How MitchelLake Intelligence is made

What a signal is

A signal is a dated, sourced record that something changed at a company: funding rounds, cybersecurity incidents, geographic expansion, layoffs, leadership appointments and departures, mergers and acquisitions, partnerships, product launches, restructuring or strategic hiring. Each signal names the company, the event, the date, the source and the geography, and carries an interpretation we call the leadership read: what the event is likely to mean for the company’s leadership needs.

Signals are facts about companies. The leadership read is our judgement. The two are kept separate on every page so that a reader can take the fact without the judgement, or disagree with the judgement while keeping the fact.

Where signals come from

Signals are detected from public sources: company announcements, press coverage, regulatory filings and podcast transcripts, gathered through a monitored set of feeds currently spanning 550 distinct publications and primary sources. Every signal links to the source it was drawn from. If a source cannot be linked, the signal is not published.

We do not use, and will never use, private or relationship-derived information as a source for published intelligence. See what we never publish below.

How signals are classified

Each source document is read and classified by the Autonodal signal engine, which assigns an event type, extracts the entities involved, and attaches a confidence score reflecting how directly the source supports the classification. Signals below our confidence threshold are held back rather than published.

Geography is attached at the level of the story, not only the company’s headquarters. An American company opening a Singapore office is a Singapore signal as well as an American one; a European fund investing in an Australian business is a signal for both markets. This is deliberate: expansion and capital movements are the events our readers most need to see from the destination market’s side.

How the leadership read is formed

The leadership read interprets the signal type and the company’s stage to say what leadership change is likely to follow: which functions, which level, on what horizon. It draws on the patterns in our own engagement history and on the corpus itself, and it is the reason we publish intelligence at all. Where a signal has no read, none is invented; the page says less rather than guessing.

What is updated, and when

The corpus updates daily. Signal pages carry the date the signal was first published and the date it was last revised. Aggregate pages (hubs, the daily read, the Talent Market Index) state the data period they cover. Dated permanent URLs exist for Index observations so that a figure cited today can be found unchanged later.

The Talent Market Index has its own methodology, published at the Index methodology and in machine-readable form at /tmi-methodology.

What we never publish

MitchelLake also holds a private relationship graph built over more than two decades of executive search. None of it feeds this site. No relationship, interaction, proximity or network-derived data appears on any public page, in any structured data, or in any feed. The only exception is a small set of whole-firm aggregate figures used as credentials, which are static, reviewed and listed, and which reveal nothing about any individual, company or relationship.

We never publish compensation data of any kind.

Client names appear only where the engagement has been approved for publication. A client named on this site is a client who agreed to be named.

Using this intelligence

Signals and reads may be cited with attribution and a link to the source page. Our open feed is at /editorial.xml. Machine-readable guidance for automated readers is at /llms.txt. We ask for attribution because it is how our readers can check our work, which is the point of publishing it.

Intelligence powered by Autonodal.