Quick answer: Break the audience into profile filters (age, gender, region, ethnicity) and behavioural criteria, estimate how common each behaviour is in each region, then multiply down through every layer and add a buffer. If the respondents you need are a large share of the realistic pool, relax the quotas or widen the definition before you launch.
What is incidence, and why does it decide feasibility?
Incidence is the share of a population that qualifies for your study. If one in five adults does the behaviour you care about, incidence is low enough that finding hundreds of qualifying respondents takes real effort. If most adults do it, feasibility is rarely a problem.
Feasibility is simply incidence combined with how many people you can reach. A niche audience is not impossible, but it needs to be checked before the questionnaire is written, not after quotas stall in the field.
How do you break a complex audience into layers?
The mistake is to treat the audience as a single sentence. Write it out layer by layer, and separate what is known from what must be estimated.
| Layer | Example | How to treat it |
|---|---|---|
| Profile filter | Age band, gender, state or region | Usually known from panel profile data, so it can be applied up front |
| Behavioural criterion | Uses a category weekly, or owns a specific device type | Estimate incidence from past studies, desk research or a short screening pilot |
| Brand criterion | Currently uses, or has switched away from, a named brand | Often the lowest incidence layer, so check this first |
| Exclusion | Remove a group from the base, such as a religious or ethnic segment | Use profile data to define the remaining base before applying other layers |
| Quota layer | Minimum respondents per region, age or ethnic group | Each quota needs its own feasibility check, not just the total |
Why treat Peninsular and East Malaysia separately?
Consumer behaviour, languages, ethnic composition and the number of reachable respondents differ between Peninsular Malaysia and East Malaysia. A single national incidence figure hides this. A behaviour that is common in one region may be rare in another, and the pool of people you can actually recruit is often smaller in East Malaysia.
The practical rule is to calculate incidence separately for each regional base, then add the results. If a regional quota matters to the decision, give it its own feasibility line and its own buffer.
What do interlocking quotas do to your numbers?
Interlocking quotas combine several characteristics in one cell, such as women in a given age band, in a given region, who use a given brand. Each added layer divides the pool again.
Here is an illustrative example, not a real figure. Suppose one in five people in a region does a behaviour, one in four of those uses a particular brand, and one in three of those falls in the target age band. Multiplying down, only about one in sixty adults qualifies. Add a gender split and two language groups, and the cells become very thin.
This is why it pays to ask whether you need every combination, or whether minimum totals per characteristic would serve the decision just as well. Nested minimums are far easier to fill than a full grid of interlocking cells.
How big should your buffer be?
Incidence estimates are assumptions, and assumptions drift. Build in a safety margin for drop-outs, quality removals and estimate error. A modest buffer is normal practice, but widen it for low-incidence audiences, for regional quotas and for any layer based on a guess rather than data.
Treat any assumed overlap between groups with caution. Do not assume that everyone who buys a product is also the decision maker, or the reverse, unless you have evidence. If the study depends on that overlap, ask a screening question instead.
Before you commission, run through this checklist:
- Write the audience as a list of layers, not a sentence.
- Mark each layer as known, estimated or guessed.
- Check regional bases separately.
- Decide which quotas are essential and which are nice to have.
- Agree a fallback if a cell stalls, such as merging age bands.
- Allow extra fieldwork time for narrow cells.
What if the audience turns out to be too narrow?
You have several options before giving up on the study. Widen an age band, merge adjacent regions, replace interlocking cells with nested minimums, extend fieldwork, or loosen a brand criterion from current use to past use. If the audience is genuinely hard to reach online, a qualitative approach or a specialist recruiter may be the better fit for that segment.
Timing matters as much as scope. Narrow audiences take longer to fill, so read our note on how long market research takes and plan the schedule backwards from the decision. For the cost side, what drives market research cost in Malaysia explains why sample size and incidence move the budget more than most teams expect.
Where does Vase.ai fit?
We build Vase.ai, and feasibility is a normal part of how studies are scoped on it. Send us the research question, the audience and the date you need results, and we will suggest a study design and sample, and flag feasibility risks up front.
The earlier you share the target audience, the easier it is to suggest quotas that can actually be filled.
Frequently asked questions
How do I estimate incidence if I have no data?
Use past studies, public statistics and a short screening pilot. Treat the estimate as a range, not a single number, and add a larger buffer where the estimate is a guess.
Can I reach rural audiences with an online survey?
Often partly, depending on connectivity and panel depth in that region. For audiences that are hard to reach online, consider a local field agency for face-to-face interviews, which covers a different job.
Should I set quotas for every combination of age, gender and region?
Only if the decision truly needs it. Full interlocking grids are the most common cause of stalled fieldwork. Nested minimums per characteristic are usually enough and much faster to fill.
