80% jobs in Switzerland: what adverts say about working days
Among 22 selected records with an 80%-compatible workload signal, 19 had substantial stored text. Seven mentioned flexible hours; three gave day counts for office attendance or travel. None of the 19 explicitly established a recurring four-day or five-shorter-day pattern. This measures recorded wording, not which arrangements employers would agree to.
- Substantial stored text19
- Fragmentary stored text1
- Missing stored text2
A workload percentage leaves the calendar open
An 80% workload gives an applicant one part of the answer. It does not establish which days are worked, how long each day lasts or whether a particular weekday can remain free. An 80–100% range also leaves the agreed percentage to discuss.
Our earlier workload investigation examined advertised percentages. This follow-up asks what explicit scheduling information appears in selected records’ stored text. We reviewed 22 records from 22 employer IDs; 19 had substantial descriptions, two were empty and one was fragmentary. None of the 19 reviewable texts explicitly established a recurring four-day or five-shorter-day pattern. That is an absence of a statement in those texts, with three cases unknown. It does not establish whether those arrangements are available.
Pro Familia’s 2025 part-time factsheet, page 6, provides context for distinguishing workload from its distribution over days. It supplies none of our advert counts.
Count the kind of information, not just the number
| Statement detected in stored text | Records among 19 substantial texts | What it establishes |
|---|---|---|
| Flexible working hours | 7 | Explicit time flexibility, sometimes a general employer benefit |
| Numeric days per week | 3 | Two office-attendance requirements and one travel allocation |
| Monday–Sunday availability, shifts and weekends | 1 | These three overlapping signals occur in the same record |
| Jobsharing | 1 | General employer wording, without a role-specific day pattern |
| Explicit four-day or five-shorter-day pattern | 0 | No qualifying statement detected |
Each selected record represents a different employer ID, so the record and ID counts coincide here. These labels overlap: they must not be added into a count of distinct vacancies. General policy also cannot establish an individual’s actual schedule. Broad flexible-working-model wording and homeoffice alone were insufficient to count flexible hours.
Office days and travel days answer different questions
The stored Swiss Re wording-specialist record required at least three office days weekly. The EPFL scientific-programme role likewise specified at least three days on site. Both describe attendance location; neither specifies the total number of working days.
The Migros construction-project role described travel assignments averaging two days weekly. Those are travel days within the role, without a complete weekly distribution. All three numeric-day findings therefore concern allocation within work.
The HOCH restaurant-service record required readiness for shifts across Monday–Sunday. This includes weekend availability but does not mean working every day. It illustrates why a compatible percentage can coexist with coverage requirements that need a separate conversation.
These links identify the source destinations recorded by Panta. We reviewed stored canonical text, without opening the original employer pages or confirming current application acceptance.
How the bounded selection was made
Capture ran from 2026-11-07T18:09:25.354Z to 2026-11-07T18:10:33.701Z. Ten public API pages returned 500 records, ordered by recent observation, with employment metadata containing 80 and stored physical_ch classification. Existing rules retained 481 compatible candidates without detected training or pool labels in the screened fields, across 22 employer IDs.
A declared hash order selected one record per ID before descriptions were requested. All 22 IDs were retained; missing or fragmentary text was not replaced. The frame reached its 500-record cap while more pages remained. It was not exhausted, and pagination was not an atomic snapshot.
This filter misses title-only and description-only 80% signals. Recency ordering, source coverage, metadata quality and location classification constrain interpretation. Fresh selected-source backing does not establish complete canonical text, independently verified Swiss geography or present acceptance of applications. Employer IDs are grouping keys, not proof of independent legal employers: two selected Migros-branded records have distinct IDs but share a brand. These results support neither national prevalence estimates nor a trend against the earlier investigation.
Two separate initial AI-assisted codings read the frozen text before comparison. One disagreement about the scope of a flexible-hours statement was adjudicated as role-specific. This was not independent human double coding.
Turn the missing details into shortlist questions
For each lead, record the percentage, total working days, required attendance or travel days, and any shifts separately. Ask whether 80% is agreed for this role, which days are fixed, and when a rota becomes available. Keep unanswered details unknown.
Use the shortlist review worksheet and its blank CSV to retain evidence. The dated audit pack is available as JSON and CSV; live adverts can change after capture.