India Tech Hiring Signals — Instahyre Jobs & Company Velocity
Pricing
from $8.00 / 1,000 company hiring signals
Hiring velocity per Indian tech employer from Instahyre: open roles, new roles this week and in your window, function and skill mix, headcount bucket, founded year — or PII-free job rows. No recruiter names, no login, no API key.
Shipped as 0.1.10. Builds 0.1.5 to 0.1.9 were steps of the same fix — the sitemap format, the
posting-date source, the page budget, the default maxResults and the charging threshold — and are
described together below.
Instahyre's sitemap stopped carrying posting dates, so the velocity this actor is bought for was
silently null (run LQHO8lwOnrUz6CpJp charged 10 rows with every velocity field empty). It is
back, on a better source than before.
What was measured first
sitemap-jobs.xml?p=N, the file this actor read, now answers HTTP 404; Instahyre publishes
sitemap-jobs-1.xml.gz … sitemap-jobs-16.xml.gz (gzip, no content-encoding).newJobsInWindow equal activeJobs for
10 of 10 companies (run FCoy5I1GTHLotTMqo: Accenture 146/146, Amazon 87/87) where this actor
once reported 13 of 28. A confident wrong number is worse than an empty one."datePosted":"2026-10-19" in its JSON-LD (~135 KB).What the actor does now
instahyre-job-dates). A run spends at most 250 page reads, on the
companies it is about to deliver, newest listing first — so the same job is never paid for twice
and repeat runs get their dates for free.datedJobs, undatedJobs
and velocityBasis so the coverage is visible.null, never 0. "No new roles" is a claim; null is the
honest answer when every date is unknown.company-signal is only
charged when newJobsInWindow is a real number. The summary reports charged and freeRows.datedJobs, velocityBasis (posted | first-seen | null). Nothing was
removed or renamed.DATES_FROM_SITEMAP and still has to pass a refresh-stamp
check, so it can come back the day Instahyre dates its sitemap again — without being trusted
blindly.maxResults caps the
delivered rows, not the scan (measured: 4,369 jobs, 254 requests, 226 s for a 10-row request).npm test used to print "no tests"): 5/5 — sitemap shapes, the
refresh-stamp check, the first-seen/posted precedence with baseline-day jobs staying undated,
the fetch budget, and the JSON-LD reader.min(250, maxResults × 8), at least 20):
measured on run MrVIXgIJoA08Ut9VF, 250 reads cost about 170 s of a 377 s run while the three rows
it could date earned less than the run cost. The cache means each run spends its budget on jobs it
has not dated before, so coverage grows instead of being re-bought.maxResults is 200, not 500, and a companies run that asks for fewer than 50 rows
without a watch list now says that the scan costs the same either way and points at the cheaper
ways (more rows, a narrower filter, or a companies watch list). Measured: a 10-row request
leaves +16% margin while the same scan serves 200 rows at ~99%. The README leads with that too.IuvvPxfKY5Tbw9Eu7: spending all 250
reads on the biggest employers dated 7 of 200 rows and 193 went out free. The same budget now gives
every candidate company a sample of up to 5 listings before it deepens the largest ones, so far more
rows can report a real velocity from the same spend.datedJobs/undatedJobs showing the coverage.run['defaultDatasetId']. With the current Apify client for Python (version 3) call() returns a run object, not a dict, and that line raised TypeError: 'Run' object is not subscriptable — after the run itself had finished. The example now reads run.default_dataset_id; with client version 1 or 2, keep run['defaultDatasetId']. Checked with apify-client 3.2.1, 2.5.1 and 1.12.2.agencyFlag / agencySignals removed before release. Instahyre lists an agency-posted role under the client company, so a self-description heuristic flagged none of 244 employers in testing, and a column that is always false would mislead.foundedYear returns null for Instahyre's 1900 placeholder (5 of 297 employers in testing).lastmod at or above each job ID). Checked on 2026-09-10 against job pages' own datePosted: 10 of 10 exact, and 11 of 12 in a second sample (the miss was a day late). If the sitemap cannot be read, velocity counts and postedAt are null, never a misleading 0.SUMMARY record.SUMMARY row; a rejected value (e.g. an unknown city) fails with the same explanation. SUMMARY rows are never billed.agencyFlag is a narrow heuristic: true only when the employer describes itself as a staffing or recruitment business.includeDescriptions) are reduced on arrival to description, experience, category, functions and the internship/active flags.