Scouting services and aggregators judge new alternative data for funds, and both Neudata and Eagle Alpha publish what they and their clients ask. This guide sets those criteria under five headings, history, coverage, mapping, point-in-time behaviour and compliance, and pairs each with a check you can run yourself on a company-level dataset, using the Fokals tables as the worked example. It is written for an analyst or data lead who holds a vendor's sample file and has to decide whether to open a trial. Every statement about the two companies was checked against the pages linked here on 4 October 2026.
Fokals is delivered direct, by REST API and as bulk files in JSON, JSON Lines or CSV, which you load into your own systems, and the checks below run on them. The tables and columns in the queries are those of the Fokals data dictionary.
Where the criteria come from
Neudata's May 2026 article for data providers names four factors that set a dataset's value to an external buyer: uniqueness, frequency, coverage and signal quality. It adds that hedge funds and quantitative managers look for a completed questionnaire, documentation of the schema and, where possible, a back-test report prepared by a systems integrator. Neudata's guide to data scouting, from July 2022, adds frequency, history and price, and asks whether the provider has permission to resell the data.
Eagle Alpha's guide to selling data describes what quantitative and discretionary funds prefer. Its August 2025 article on data sourcing names what to weigh: how broad and deep a dataset is, how accurate its history is, what it costs and what legal risk it carries, then ticker coverage, API or FTP delivery, back-testing potential, the vendor's reliability and regulatory exposure. The five headings below follow from those lists. The checks are this guide's own.
History
What scouts ask. Eagle Alpha says quantitative funds need large volumes of historical data to back-test, and its guide advises a trial of at least three months. Neudata's May 2026 article says trials usually run on historical periods that represent relevant market cycles, not on current data.
The check. Find the first day, the last day and the gaps. A table that starts later than the vendor says, or skips days, fails here before any modelling begins. The query runs on Hiring Activity, company_hiring_daily.
select min(day) as first_day,
max(day) as last_day,
count(distinct day) as days_present,
max(day) - min(day) + 1 as days_in_span -- PostgreSQL date arithmetic
from company_hiring_daily;If days_present is lower than days_in_span, find the missing days before any test. The reconstructed flag marks a period that was written late, more than seven days after it closed.
Where Fokals stands. Every observation is dated, and each daily or weekly table is written once, after its period closes, and never revised. A test on Fokals data therefore reads each day as it was known on that day, which is the property a back-test depends on. Hiring Activity gives a daily series for each company, and Market Series gives weekly series on same-store cohorts, as of each Sunday. The methodology describes each dataset.
Coverage
What scouts ask. Neudata's May 2026 article says coverage decides some trials: it reports trials failing at the last stage because the data did not cover the buyer's index model, with no quality issue. Eagle Alpha's guide says quantitative funds want data that spans a wide set of investable names, which it puts at typically more than 75 tickers.
The check. Match the sample to your own list, not to a headline total, and split the result by the groups you trade. A total can hide a weak sector. In the query, my_companies is your list with an isin and a sector column.
select u.sector,
count(*) as names,
count(v.isin) as covered,
round(100.0 * count(v.isin) / count(*), 1) as covered_pct
from my_companies u
left join (
select distinct isin
from company_hiring_daily
where day = date '2026-10-01' and isin is not null
) v on v.isin = u.isin
group by u.sector
order by covered_pct;Where Fokals stands. The index covers equity listings in 79 countries, the brands those companies own and verified private companies, each with one stable company ID. Hiring Activity gives daily open, new and closed postings for each company in it. Match on the companies you track before relying on any total, and split the match by the groups you trade.
Mapping
What scouts ask. Eagle Alpha's May 2025 article says most financial institutions will not consider buying a dataset that is not mapped to identifiers, and that funds prefer mapped data even where it is optional, because it is easier to back-test. It names tickers, ISIN, CUSIP and FIGI, and advises vendors to keep reference data current through mergers, acquisitions and ticker changes.
The check. Measure identifier completeness, then join to your security master. The first number says how much of the file can be mapped, the second how much of your list it reaches.
select count(*) as rows_total,
count(isin) as with_isin,
count(figi) as with_figi,
round(100.0 * count(isin) / count(*), 1) as isin_pct
from company_hiring_daily
where day = date '2026-10-01';A row carries an ISIN when the company or its parent is listed. The rows to read are those that carry an identifier and still fail to join to your master: a share class, a delisted line or a different identifier scheme.
Where Fokals stands. Each listed company is identified by ticker, MIC, ISIN, LEI and share-class FIGI, plus an SEC CIK where one exists, and a brand or subsidiary takes the identifiers of its listed parent. A code is stored as an ISIN or LEI only if its check digit passes. Join on ISIN or FIGI, because a ticker can be reassigned; if your security master is keyed on another identifier, map through a crosswalk. The use case on mapping alternative data to a security master works through the join.
Point-in-time behaviour
What scouts ask. Eagle Alpha's May 2024 article says point-in-time data prevents look-ahead bias, because a back-test then uses only what was known at the time. It tells providers to timestamp records, keep versions, hold a record of every correction and leave historical identifier records unaltered.
The check. A table that is written once cannot change between two deliveries. Export the same period twice, a week apart, and count the differences.
select count(*) as changed_rows
from export_week_1 a
join export_week_2 b using (company_id, day)
where a.open_postings is distinct from b.open_postings
or a.new_postings is distinct from b.new_postings
or a.closed_postings is distinct from b.closed_postings;On a write-once table the count is zero. A vendor that corrects history will show a number here, and a back-test on the later file then uses information that did not exist on the day.
Where Fokals stands. Every record carries the time it was observed. A daily or weekly table is written a single time, once its period has closed, and is never revised. A first observation sets a baseline and is never counted as a change. Each label and score belongs to a named, frozen version, and a change that breaks compatibility arrives as a new version, announced at least 90 days ahead.
Compliance
What scouts ask. Neudata's April 2023 article on compliance lists questions a vendor should expect: where the data comes from, on what legal basis the vendor distributes it, whether it holds personal data and whether the company has faced lawsuits or investigations. It names material non-public information, data privacy and intellectual property as the areas a buyer checks. Eagle Alpha's guide describes the due diligence questionnaire as a structured form covering sourcing, collection methods, privacy and the legal right to sell.
The check. Ask for the questionnaire, the data dictionary and a small sample, then test the claims. Open the URL of a sample of rows in Job Postings and Company News and confirm that each is a public page or filing. Read the data dictionary for any column that holds a name, an email address or a phone number.
Where Fokals stands. Fokals data is collected from first-party company sources and public records, and processed in-house: what companies publish on their own websites and careers pages, what they announce and what they file. It is company-level data throughout, and every input is public when it is observed. The sourcing statement sets this out, and alternative data due diligence questionnaires, question by question goes through the standard questions.
Running the checks in order
Run the checks in order of cost. Read the data dictionary and the sourcing statement first, because they take minutes and rule a dataset out most cheaply. Load the sample next and run the coverage and mapping queries, which need only your list. Read the first and last day for history, and request a second export a week after the first so that the point-in-time test has two files to compare.
Give the compliance documents to your compliance team early. Neudata's May 2026 article lists compliance sign-off among the causes of delay on a provider's path to first revenue, and its April 2023 article says the compliance team should be involved at every stage.
Two criteria no vendor can document for you
Uniqueness and signal quality come last because the vendor cannot settle them. Neudata's article says data that is widely available loses value quickly as more buyers trade on it, and that buyers need evidence of predictive value before committing. Fokals data is a dated, labelled and mapped record of what companies do in public, refreshed daily, and every intent score carries the signals behind it. Whether it adds signal to your strategy is a result you produce on the companies you track, and sample data for that list is sent on request.
Frequently asked questions
What do alternative data scouts check first?
Neudata's 2022 guide says a scout wants to know how reliable the data is, how much value it adds and whether it is compliant, along with its frequency, history and price. The cheapest checks to run first are where the data comes from and whether the provider may resell it, whether it covers the companies you track, and whether every record is dated.
How much history does an alternative dataset need?
There is no fixed minimum, because it depends on the use. Eagle Alpha says quantitative funds need large volumes of history to back-test, while Neudata says trials usually run on historical periods that represent relevant market cycles. Whatever the length, a back-test is only as good as the dating: ask for a timestamp on every record, versions rather than overwrites and tables written once, so that each day reads as it was known.
Why do funds ask for identifiers such as ISIN and FIGI?
A dataset has to join to the fund's security master before anyone can use it. Eagle Alpha writes that most financial institutions will not consider buying an unmapped dataset, and that even where mapping is optional funds prefer it for back-testing. ISIN and share-class FIGI identify a security, whereas a ticker can be reassigned.
What is point-in-time data and why do scouts ask for it?
Point-in-time data records what was known on each date, so a back-test cannot use information that arrived later. Eagle Alpha's article says providers need timestamps, version records and preserved corrections, and that identifier history must stay unaltered. A quick test is to compare two exports of the same period: a write-once table shows no differences.
Do funds ask data vendors for a due diligence questionnaire?
Yes. Neudata's May 2026 article says hedge funds and quantitative managers expect due diligence documents, including a completed questionnaire and schema documentation, and Eagle Alpha describes the questionnaire as a structured form covering sourcing, collection methods, privacy and the right to sell. The guide to due diligence questionnaires goes through one question by question.
The queries and code on this page are examples to adapt. Test them in your own environment before you rely on them.
What this page says about the products it names was checked against their public documentation on 4 October 2026. Product and company names are trademarks of their owners. Fokals is not affiliated with them or endorsed by them.