Most of the serious research about work is published by companies that sell something next to it: a job board, a payroll processor, a staffing firm. Indeed Hiring Lab, ADP and Gusto all do good work, and I use some of it. This is less a criticism than an observation about incentives. That research exists because it makes the parent company interesting, and that decides which questions get asked.
The rest of the field runs on grants, which answers to a different set of priorities but answers to something all the same.
No Talent, No Alpha has neither. It is paid for by the people who read it. That is the reason it can spend a week on a question nobody is asking, and say plainly when the numbers will not support the answer everyone wants.
No Talent, No Alpha is read by people at Korn Ferry, Iron Mountain, Fox, SS&C, NSF, Luxco, IntelliSurvey and Atomic Brands: people who hire, people who run companies, and people whose job it is to know what is happening to work before somebody asks them.
Author|How this is made|Ethics|What this does not cover|Contact
Author
No Talent, No Alpha is written by me, Kane Carpenter.
I've spent more than a decade inside human capital work: hiring, restructuring, pay and retention. I have sat in the meetings where a headcount number gets decided, and I have watched the reasoning get tidied up afterwards into something that sounds strategic. That is why I do not take the public explanation at face value, and it is why the name is an argument rather than a joke. Without the people there is no outperformance. They are not a cost line that happens to breathe.
I have an MBA from the University of Chicago Booth School of Business.
How this is made
Every number here comes from a primary source: state labor agencies, the Bureau of Labor Statistics, the Census Bureau, Indeed Hiring Lab. Nothing is licensed from a data vendor and nothing is modeled or estimated.
The datasets are rebuilt every morning by a script that reads the government files directly. That script is public. You can read it, run it yourself, and check that the numbers on this site are the numbers that were published.
An example of what that catches. In March the Census Bureau quietly changed the question it uses to measure AI adoption, went back to an older wording for one fortnight, and then changed it back. The national figures either side of that fortnight read 18.2%, 18.9% and 19.1%. Perfectly smooth. Nothing in the numbers tells you that the middle one answers a different question. The only way to see it is to read the question text back on every single period, which is what the script does, because a trend built out of two different questions is not a trend.
Ethics
I want my incentives pointed at readers rather than at the companies I write about.
- I take no advertising and no sponsorship.
- Where the data is thin or missing I say so, rather than fill the gap with an estimate.
- When I get something wrong I correct it in public and say what changed.
What you get
The monthly read on what the numbers did is free.
Everything else is $20 a month, or $200 a year. Every issue in full: one claim about work at a time, taken apart, ending somewhere you can use on Monday. Published whenever the data gives us something worth saying, not to a schedule. The Data Room as well: every dataset behind the writing, every layoff notice as the notice is filed, rebuilt every morning, searchable and exportable. Ask a question in plain English and the answer comes back from those datasets with every source named.