Alternative data on Bloomberg: what buyers should know

Bloomberg lists alternative data among the data it provides to firms, and it offers that data in two forms: a function on the Terminal that people read, and data feeds that a data team loads. This guide sets out what Bloomberg's own pages say about each, what to test before you rely on either, and how to join the feeds to company-level data that you load into your own database. Every statement about Bloomberg was checked against the pages linked here on 4 October 2026.

Fokals is delivered direct, by REST API and as bulk files, which you load into your own database with your warehouse's loader, next to whatever you take from Bloomberg's feeds. The section on joining shows how.

What Bloomberg says it offers

Bloomberg's Enterprise Data page names alternative data next to reference, pricing and regulatory data. A separate alternative data page names the datasets. Two of the six carry the Bloomberg Second Measure name, and the page calls that service's feeds proprietary; the others are named for their providers.

The same page presents ALTD, a function of the Terminal, as a way to see how public and private companies are performing, shown beside market data, and suggests setting KPI estimates built from the data against consensus during the quarter. It states 4000+ supported securities and 7000+ KPI estimates in ALTD, 10+ years of history for the Second Measure data products, and a one-day lag on the Second Measure transaction-level product. Those four figures are Bloomberg's, and they are the ones to test against the companies you track.

The page lists four uses: assessing performance within the quarter before earnings are announced, validating a thesis across several datasets, tracking product launches and analysing custom groups of consumers. The second is where data from other sources enters, because validating a thesis across datasets means joining them, and a join needs a shared identifier and a date on every value.

Where you meet it: the Terminal, feeds and Data License

Bloomberg's alternative data page says the data reaches users every day through ALTD on the Terminal and, as aggregated and transaction-level feeds, through DATA <GO>. The Enterprise Data Catalog page calls data.bloomberg.com Bloomberg's self-service data website, a place to look through, trial and obtain datasets from Bloomberg and from other providers, and says its REST API interface returns standardised data that machines can read. The Data License page names DATA <GO>, a REST API, SFTP, Web Services and the cloud providers as the ways in to its content.

The route decides what you can do with the data. A function you read on a screen supports discretionary work. A systematic process needs a table with a date on every value, so the first question for any dataset you meet on the Terminal is whether the same dataset is offered as a feed, and with how much history.

The pages linked here do not give prices, entitlements or the identifiers each feed carries. Ask Bloomberg which datasets a Terminal subscription includes and which are licensed separately, what each feed is keyed on, and whether a value is ever revised after it is delivered.

Joining it to company data you load yourself

The join is yours to build, and the identifiers decide it. A listed company in the Fokals index carries its ticker and MIC, its ISIN and LEI, and its share-class FIGI; a brand or subsidiary carries the identifiers of its listed parent. Every ISIN and LEI delivered has passed its check digit.

FIGI is an open standard of the Object Management Group, and OpenFIGI names Bloomberg as its Registration Authority. A feed keyed on the share-class FIGI joins to Fokals tables directly. A feed keyed on a listing-level FIGI joins through listed_securities, which holds both values.

Suppose you hold a table of KPI estimates keyed on ISIN and date, loaded from a feed. The table and its column names are yours, because Bloomberg's pages publish no schema. The query adds to each estimate the postings a company opened and closed in the 28 closed days before the estimate date, from Hiring Activity in the hiring dataset, where the table is company_hiring_daily.

-- kpi_estimates is your own table: one row per isin and as_of_date.
-- Write the date arithmetic in your warehouse's syntax.
with hiring as (
  select
    isin,
    day,
    sum(new_postings)    as new_postings,
    sum(closed_postings) as closed_postings
  from company_hiring_daily
  where isin is not null
  group by isin, day
)
select
  e.isin,
  e.as_of_date,
  e.estimate,
  sum(h.new_postings)    as new_28d,
  sum(h.closed_postings) as closed_28d
from kpi_estimates e
left join hiring h
  on h.isin = e.isin
 and h.day >= e.as_of_date - 28
 and h.day <  e.as_of_date
group by e.isin, e.as_of_date, e.estimate;

Three details carry the weight.

• The condition day < as_of_date keeps only days that had closed by the estimate date, which matters because each daily row is written once, after its day closes, and is not revised. For a stricter test, also lag by the time your deliveries show between a day closing and its row arriving.
• Hiring is summed by ISIN first because a brand or subsidiary carries its parent's identifiers, so one ISIN can sit on several company IDs.
• A first observation of a company's postings sets a baseline that the new-postings count leaves out, so a company's first day carries no new postings however many roles are open. Do not read the first day of a series as a lull in hiring.

Read the result as two dated observations side by side: an estimate and the hiring evidence that stood on its date. The data dictionary names every table and column used here.

What to test before you rely on any alternative dataset

Each figure Bloomberg states points to a test you can run on the companies you track.

The third test decides whether a back-test can be trusted. A series that is revised after the fact lets a point-in-time study see information nobody had on the day. Fokals writes daily and weekly tables once, after the period closes, and does not revise them, and each record carries the time it was observed. The same question applies to every vendor, Fokals among them.

In order, the first five steps for a buyer are these.

1. Decide the route. Terminal-only data serves people; a feed serves a process.
2. Ask for a sample and a data dictionary for the datasets you want, before any contract.
3. Run the coverage test: your own holdings against the securities the dataset supports.
4. Run the date test: the date each value describes, the date it became available and any later revision.
5. Run the join on a shared identifier, count the rows before and after, and read the rows that did not match.

Where the two sit side by side

The datasets Bloomberg names measure consumer spending, foot traffic, web visits, app engagement and prescription data. Fokals delivers firmographic, technographic, hiring and intent data on public and private companies worldwide: the technologies a company's website runs, the roles it advertises, what it announces and files, and weekly Intent Scores in which every score carries the dated signals behind it. Web Traffic gives the monthly traffic tier of each website. Each Fokals dataset carries the identifiers a security master already holds, so a screen or a model can hold it in one table with the Bloomberg feeds you load.

The guides to the Open:FactSet Marketplace and the S&P Global Marketplace cover two other places where firms find alternative data, and the page for investors sets out how Fokals maps its data to securities.

Frequently asked questions

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.