Incidence, not budget, decides whether a narrow B2B survey is feasible. On low-incidence screens the achieved sample routinely lands well short of a 200 target, and the cost per complete runs at a large multiple of general-population work — a multiple that tracks incidence almost linearly.
What sample is actually achievable?
Often less than the brief specifies, and the gap is knowable before fielding rather than after. Feasibility is a function of how many people match the screen and can be reached, and both are estimable from the screen definition alone — before you decide how many responses the analysis actually needs.
The conversation worth having at scoping is what sample the screen supports, not what sample the analysis would prefer. Those are different numbers and only one of them is achievable.
Estimating it in advance is also a service to the analysis rather than only to the budget. Knowing the achievable sample before fielding lets you decide what the study can claim, and design the instrument accordingly — fewer closed questions, more open ones, and a reporting plan that does not depend on cuts the sample will not carry.
What does incidence do to cost?
Almost everything. Incidence is the share of contacted people who qualify, and at two percent you screen fifty people to get one complete — and you pay for all fifty, because screening is work whether or not it produces a respondent.
That is why cost per complete on a narrow B2B screen bears no relation to consumer survey pricing. The respondent fee is a minor component; the finding of the respondent is the cost.
How narrow is too narrow?
When the population that matches is smaller than the sample you need. That sounds obvious and it is routinely missed, because screens are written as a list of desirable attributes rather than as a definition of a countable group.
Each additional criterion multiplies rather than adds. Role, sector, company size, geography and recency together can reduce a population of tens of thousands to a few hundred — the same arithmetic a competitive intelligence screen runs into.
The multiplication also means the order in which criteria are relaxed matters. Dropping the recency clause and dropping the company-size band do not have the same effect on either the population or the quality of the evidence, and the choice should be made deliberately rather than by whichever is easiest to argue for. Recency is usually the one worth defending longest.
Why do consumer panels not work?
They are built for demographic screens, not occupational ones. A panel can reliably find two hundred people who bought coffee last week; it cannot reliably find two hundred who specify industrial valves, because it was never assembled to know that about anyone.
Where panels do return B2B completes at volume, the qualification is usually self-reported and unverified, which is a different problem and a worse one — and it is fatal to a pricing study.
When should you interview instead?
When the achievable sample is too small to quantify anyway. Twenty completes will not support a confidence interval, and presenting them as though they might is worse than not running the survey.
Twenty expert calls with the same people, however, will tell you why the market behaves as it does — which is frequently the question underneath the one that was asked.
The switch to interviews is also easier to sell internally when it is framed at scoping rather than after a failed field. A study that was always going to be qualitative is a method choice; one that becomes qualitative because the sample did not arrive is a failure, and the two produce identical evidence and very different receptions.