What Workday actually asks

Workday feels like it asks endless questions. Counted, it asks remarkably few — the same ones, over and over, per employer.

293 questions, asked 10,708 times

Workday's questions are configured once per employer tenant and then reused on every posting that tenant publishes. The consequence is measurable: 10,708 logged questions across only 293 distinct labels — each one reused about 37 times. Greenhouse, where questions are written per posting, is the opposite: a long tail of labels that rarely repeat.

If you are applying to several roles at one Workday employer, that is the single most useful fact on this page. The second application at that employer will ask you almost exactly what the first one did. Writing your answers down once is not busywork; it is most of the work.

What the 293 are about

Dated snapshot, computed 2026-11-07 by the same 17 ordered match rules published in full on the all-platforms page; the rows, residual included, sum exactly to the 10,708 logged questions.

44.8% of Workday's questions are about who you already know there. That is the finding that most surprised us, and it is not close. Conflict-of-interest and prior-employment questions — are you a former employee, do you have a relative here, do you have a close personal relationship with anyone in the organisation, do you hold an interest in a competing business — dominate the Workday question set. Only 45 distinct labels, but 4,793 encounters. They come from large regulated employers (defence, audit, utilities, government contracting) whose compliance questionnaires are tenant-wide.

Two practical consequences. First: these are the questions to have a considered answer for before you start a Workday application, because several of them are long and some are multi-part. Second: they are facts about your history, not opinions about the job, so your answer to them is identical across every employer who asks — which is exactly why they are worth writing down once.

Compare the bottom of that table: Workday asked about compensation expectations 15 times and "how did you hear about us" 6 time(s). Across all platforms together the same rules matched 3,306 compensation encounters and 3,886 how-did-you-hear encounters — those questions live almost entirely outside Workday. The platforms ask genuinely different things.

Where Workday applications actually stall

Of 1,620 Workday applications attempted, 1020 reached submit and 600 did not — so 37.0% of attempts stopped somewhere in the flow. Each abandoned attempt recorded why. Grouping them (the residual row keeps the table summing to its own denominator):

The stall records name the step they died on, and the shape they draw is consistent: the work-history step ("My Experience") and the account/sign-in wall before the form even opens are the two hard gates of Workday's multi-step wizard, and the questionnaire step is where required questions pile up.

Greenhouse fails for a completely different reason

The same instrument abandoned 1,644 Greenhouse applications, and the reasons barely overlap with Workday's.

Greenhouse has no wizard to get lost in — it is one page, so it does not stall, it just refuses to submit until a required question is answered. And 448 of those 1,644 were not form problems at all: the job had already been removed, or the employer's board handed off to a different system mid-apply. Postings disappear faster than most job seekers assume.

Reproducing this

Run 2026-11-07 against the live database (the same strings, machine-readable, are at /stats.json):

SELECT count(*) FROM direct_applications WHERE ats='workday' AND status IN ('applied','interview','rejected'); → 1020 submitted

SELECT count(*) FROM direct_applications WHERE ats='workday' AND status='apply_failed'; → 600 not submitted; the stall split groups skip_reason with a first-match-wins CASE whose ELSE is the published residual row (full text at /stats.json)

The question categories come from the same logs/answer_gaps.log described on the all-platforms page, filtered to source=workday and bucketed by answer_gap_buckets.py, which exits nonzero if the rows do not sum to the denominator.