The US economy, seen through the eyes of the employer.

The problem

Workforce change at other employers is invisible until it is news. Most of it never becomes news.

The story of one employer arrives in pieces, months apart, each read by a different person. A labour reporter sees the union petitions. A filings desk sees one line in an 8-K. A state web page shows a notice for 224 workers to whoever happens to open it. Nobody sees the thread.

By the time a company’s situation is understood, it is months old. And for most employers it is never understood at all: of 32 listed employers with a layoff notice since 1 October 2026, 22 had no press report. The talent, the warning and the time go to whoever read the state’s page.

What Vindas does

Every public workforce record, under one name, dated the day it became public.

WARN Act layoff notices

35 states and DC, read the day a notice is posted, with site and effective date

SEC 8-K filings

Item 2.05 (exit or disposal costs — a restructuring) and item 1.03 (bankruptcy, receivership, or an assignment to creditors)

Public announcements

The press, national and local, kept as what it is: the company’s word, not a record

Union petitions

NLRB representation petitions, by store and state

Joined per employer, the records read as a timeline. Beside each change, what followed for employers whose records looked like it — “of employers that looked like this, X% did Y within a year, against Z% of all” — as a count with its n, never a forecast. Every sentence is kept as written on its day, so last quarter’s picture can be replayed as it looked then.

It reaches you as a weekly brief of what changed at the employers you watch, a page per company with its record in full and its peers beside it, watchlists with alerts the day a record lands, and an API with point-in-time reads for your own stack.

Console screens on development data, as of 26 October 2026.

Who it is for

Two readers, in order.

A worked case, still open

Starbucks: one employer, four kinds of record, about 90 days.

The 90 days to 26 October 2026, every record at the date it was first public. Four layoff notices for 428 workers, one restructuring filing, 23 union petitions in 16 states, and the press — read apart by four different desks, read together by nobody.

1. 31 JulFirst union petitionNLRB · 23 petitions in 16 states follow through 16 October
2. 20 AugLayoff notice, 224 workersWARN · Washington
3. 24 SepRestructuring filingSEC 8-K item 2.05 · “costs associated with exit or disposal activities”
4. 24 SepNational press“another 250 stores to close”
5. 7–8 OctTwo more notices, 51 + 18 workersWARN · Washington, Pennsylvania · five local outlets confirm on 8 October
6. 23 OctFourth notice, 135 workersWARN · Seatac and Seattle

Read together: a notice for 224 workers in August, a restructuring filing five weeks later, three more notices after it, while organising spread to 16 states. One employer, one dated thread.

A worked case, closed

Lazydays Holdings, 2025: the record, and what the console would have said on 30 April.

An RV dealer in Florida that no national desk followed. Every item below is in the filing index and the Florida WARN list.

1. 23 JanContinued-listing deficiency noticeSEC 8-K item 3.01
2. 30 AprRestatement: past accounts not to be relied onSEC 8-K item 4.02 · the replay point
3. 16 SepLayoff notice, 76 workersWARN · one Florida site, effective 16 November
4. 10 OctRestructuring filingSEC 8-K item 2.05 · an asset sale put to shareholders two weeks later
5. 7 NovDelisting filingSEC 8-K item 3.01 · off Nasdaq on 28 November
6. 1 DecWound upSEC 8-K item 1.03 · a general assignment for the benefit of creditors and a certificate of dissolution — a wind-up under state law, not a court petition; nothing for shareholders

On 30 April 2025 the console would have said: after a restatement, a delisting notice followed within a year at 37% of listed employers, against 14.3% (1,787 cases); after January’s deficiency notice, a delisting within a year at 28%, against 10.5% (9,106 cases).

The delisting came 212 days after that replay point. These are counts of what followed other employers, re-measured on the record as of November 2026. They were never a forecast for Lazydays, and one case is an anecdote; the measured rates are the evidence.

What a pilot measures

Two numbers from the record so far, and what the first month adds.

A pilot with a design partner measures its own gain on its own accounts: how many of the employers it pursued surfaced in the record first, how many days before the team would otherwise have heard, and how many it never heard about at all. No gain is claimed in advance; it is measured in the first month and written down with its n.

What it does not claim

Data, not advice.

• No layoff prediction, and no “will”, “likely” or “risk” on any line about a named company. What Vindas prints is a count: of employers whose records looked like this, N% did Y within a year, against Z% of all, with the n.
• Not “weeks before it becomes news”. Where a notice and a press report both existed (10 employers), the notice came first in 4 cases and the press in 5. What the data backs is narrower: most never make the press.
• One case is an anecdote. Starbucks and Lazydays are shown because every date is a row in the record; the rates beside them are measured across every listed employer, on years the model never saw.
• Coverage is stated per figure, with its read date. Thirty-five states and DC publish notices; the rest do not, and the page says so rather than filling the gap.