Can we use predictions to personalize course recommendations?

We recommend against using gender to steer course choices. Recommendations based on stated interests and past activity are more accurate and avoid reinforcing stereotypes about who studies what.

Why does Andrea return different results by country?

Some names lean differently by country. Andrea is commonly male in Italy and commonly female in the United States, so pass a country code when you have one.

How are batch lookups billed?

Each batch item costs 1 credit, or 3 credits if it requests AI. Batch items default to ai_mode off, and the whole batch counts as one request toward the rate limit.

Gender prediction · Education and edtech

Aggregate gender analytics for adult education programs

Continuing education teams, bootcamps and professional certificate programs often ask whether their marketing reaches a broad mix of adult learners. Where people have not been asked, or chose not to say, Finowl gender predictions estimate how a first name is typically used. Rolled up across thousands of anonymized inquiries, that estimate can show whether a campaign for a coding course or a nursing refresher skews heavily in one direction.

A prediction is a statistical estimate, not how a person identifies, and it is wrong for some people. Use it only for aggregate reporting about outreach, never to decide admissions, scholarships, pricing, grading or access to any course. Let learners state or correct their own gender in their profile, and avoid any processing of data belonging to minors. Self-reported answers should always override a prediction.

Try it in the sandbox About gender

USE CASES

Where it fits in education and edtech

Campaign reach reporting
Estimate the aggregate gender mix of inquiries generated by each adult education campaign, so marketing teams can see whether messaging and channels reach the broad audience they intended.
Program outreach reviews
Compare anonymized, aggregate estimates across programs such as data science and early-years teaching to inform where outreach to underrepresented groups could be broadened, without touching individual decisions.
Backfilling historical reports
Older enrollment exports often lack self-reported fields. Batch predictions on first names can produce rough aggregate trends, clearly labeled as estimates, for institutional reporting on outreach.
Survey weighting checks
When an alumni or course feedback survey has low response, compare the estimated aggregate mix of respondents with the invited population to see whether results may be skewed.

HOW TO INTEGRATE

The workflow

1
Export first names and an optional country code for adult inquiries or alumni, removing any records that belong to minors.
2
Send up to 50 items per call to POST /v1/gender/batch with opaque ids that do not reveal identity.
3
Keep only identified results above your confidence threshold and count unknown results as their own category.
4
Aggregate counts by campaign or program and delete the per-record predictions once the report is produced.

POST /v1/gender/batch

curl https://api.finowl.co/v1/gender/batch \
  -H "Authorization: Bearer $FINOWL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "items": [
    {
      "id": "inq-0001",
      "type": "name",
      "value": "Andrea",
      "country": "IT"
    },
    {
      "id": "inq-0002",
      "type": "name",
      "value": "Samuel",
      "country": "US"
    },
    {
      "id": "inq-0003",
      "type": "email",
      "value": "k.okafor@example.com"
    }
  ]
}'

Estimates gender for anonymized adult inquiries in one batch. Results are aggregated by campaign for outreach reporting and never attached to an admissions or enrollment record.

GOOD PRACTICE

Before you ship

Never use predictions for admissions, scholarships, grading, pricing or course access. They are aggregate estimates of name usage, and acting on them for individuals could cause real harm.
Exclude data belonging to minors entirely, list gender inference in your privacy notice, and keep only aggregated counts after each reporting run.
Let learners self-identify, include non-binary and prefer-not-to-say options, and always let a stated answer replace any estimate in every report you produce.

Read the API reference for every field and error code. Each lookup costs 1 credit (3 when it requests AI); see pricing.

A FEW GOOD QUESTIONS

Common questions