One method, from a million profiles to one hire.
Everything we do exists to put the right person in front of you, with the reasons written down. This page explains how, in as much detail as we can give without handing the method to our competitors.
1.We read the role, not the job ad.
Thirty minutes with the person the engineer will report to. What they will ship in their first year. What your best engineers have in common. What would make you say no. That becomes a written scorecard, with weights, that you approve before we search. Everything after is measured against it.
- 1
We read the role, not the job ad.
Thirty minutes with the person the engineer will report to. What they will ship in their first year. What your best engineers have in common. What would make you say no. That becomes a written scorecard, with weights, that you approve before we search. Everything after is measured against it.
- 2
We start from the whole market.
Our sourcing reads licensed professional databases, public code, public job-seeker posts and the profiles companies publish, across North America. Hundreds of profiles per role, not dozens, including the people who are not looking. The rules that disqualify someone are applied first, to everyone, the same way.
- 3
We score every profile, with the evidence.
Each remaining profile is scored against your scorecard on the attributes that predict performance. Every score comes with the evidence it rests on and a confidence level. Where the profile is silent, we say “unknown” rather than guess. The top of the ranking is then compared head to head, so the order holds.
- 4
We screen before you spend an hour.
A structured call, twenty minutes, built from your scorecard: the same questions in the same order, scored against written anchors, with a transcript. Every call is rated twice, by the interviewer during the call and independently from the transcript afterward; where the two disagree, a person decides. Structured interviews predict performance about twice as well as the conversational kind, which is why we do not do the conversational kind.
- 5
We write it down, including what we did not like.
One page per candidate: what they have built, where they are strong, what to probe in your interview, and the one thing we are unsure about. You get the shortlist within ten business days of the signed agreement.
- 6
You decide.
You interview who you choose. We do not push, and we do not bury the second-best candidate to protect the first. If the hire leaves within ninety days, we run the search again at no charge.
Powered by research,run by software,judged by people.
Eighty-five years of selection research say the same thing in different words: some attributes predict how someone will perform, most do not, and people choosing by instinct reliably weight the wrong ones. Our scoring combines the attributes with the strongest evidence, reads them from what a person has actually done rather than from how they describe themselves, and weights them the way the research says to, not the way a resume makes them look.
The system is not a filter on keywords. It reads the whole profile, the code where there is code, the writing where there is writing, and it quotes what it finds. It is also not the decision. A ranking is where our work starts, not where it ends.
Why most recruiting does not work this way. The findings are not new or disputed. Structured interviews predict performance about twice as well as conversational ones; a fixed scoring formula beats expert judgment at ranking candidates; confidence and polish move interviewers far more than they move results. Surveys of the profession have found for twenty years that a large share of practitioners simply do not believe it. The first impression is the judgment people enjoy making most, and it is the one the evidence trusts least. We built the firm around the evidence.
What we will not tell you on this page is which attributes, at which weights, read from which signals. That is the method.