist buyers in a single country, take longer to recruit than broad professional groups.
We screen every respondent with questions about their actual responsibilities, check stated roles against [verification source], and remove responses that fail quality checks for speed, consistency or duplication. Recruiting through an expert network also means many respondents have already been profiled and verified.
Cost depends mainly on sample size, incidence, questionnaire length and method. A short online survey of a broad audience costs far less than a telephone study of hard-to-reach specialists. [Add a starting price or typical range if TEP is willing to publish one.]
Five questions to answer before you commission a survey
1
What decision will this survey change, and by when?
2
Who exactly holds the answer, and roughly how many of them exist?
3
Which groups will we need to compare, and how many responses does each need?
4
What do we already know from published sources?
5
How will we confirm that respondents are who they say they are?
If you can answer the first two, you’re ready to brief a survey. If you can’t, a few expert calls are usually a better place to start.
Send us the decision, not just the questionnaire
Tell us what you need to decide, who you think holds the answer and when you need it. A research manager will come back with a suggested sample, method and timeline, usually within [X] hours.