A top-10 US bank and a global credit-ratings group
Joined Import.io through its 2019 acquisition of Connotate.
Alternative data you can backtest. Captured as it happened, never restated.
Import.io collects point-in-time data from company sites, job pages, filings, registries, sanctions lists and pricing sources. Investment, lending and risk teams receive timestamped, entity-mapped records through an API or managed feed.
| Signal | Entity | Ticker | Value | Δ WoW | As of (UTC) |
|---|---|---|---|---|---|
| Open job postings | Kestrel Logistics | KSTL | 1,284 | +8.2% | 09:30:04 |
| Active SKUs online | Halden Home | HLDN | 41,902 | +1.1% | 09:30:11 |
| Avg. discount depth | Northmoor Retail | NMRT | 18.4% | +3.1 pt | 09:30:19 |
| Store locator count | Brightwater Foods | BWTR | 612 | +4 | 09:30:22 |
| Hotel ADR, key metros | Harbor Hospitality | HRBR | $289 | −2.4% | 09:30:27 |
| 10-Q filed | Kestrel Logistics | KSTL | 38 tables parsed | new | 09:31:02 |
Production examples from organisations using Import.io today and in recent years.
Joined Import.io through its 2019 acquisition of Connotate.
Has run managed web data feeds for its research teams.
Runs the web data behind its risk models on Import.io.
Web data is only an edge if it survives compliance and a backtest. That means every value timestamped at capture, never overwritten, mapped to the right entity and security, with a collection method you can document. Import.io delivers web signals the way a quant, a credit committee and a compliance officer all need them.
Every fund can buy web data now. What separates a signal from a liability is whether it was collected the way it claims: values stamped when they were seen, a stable panel that doesn’t quietly lose the companies that failed, identifiers that map cleanly to securities, and a methodology your compliance team can sign.
We capture point-in-time by default, keep every version, document every source, and map entities to tickers and legal identifiers. Backfill comes from our own captures, not from today’s page pretending to be last year’s.
Six programs we run most often in this industry.
Related guidance: alternative data, point-in-time data, external data, data quality.
Price indices, discount depth, assortment and stock-outs by company, as leading indicators of revenue and margin.
nowcastingJob postings by company, role, location and salary band, deduplicated across boards and careers pages.
hiringTables and text from 10-Ks, 10-Qs, prospectuses and registry filings, with page and table citations.
filingsCompany registries, officers and sanctions lists, monitored for changes that matter to onboarding and exposure.
riskListings, price cuts and building permits as supply and demand signals by metro.
propertyCompany events, store openings and closures, recalls and management changes as structured events.
events| field | type | example |
|---|---|---|
| entity_id | str | kestrel-logistics |
| ticker / figi | str | KSTL / BBG000… |
| signal | str | open_job_postings |
| value | num | 1284 |
| unit | str | postings |
| period | date | 2026-09-25 |
| as_of | ts | 2026-09-25T13:30:04Z |
| source_url | url | careers page |
| method_version | str | v3.2 |
| content_hash | str | sha256:9f2c… |
What breaks when this is done with scripts, and how Import.io handles it.
Values are versioned at capture and never overwritten, so backtests see only what was knowable on each date.
Coverage is fixed and changes are logged, so failed companies don’t silently disappear and bias the history.
Brands, subsidiaries and domains are resolved to legal entities and securities, with the mapping itself versioned.
Source lists, collection methods, personal-data handling and data processing terms are documented for your vendor review.
Same capture engine underneath each one.
Quant and data teams who want point-in-time signals delivered to Snowflake, S3 or BigQuery before the open, under an SLA.
learn more →Filings, job postings, company registries and permits, maintained and ready to trial.
learn more →Research analysts who want to prototype a signal before it becomes a feed.
learn more →Programs collect publicly available information only. Collection methods, sources, personal-data handling and data processing terms are documented for vendor due diligence, and every value carries its source URL and capture time.
Straight answers.
Yes. Every value is stamped when it was captured and kept as a version, so history reflects what was knowable on each date, not today’s restated page.
History starts from our first capture of each source. We don’t reconstruct history from current pages; where a customer needs earlier history, we say what exists and where it came from.
Yes. Companies, brands and domains are resolved to legal entities and security identifiers, and the mapping is versioned alongside the data.
Yes. We document sources, collection methods, rate-aware collection, personal-data handling and data processing terms, and answer vendor questionnaires.
To Snowflake, BigQuery, Databricks, S3 or SFTP on a fixed schedule, typically before the market open, with a manifest on every delivery.