What Is Survey Incidence Rate and How Does It Affect Cost?
Incidence rate decides how many people you must screen to get one complete. Here is how to estimate it, and how it drives price and timelines.
In this article
If you only learn one number before asking for a sample quote, make it incidence. It drives price, timing and, more often than people admit, data quality.
Incidence rate (IR) is the share of people who start your survey and actually qualify. Screen 1,000 people, 150 qualify, your IR is 15%. Simple math, big consequences.
Key takeaways
- IR is qualified respondents divided by everyone screened. It is not the response rate.
- Halving the IR roughly doubles the number of people you have to reach.
- Low-IR studies attract fraud because the incentive is high and the screener is easy to guess.
- Estimate IR from real data, then confirm it in a soft launch before full field.
Why low incidence multiplies cost
Say you need n=500 completes.
| Incidence | People you need to screen | What that means |
|---|---|---|
| 60% | about 830 | Gen pop, quick field |
| 20% | 2,500 | Category buyers |
| 5% | 10,000 | Niche owners, specific conditions |
| 1% | 50,000 | Rare audiences, often not feasible online |
Every one of those screen-outs costs panel time and respondent goodwill. That's why CPI climbs as IR drops. A 5% IR study isn't a little more expensive than a 50% one. It's a different kind of project.
How to estimate IR before you field
Start with outside data: census tables, category penetration from syndicated sources, or your own CRM if you're a brand. Then adjust for your screener. Every extra qualifier ("purchased in the last 3 months," "primary decision maker," "not employed in marketing") cuts IR further, and the effects compound.
When in doubt, run a soft launch of 50 to 100 completes. Real IR shows up fast.
When actual IR comes in lower than quoted
It happens. A fair provider flags it early, shows you the screener drop-off by question, and gives options: loosen one criterion, extend field time, or re-quote. What you don't want is a provider who "finds" the completes anyway. Sudden rescues on a low-IR study are the most common place I see bad data sneak in. Our data quality checks get stricter, not looser, when incidence is low.
Cutting the cost of low-IR work
Use profiled sample where the provider already knows who owns a pet, drives an EV or manages IT purchases. That pre-targeting can lift effective IR a lot. Keep the screener short and hide the target with multi-select lists (more on that in our screener design post). And ask about feasibility by country before you lock quotas.
FAQ
What is a good incidence rate for an online survey?
Anything above 30% is comfortable for most online panels. Between 5% and 30% is normal for category studies. Below 5%, expect higher CPIs, longer field and extra verification.
Who pays when incidence is lower than expected?
It depends on the contract, but a good provider tells you as soon as the soft launch shows a gap and agrees the change with you before fielding more. Surprise invoices at the end are a sign of weak project management.
