Quotas: Control Sample Structure

Quotas control the composition of valid completes. They do not decide who may answer; they decide how many qualified respondents are needed in each category.

For a target of 300 completes, for example:

• Gender: 50% men / 50% women;
• City: 40% Shenzhen / 30% Guangzhou / 30% Beijing;
• Education: 60% bachelor’s / 40% associate.

Wejot controls the marginal distribution of every dimension at the same time. You do not configure a cross-combination matrix such as “men in Shenzhen with a bachelor’s degree.”

Current scope

Advanced quotas currently apply to standard survey panel distribution. AI interview samples do not yet support them.

Quotas versus screening

A common setup screens out non-buyers first, then controls the distribution of purchase frequency.

Available quota dimensions

Panel attributes

Gender, age, location, and other profile attributes can be quota dimensions.

• Without a distribution-range restriction, every available value remains a quota bucket. This allows random distribution with a balanced quota such as 50/50 gender;
• If the range restricts an attribute, only its remaining selected values become buckets;
• If only one value remains, that attribute is not offered as a quota dimension because there is nothing to allocate.

Survey questions

Choice questions are bucketed by option. Scale, rating, slider, and number questions are automatically split into three ranges.

When the same question is screened, its quota buckets narrow to answers that can still qualify.

Custom screener questions

A complete custom screener with options can become a quota dimension. Removing a referenced screener also removes its quota dimension so that the task does not keep an unresolvable configuration.

How multiple dimensions work

Wejot uses independent marginal quotas. Every dimension must sum to the total, but dimensions are not multiplied into a visible cross-table.

For a target of 100:

• Gender: 50 men / 50 women;
• City: 40 Shanghai / 30 Beijing / 30 Guangzhou.

The target is to satisfy both distributions at once. The platform coordinates the cross-combinations during collection; you do not specify how many Shanghai men are required.

The current interface no longer imposes a fixed four-dimension limit. Availability depends on the eligible dimensions in the task. More dimensions can still make simultaneous fulfillment harder, so control only variables that matter to the conclusion.

Percentage and headcount modes

Switching units converts the current allocation and corrects integer rounding while preserving the structure as closely as possible.

Lock, distribute evenly, and auto-balance

• Lock fixes a bucket so automatic changes do not move it;
• Distribute evenly splits one dimension equally;
• Auto-balance keeps locked buckets fixed and distributes the remainder among unlocked buckets;
• Use the allocation bar or enter a number directly.

Set and lock hard requirements first, then balance the remainder. Payment is blocked until every dimension sums to 100% or the target headcount.

What happens when a quota fills

Once a bucket is full, later responses belonging to that bucket are screened out:

• they do not count as valid completes;
• they do not incur a valid-sample charge;
• collection continues for buckets that are still open.

This is now the fixed behavior so the final structure matches the bucket counts purchased. The former “overflow into total” option has been removed.

How quotas affect price

Attribute quotas: price each bucket

Profile attributes have explicit tag prices, and buckets cannot substitute for one another:

bucket subtotal = bucket headcount × bucket tag unit price
attribute-quota total = sum of all bucket subtotals

Percentage targets are first converted to rounded headcounts.

For 100 respondents, with men priced at ¥2 and women at ¥1 and a 50/50 quota:

Men: 50 × ¥2 = ¥100
Women: 50 × ¥1 = ¥50
Gender quota total: ¥150

The payment breakdown shows each bucket’s headcount, unit price, and subtotal. If the same attribute is also in the distribution range, it is priced only as quota buckets, not twice.

Survey-question and custom-screener quotas: scarcity uplift

These dimensions do not have direct tag prices, so difficulty is estimated from the narrowest non-zero bucket:

• below 25% of the total triggers the lower scarcity tier;
• below 10% triggers the higher tier;
• an existing screening fee is adjusted; without one, a fixed scarcity fee is used.

Attribute quotas are already priced by real tag costs and do not receive this extra uplift. See How Pricing Is Calculated for exact tiers.

Recommended workflow

1. Enter the total completes;
2. Set the distribution range;
3. Complete screening;
4. Enable quotas and select the first dimension;
5. Choose percentage or headcount and lock hard requirements;
6. Add only necessary dimensions;
7. Resolve balance errors, conflicts, and narrow-bucket warnings;
8. Verify every bucket price in the payment confirmation.

FAQ

Can I set a gender quota without targeting gender?

Yes. An unrestricted distribution range keeps all gender values available, which supports random distribution with a 50/50 quota.

Why is an attribute or option missing?

Targeting or screening may have narrowed it to a single possible value, leaving nothing to allocate. The question may also be outside the supported choice or numeric types.

Are multiple dimensions cross quotas?

They are not a user-editable cross quota. You set marginal targets by dimension; the platform makes them hold simultaneously.

Why did a small bucket change the quote sharply?

Small buckets are harder to fill. Attribute buckets use their actual tag prices; survey-answer or custom-screener buckets use a scarcity uplift. Consider merging ranges, relaxing eligibility, or increasing the bucket share.

Other names used in the industry

Wejot currently uses independent marginal quotas: each dimension has its own targets and the platform coordinates them simultaneously. A cross or interlocking quota directly targets combinations such as “30 men in Shanghai”; that is not the user-configured model on this page.

Quota sampling is not stratified sampling

Quota sampling is usually a non-probability method that fills category targets. Stratified sampling divides a known population into strata and samples with a probability design. Both can control group proportions, but their statistical assumptions differ.