# Niche job board launch worksheet
All numbers and targets below are yours to choose. This is not a forecast.

## Audience and scope
- Reader and decision the board helps:
- Country/region and work arrangements:
- Include these occupations and employment types:
- Exclude these roles, campaigns and locations:
- Required fields before publication:
- Cases requiring human review:

## Evidence of supply (repeat on several dates)
| Date | Query | Returned rows | Relevant distinct jobs | Unique employers | Known omissions | Rejection reasons |
|---|---|---:|---:|---:|---|---|
| | | | | | | |

## Launch implementation
- Source/data plan and permitted usage:
- Stable job identity and canonical page path:
- Backfill choice, start time and continuation state:
- Update, expiry and out-of-scope handling:
- Application destination fallback:
- HTML sanitizer and allowed tags:
- Empty-search behavior:
- Page/index/cache removal procedure:
- Owner and alert for failed syncs:

## Monthly usage estimate
- Initial returned records:
- New jobs returned:
- Changed records or change events returned:
- Separate detail calls:
- Rechecks and retry allowance:
- API request total:
- Hosting, email, editorial and support costs:
- Plan headroom and expected total cost:

## Reader validation
- Five people to interview (keep private):
- Tasks they will attempt:
- What made a job irrelevant:
- What information they could not find:
- What would make them return:

## Release gate
[ ] A relevant job can be found and its application destination works.
[ ] No salary or location is invented to fill a blank.
[ ] A closure disappears from search, cache and queued alerts.
[ ] A replayed batch creates no duplicate records.
[ ] A failed request preserves the checkpoint.
[ ] Mobile pages, keyboards and empty states work.
[ ] Public job markup matches visible facts; indexable pages have a purpose.
[ ] Analytics distinguish source traffic, searches and application-link clicks.
[ ] Targets and a review date are written down before launch.