46% fail within 18 months
Half of hires fail, and the clues were there
In 82% of those cases the interviewers saw the warning signs and did not act on them.
Source: Leadership IQ, Why new hires fail (n = 5,247 hiring managers) ↗Intelligence-led recruiting / How We Scope and Screen
How We Recruit: Evidence Before Introductions
What companies at your stage posted defines the seat. A person decides who earns the introduction.
A strong search begins before sourcing. First we define what success looks like and look at what comparable companies posted for the same seat. Next we write the questions the seat turns on, which test ownership, operating context and results, not keyword overlap. Every candidate answers them in writing; we write a read on each one who passes, with what we would push on next, and a plain verdict, including when the answer is not a hire. You meet people with their answers and our read in front of you, not a pile of profiles. For GTM engineering seats a working GTM engineer runs the screen; for every other seat the screen is ours. You see the questions, and can change them, before the search begins.
Book a call A 30-minute search strategy session. No obligation to engage.
Why This Search Is Hard
Why Most Screens Miss
Not for lack of candidates. Because the hard questions are available and are not asked before the offer.
Madrona's rule for search partners
Talk to every candidate before the founder does
Madrona tells founders to personally talk to every candidate before they do and to run from any search model that will not let them. The written read is how we make that possible without spending the founder's hours.
Source: Madrona, Hooked on search partners ↗300+ applications per hire
Volume is not a shortlist
A technical seat draws more than 300 applications and 17.6 interviews per hire. A list is the cheap part; what reaches your inbox should be the few who passed, not the many who applied.
Source: Ashby, 2023 recruiter productivity trends report ↗For the Hiring Manager
Reading a Screen Report
What each part of the written read is for, and how hiring managers use it.
- 01
The answers, verbatim
What the candidate said to the questions the seat turns on, in their words. You read the conversation before you take the call, so your first interview is a second conversation.
- 02
What we would push on
The one or two things the screen could not settle, so your interview spends its time there rather than re-asking what has already been answered.
- 03
The verdict, and for which kind of company
Put forward, or not, and for what: a founder-led company selling to mid-market operations teams is a different verdict from an enterprise-first motion. The read says which.
Why Us for This Seat
How We Run This Search Differently
Fifteen-plus years placing technical talent, and what several hundred companies at your stage posted, decide how we define, assess and close a search.
The scope comes from what your peers posted
We track the public postings of several hundred companies at your stage and publish what we find: how many posted a sales leader, an account executive or a first applied AI engineer, and at what size. The seat is scoped against that, not against a template or the last job description someone had to hand.
Quality, not quantity
Each role family has a screen written for it: ownership of the work, the scope actually held, the number moved, stage fit and range, in questions that separate people who owned the work from people who sat near it. Most candidates do not pass it, and that is the point: only the best-matched reach your inbox, never a stack of resumes to sort. You see the screen, and can change it, before anyone is contacted.
The read is written, and it comes with the person
We record what the candidate said, what we would push on in a second conversation, and a plain verdict: put forward, or not. You receive it with the profile, so the first interview is a second conversation, not a screen.
No stake in who wins
We are paid on the hire, so we have a stake in the search finishing, and we say so. We have no stake in which candidate wins: nobody at the firm earns more if a particular person is hired, and the read says when neither of the two is right.
What a read looks like (illustrative)
An example in the format we use, with no real candidate behind it: "Closed $1.1M in new ARR in year one on a $90k base, median cycle 47 days, all outbound, founder on two of eleven deals. Playbook: wrote the discovery script and the pricing objection tree the second AE now uses. Would push on: enterprise cycles over 120 days, none closed yet. Verdict: put forward for a founder-led company selling to mid-market ops teams; not for an enterprise-first motion."
For GTM engineering, a working GTM engineer runs the screen
GTM engineering is the one seat where we put a person currently doing the job between the market and your calendar, because the role is new enough that the questions change every quarter. For every other seat the questions are written for the role and the screen is ours.
A person decides
Software builds the list and drafts the notes. Whether a candidate is put forward is decided by a person, every time, and that is the sense in which there is a human in the loop.
How a Search Runs
From the First Call to the Signed Offer
The same shape for every seat, from a founding engineer to a VP. What changes is what peers posted for the seat, and the questions the seat gets.
-
01 · Before anyone is sourced
The role, in writing
A thirty-minute call settles the constraint the seat relieves, the number it has to move, the level, the scope and the compensation band. It is written down and you sign off on it. If the brief is asking for two jobs in one seat, or the honest answer is tooling, we say so here, before a fee exists.
-
02 · Within a day of go
Two people, with their answers
Every candidate is screened against your exact requirements before reaching you: ownership of the work, the scope they actually held, stage fit, range, and the questions written for this role. We screen for quality, not quantity, so only the best-matched reach your inbox, and we write down what they said and what we would push on next. You get two people who passed, their answers and our read. On an exclusive search the stake is in writing on the terms page, in the words our first email uses: Two who passed, with their answers and our written read, by this time tomorrow. If not, no fee on whoever you hire from us on this search.
-
03 · Weeks one to three
Your loop starts at the second conversation
You interview people who have already answered the hard questions in writing, so the loop is one round shorter and the first interview starts where most screens end. One firm, one submission channel: your candidates are not being called by four agencies about the same seat.
-
04 · The close
Compensation, equity, decision rights
We calibrate the offer against the market, surface what the candidate is weighing (equity, reporting line, the founder relationship) and manage it through to a signed acceptance.
-
05 · After the hire
The read stays with you
The seat's questions and the written assessments become the rubric your team uses for the next hire in that function. Replacement terms are in the written agreement for the search.
What you hold at the end
- 01A one-page scope: the constraint, the number, the level and the screening questions, signed before outreach
- 02Two profiles that passed the seat's screen, with their written answers and our read, within a day of go on an exclusive search
- 03One submission channel and a weekly note on the market: who is interviewing, who declined and why
- 04A calibrated offer and a signed acceptance, with the compensation reasoning written down
- 05The rubric for the next hire in that function
What the Market Figures Say
The Numbers Behind the Search
Third-party figures, each linked to the page it comes from. We publish no placement statistics of our own; these are the market facts the search is built around.
-
Founders most often fail at early hiring by hiring for roles they do not fully understand and failing to clarify the role first.
Source: a16z, How to hire a strong founding team ↗ -
46% of new hires fail within 18 months; in 82% of those cases the interviewers saw the clues and did not act on them.
Source: Leadership IQ, Why new hires fail (n = 5,247 hiring managers) ↗ -
A technical role takes a median 76 days to fill, with 17.6 interviews and 23.3 interview hours per hire.
Source: Ashby, 2023 recruiter productivity trends report ↗ -
Marketplace and AI sourcing tools leave screening, engaging and closing to the hiring team; their own pricing pages say so.
Source: Wellfound pricing, as reviewed by Hire in South ↗
Avoidable Hiring Errors
Where Searches Go Wrong
- Letting a keyword screen reject the candidate who owned the work under an unusual title
- Interviewing from a resume when the hard questions have not been asked yet
- Scoping a first-of-its-kind seat from a template instead of from what peers posted for it
- Starting the loop before the level and the scope are settled in writing
Frequently Asked
5 questionsWho screens the candidates?
The firm does: fifteen-plus years placing technical talent, against questions written for the seat, several dimensions of fit, and what companies at your stage posted for it. For GTM engineering seats a working GTM engineer runs the screen. The scope document names who screens your search before outreach starts.
Where does the peer data come from?
From the public job postings of the companies we track, Seed to Series B across the US and Canada, which we publish as research on this site: which seats were posted, at what level, and which seats were posted beside them. When your company's investor portfolio or category has a published study, the scope cites it.
Do you use AI in screening?
For the mechanical parts: building the long list, scheduling, drafting the notes we review. Not for the decision. Whether a candidate is put forward is a person's call.
Can we see the questions before the search?
Yes. The screen for the seat, the questions and what we test for, is in the one-page scope you sign off on before anyone is contacted, and you can change any of it.
What if neither of the two is right?
Then the scope was wrong or the market is, and the read will usually say which. We revise the scope with you and run it again. On an exclusive search, once the seat's two questions are agreed, the stake is in writing on our terms page in the words our first email uses: Two who passed, with their answers and our written read, by this time tomorrow. If not, no fee on whoever you hire from us on this search. Nothing is owed unless you hire.
In their words
“He valued relationships over quick wins. Refreshingly human.”
41 41 people have recommended the person who runs your search. Read them on LinkedIn
Next step
See the Screen Before You See a Candidate
The scope you sign off on names the dimensions and the lead questions for your seat before anyone is contacted, and you can change them. Thirty minutes is enough to write it.