Synthetic Onboarding — Benefits Enrollment Data

What this document is

The benefits enrollment form is where a new hire chooses coverage: a medical plan or a waiver, a coverage tier for employee, employee plus spouse, employee plus children or family, dental and vision elections, the per-paycheck premium for each, and a Section 125 authorisation to take the premiums pre-tax. It drives payroll deductions from the first cheque and determines what a payer sees when the employee's first claim arrives.

Why generate synthetically

Enrollment forms are conditional documents pretending to be flat ones. Electing a plan unlocks a tier and a premium; waiving blanks all three and leaves the labels behind. That branching is the whole difficulty, and it is invisible in a corpus that only contains enrolled employees — which is what a vendor's internal sample almost always is, because employees who waive rarely generate a support ticket. Generating the form lets you specify how often each branch appears instead of inheriting whatever the sample happened to hold.

What makes synthetic data useful

Elections follow a realistic benefits mix rather than a uniform draw: 81.4% of employees enroll in a medical plan and 18.6% waive, dental is elected by 69.3% and vision by 61.2%, and 73.6% authorise pre-tax treatment under Section 125. Premiums are plan-consistent — medical premiums cluster at $82, $176, $228 and $315 per pay period depending on plan and tier, dental between $12 and $18, vision between $6 and $9 — so a validator can be trained to notice a premium that does not belong to the elected tier.

Training challenges

The conditional cascade is the core problem: on the 18.6% of forms where medical is waived, the plan tier group is unmarked and the medical premium field is empty while every label stays on the page, so an extractor with a positional prior fills in a plan the employee explicitly declined. Three separate mutually exclusive checkbox groups — plan, tier, and each ancillary election — mean group-level correctness matters more than per-box accuracy. And the three premium fields are small currency values in similar positions with very close magnitudes for dental and vision, which is where they get swapped.

Who uses this data

HR-tech onboarding and benefits-administration platforms capturing elections from scanned forms, payroll integrators setting up deductions, benefits brokers and enrollment-outsourcing firms processing paper at open-enrollment volume, and payer-side eligibility teams tracing a member's coverage back to the election that created it.

Document complexity profile

18 fields on a single page: 6 text and 12 checkbox targets forming several mutually exclusive groups — medical plan, coverage tier, dental, vision — plus three premium fields and a Section 125 authorisation, with one label-to-value relation per field. 1 arithmetic binding and 2 function calls, at maximum expression depth 2. Premium fields are conditionally present, gated on the corresponding election.

Key stats from our synthetic corpus

Quantitative characteristics of the Onboarding — Benefits Enrollment documents our generator produces.

How this document co-occurs with others

Rates at which identities in our corpus that produce a Onboarding — Benefits Enrollment also produce other documents.

Election rates and premium values above are corpus-derived: they were computed over the 641 employment-eligible identities inside a local synthetic corpus of 1,000 identities generated by SymageDocs' World Simulation Engine at seed 20260421 — the shipped definition gates onboarding forms on active employment. Field, type and relation counts and the conditional premium structure come from the shipped benefits enrollment definition in the SymageDocs form library. No real employee or benefits data was used at any stage.

Frequently asked questions

What data format do synthetic benefits enrollment forms include?

Each generated identity produces a rendered PDF plus a structured JSON annotation file with bounding boxes, field types, and ground-truth values for all 18 fields — 6 text and 12 checkbox targets — with one label-to-value relation per field. Checkbox group membership is recorded so exactly-one-of-N correctness can be scored. COCO, YOLO, FUNSD, and BIO/NER exports come from the same job.

How often do employees waive medical coverage?

18.6% waive. On those forms the coverage tier is unmarked and the medical premium is blank while all the labels remain printed, which is the branch that most reliably produces hallucinated elections.

Are the premiums consistent with the elections?

Yes. Medical premiums correspond to the elected plan and tier and cluster at $82, $176, $228 and $315 per pay period; dental runs $12 to $18 and vision $6 to $9, present only when that coverage is elected. That consistency is what makes the corpus usable for validation rules and not only for extraction.

What is the Section 125 box for?

It authorises the employer to deduct premiums pre-tax under a cafeteria plan. 73.6% of employees authorise it. It matters downstream because pre-tax premiums reduce the federal wages reported in Box 1 of the employee's W-2 — a cross-document relationship the catalog can actually exercise.

Can I use this data commercially?

Yes. Every employee, election and premium is synthetic, contains no real personal or benefits data, and is licensed for commercial use including model training, benchmarking, and redistribution inside your own products.