Refolk
FrameworkInvesting and deal sourcing

The Hiring-Plan Seat Read: On Plan, Slips, Reprice, or Not Fillable

Score each seat on a founder's hiring plan on four measured dimensions, give it one of four verdicts, and restate the round's month-by-month burn and runway.

23 min readLast reviewed November 17, 2026Read as Markdown

Key takeaways

  • Refolk's index holds 237 Directors, Heads and VPs of machine learning in the US against 4,237 senior ML engineers, about one leader per 18 engineers, so leadership seats are usually what holds an AI hiring plan back.
  • At the published outbound benchmark of 2.5% sourced-to-hired, one hire uses up about 40 sourced candidates, so a raw pool count is the most supply could be, not the supply itself.
  • Time-to-fill benchmarks stop at offer acceptance (about 39 to 44 days in SHRM data, about 48 days for tech roles), so a realistic start month needs a senior premium of roughly 20% plus a written notice-period lag.
  • Engineers reply to outreach at 4.64% on a delivered basis, and the 21.3% blended reply rate overstates engineering conversion more than fourfold, so always use the rate for the role family.
  • A one-quarter slip on a $180,000 seat defers $58,500 of loaded burn at 1.3x, but it also pushes the milestone that seat drives further out against a Carta seed-to-Series A median of 1.9 years.
  • In Refolk's index the UK has 666 senior ML engineers and Meta holds 6 of 25 sampled profiles, so one large employer in a small pool counts as competing demand, not available supply.

This is the test I run on a hiring-plan slide, one seat at a time, before committing to a round: can each planned hire be found, afforded and signed by the month the financial model assumes? It is for early-stage investors, platform and talent partners at funds, and angels doing use-of-funds diligence. By the end, every seat has one of four verdicts, and you have a restated month-by-month burn to set against the runway the deck claims.

What does a hiring-plan seat read test?

A hiring-plan seat read tests whether the labour market can supply the plan. For each planned seat it measures four things: the reachable pool, the competing demand, the gap between budgeted and posted pay, and a realistic date when the hire signs and starts. Standard use-of-funds headcount plan due diligence checks what the seats cost. This read checks whether those seats can be filled at that cost by that date.

Most of the guides that rank for how to diligence a startup hiring plan stick to arithmetic: loaded-cost multipliers, months of runway, hires per milestone. Startupfundraising.com's hiring-plan framework checks whether the first 3-5 hires are named or actively in flight. Guru Startups asks for sensitivity analyses on time-to-fill but gives no way to measure it. Both checks are useful. Both also assume that a senior machine learning engineer can be hired in London at the planned salary by month four, and that assumption is where plans quietly break.

Four terms, defined once:

  • Seat: one row of the hiring plan, made up of a role, a level, a location, a budgeted base salary and a modeled start month.
  • Reachable pool: the number of profiles in a profile index that match the seat's title, seniority and location, after you remove the people who are not real candidates.
  • Competing demand: the open postings for the same profile in the same location, plus any large employer that already holds a big share of the pool.
  • Time to signed start: the days from opening the requisition to the hire's first working day. It is longer than time-to-fill, which stops at offer acceptance.

This read grades the market, not the founder. A founder with unusual pull can beat these numbers, and that deserves its own assessment. The seat read tells you what an ordinary, competent hiring process would achieve.

The stakes are high because headcount is typically 60-75% of burn. If the plan is wrong about its seats, it is wrong about most of the spend and about when the milestones land.

Dimension 1: how big is the reachable pool in the stated location?

The reachable pool is the number of matching profiles in the seat's stated metro, minus the profiles that are not real candidates. The raw national count is only the most it could be. Score the pool against how many sourced people one hire actually uses up, which the published outbound benchmarks put at about 40.

What the counts look like

Machine learning makes a good test case because most AI-era hiring plans include it. These are the counts in Refolk's index of professional profiles:

Band and countryProfilesMain hub share (25-profile sample)Derived
Senior ML Engineer, US4,237Bay Area, 40%6.4x the UK pool
Senior ML Engineer, UK666London, 48%1.0x
Director/Head/VP of ML, US237Bay Area, 64%5.6% of the US IC pool

The hub shares come from 25-profile samples, so treat them as rough guides. The multiples are my own arithmetic.

237
Directors, Heads and VPs of machine learning in the US, in Refolk's index
That is about one leader for every 18 senior ML engineers, and 16 of the 25 sampled are in the Bay Area.

Multiply the national count by the hub share to get the count for the stated location. A London senior MLE seat draws on roughly 320 profiles (666 x 48%), not 666. A Bay Area senior MLE seat draws on roughly 1,700. A US ML leadership seat outside the Bay Area draws on something like 85 people, spread across every other US metro. These are rough figures, but the gaps between them are too large to be sampling noise.

The leadership row is the one that breaks plans. Each team has only one leader, so the supply of leaders grows far more slowly than the number of funded startups that want one. The top employers in the leadership sample include Pinterest, ServiceNow, Upstart and Lightmatter. These are established companies, and they pay to keep their ML leads.

Why a pool turns into so few hires

Outbound sourcing, per 100 people sourced

  1. Sourced
    100%

    every profile contacted

  2. Replied
    16.9%

    blended across role families

  3. Entered the ATS
    8%

    became a tracked applicant

  4. Hired
    2.5%

    about 1 in 40

At the published outbound benchmark, 100 sourced candidates produce about 2.5 hires, so one seat needs roughly 40 sourced people.

At 2.5% sourced-to-hired, one hire uses up about 40 sourced candidates. If you sourced every one of 4,000 profiles, you would get about 100 hires at most, and every other company hiring that profile is sourcing the same people. Once you take out stale and unwilling profiles, the honest yield is dozens of hires, not hundreds.

Discounting the raw count

No public procedure exists for discounting a raw pool count. The closest evidence comes from Pin's study of more than 940,000 contacted candidates. People between roles replied 74% more often than candidates with no signal, and people with recently updated profiles replied 23% more often. That supports weighting a count towards job-change and recency signals. Nobody has published a deduplication rate or a discount for stale profiles, so choose one, write it down, and apply it the same way to every seat.

  • What the pool proves: the people exist where the plan says the seat sits.
  • What it looks like when it lies: a count padded with founders, people in adjacent titles, and profiles nobody has touched in years. The only reliable check is to read the 25-profile sample yourself.

Dimensions 2 and 3: competing demand and the posted-pay gap

Competing demand is who else is bidding for the same pool right now. The posted-pay gap is how far the plan's budgeted base sits below the pay those bidders publish. Together they tell you whether the seat can be won at the planned price.

Competing demand

Count the open postings for the same title, seniority and location, and list the top current employers in the sampled profiles. No public benchmark exists for a healthy ratio of postings to pool, so the cut lines in the rubric below are my working ones.

In small pools, how concentrated the employers are matters more than the raw number of postings. In Refolk's UK senior ML sample, Meta is the top current employer, with 6 of 25 profiles. In a country pool of 666 people, a quarter held by one large employer is demand, not supply, because those people are being paid to stay. The same share would matter less in the US pool of 4,237, since the UK pool is about one-sixth that size.

Checking concentration by hand means paging through profiles one at a time. Refolk does it with a plain-English query: you ask for the profile at the named employers in the city, and the count and the sample come back together.

Posted-pay gap

Collect posted ranges for the same profile from employers in jurisdictions that require them. Take each range's midpoint, find the median of those midpoints, and compare it with the plan's budgeted base. Express the gap as a percentage below (or above) that median.

How much posted ranges tell you depends on where the posting is:

SignalFigureHow to read it
Software development postings disclosing salary (Indeed)63%Enough to sample in most US metros
UK postings with pay information (secondary, citing Indeed Hiring Lab)58%Usable, but check how wide the range is
German postings with pay information (same source)12%Too thin; treat pay as unmeasured
Ontario "transparent" postings quoting exact pay16%Mostly ranges, often wide ones

Colorado was the first US state to require salary ranges in postings, and trackers disagree on how many states require them now. One tracker counts 13 states plus DC, with more on the way. Check a current tracker rather than trusting any count, including this one. California's SB 642 defines the posted pay scale as a good-faith estimate of the range the employer reasonably expects to pay at hire, so California midpoints are the most trustworthy anchor. A secondary source reports about 24% non-compliance among legally covered ads, but that figure has not been verified at the primary source.

  • What the pay gap proves: whether the budget sits inside the range the market pays.
  • What it looks like when it lies: a very wide posted band whose midpoint means nothing, or a midpoint taken from a jurisdiction with no good-faith rule.

Dimension 4: how long until a signed hire actually starts?

Time to signed start is a benchmark time-to-fill, raised for seniority, plus a written lag for notice periods and onboarding. For senior technical seats it often runs past the 60-90 days that model guidance recommends. Compare it with the plan's modeled start month: the difference is the slip.

BenchmarkFigureCaveat
Median time-to-fill, all roles (SHRM)About 39-44 daysDepends on the report edition; check shrm.org
Median time-to-fill, tech roles (KORE1)About 48 daysRecruiting agency, so not neutral
Senior premium, tech roles (KORE1)About 20% longerApplies at every level
Senior roles past 90 days (Pin, citing HR.com)Nearly 40%The long tail, not the median
Offer accepted to first dayNot publishedState your own assumption

The formula is: realistic start = requisition open date + (benchmark time-to-fill x seniority premium) + start lag. For a senior tech seat, 48 days x 1.2 comes to about 58 days to offer acceptance before any lag, and nearly two in five senior roles take longer than 90 days. For leadership seats, plan around the 90-day tail rather than the median. This is the check most time-to-fill assumptions in a financial model skip.

SHRM's data also carries a warning about leadership seats. Executive roles used to take 16 to 17 days longer to fill than nonexecutive ones. That gap has closed, but the cost of executive hires has risen, and SHRM suggests organizations may be investing more to expedite hiring for critical roles. Average cost per hire is about $5,475 for nonexecutive hires and about $35,879 for executive hires. A plan that expects executive hires at nonexecutive speed and cost will miss either the cost or the date.

How quickly people respond to outreach also depends on the role family. Engineers are the slowest to reply, at a median of about 4.1 days, and on a delivered basis they reply at 4.64%, against 9.94% for product managers. A 4-step outreach sequence gets twice the replies of a single email and a 68% higher rate of interested responses. If the plan expects one recruiter to fill three engineering seats in a quarter, test that against these engineering rates, not a blended figure.

The four verdicts and how to assign them

Every seat gets one verdict: on plan, slips, reprice, or not fillable as specified. Give each seat the most severe verdict that any of its dimensions triggers, and record the next most severe as a secondary note.

Pool depth against pay fit

Budget at posted payBudget under posted pay
Thin pool, pay fits
Slips or not fillable; widen the location or move the date
Deep pool, pay fits
On plan if the start month holds, otherwise slips
Thin pool, budget short
Not fillable as specified; change level, location or seat
Deep pool, budget short
Reprice; raise the base, add a fee line, or both
Thin pool after discountsDeep pool after discounts
A deep pool fixes nothing if the budget is under market, and a full budget fixes nothing if the pool is not there.

Score each dimension green, amber or red:

DimensionGreenAmberRed
Pool per seat, after discounts400 or more160 to 400Under 160
Competing demandPostings under 5% of pool5-15%, or one employer near a fifthOver 15%, or one employer a quarter or more
Pay gap vs posted midpointWithin 5%5-15% underMore than 15% under
Start vs modeled monthOn time or early1-2 months late3 or more months late

The pool cut lines are multiples of the roughly 40 sourced candidates one hire uses up: 10x for green and 4x for amber. They, and the demand and pay cut lines, are my working thresholds, not a published standard. Adjust them for your own market and say that you did. Competing demand never produces a verdict on its own. When demand is red, move pay and time one band worse each, because being outbid costs either money or months.

How the scores map to verdicts:

  • Not fillable as specified: the pool is red; or the pool is amber and pay is red; or a leadership seat sits outside the main hub and time is red. Change the seat itself (its location, level or remote policy), then score it again.
  • Reprice: pay is amber or red while the pool is green or amber; or the seat will clearly need a recruiting fee the plan does not budget for. Raise the base, add a fee line, or both.
  • Slips: pool and pay are fine but the start month is late. Move the start month.
  • On plan: every dimension is green, apart from demand, which can be amber.
Seat read worksheet, one block per seat
SEAT READ: [Company] / [Round] / Seat [ID]
Role and level:
Location as stated in plan:
Budgeted base:            Modeled start month:

POOL
Raw count (country):      Hub share (25-sample):      Metro count:
Non-candidates struck in sample (x of 25):      Discount applied:
Discounted pool:          Score (G/A/R):

DEMAND
Open postings, same profile and location:       Postings / pool:
Top current employers in sample (count of 25):
Score (G/A/R):

PAY
Postings sampled (jurisdiction):      Median posted midpoint:
Gap vs budgeted base (%):             Score (G/A/R or unmeasured):

TIME
Benchmark time-to-fill (source, edition):       Senior premium:
Start-lag assumption (days, why):
Realistic start month:    Months late vs model:      Score (G/A/R):

VERDICT: On plan / Slips / Reprice / Not fillable as specified
Secondary verdict:
One-line reason:
Model change: [new start month] [new base] [fee line]

Copy one block per seat into the deal folder; leave a field blank and mark it unmeasured rather than guessing.

How to run the seat read, step by step

The procedure takes about an hour to set up, plus roughly 75 minutes per seat for the three counts, plus time for the date estimate. A 12-seat plan is about two working days for one analyst. Run every step on every seat; a plan rarely fails on the seat you would have guessed.

The seat read, from slide to restated runway

  1. Pull the seat list
    Extract role, level, location, budgeted base and modeled start month for every seat, and check that the deck's hire count matches the model's. Done when there is one row per seat and both documents agree. Takes about an hour.
  2. Count the reachable pool per seat
    Query a profile index by title, seniority and country or metro, then read a 25-profile sample to get the hub share and strike out non-candidates. Done when each seat has a raw count, a metro-adjusted count and a written discount. About 15 minutes per seat.
  3. Count competing demand
    Count open postings for the same profile and location, and list the top current employers in the sample. Done when each seat has a postings count, a postings-to-pool ratio and a flag for any employer holding a fifth or more of the sample. About 30 minutes per seat.
  4. Measure the pay gap
    Collect posted ranges from employers in jurisdictions that require good-faith ranges, take the median midpoint, and compare it with the budgeted base. Done when each seat has a percentage gap or is marked unmeasured. About 30 minutes per seat.
  5. Estimate time to signed start
    Take a benchmark time-to-fill for the role family, apply the senior premium, and add a written start-lag assumption for notice and onboarding. Done when each seat has a realistic start month next to the modeled one.
  6. Assign a verdict per seat
    Score the four dimensions green, amber or red, then map them to on plan, slips, reprice or not fillable, taking the most severe. Done when every seat has one verdict and a one-line reason.
  7. Restate the model and the runway
    Move start months, apply a loaded multiplier, add fee lines in hire months, and recompute month-by-month burn. Done when the restated runway and milestone dates sit next to the deck's figure and the Carta seed-to-Series A median.

Sources disagree on which should come first. Startupfundraising.com says the hiring plan should work backward from the metrics needed to raise the next round. Fiscallion says to build the capacity model first. For a seat read the plan already exists, so I take its milestones as given and test whether the capacity it describes can be hired in time to reach them.

Steps 2 and 3 are where most of the hours go. Writing each seat as a plain-English request in Refolk ("senior backend engineers with Rust experience in Berlin who changed jobs in the last two years") returns the count and the sample from one query, which cuts the per-seat time sharply.

How to restate burn and runway from the verdicts

To restate the plan, move each seat's cost line to its realistic start month, apply a loaded multiplier, and add a recruiting fee in the hire month. Then recompute burn month by month and compare two things with the deck: the month cash runs out, and the month each seat that drives a milestone is actually working.

Book cost only from the hire month. As Forecastr puts it, a slip then updates the runway projection automatically. For the loaded multiplier, published guides use 1.25x (lean benefits) to 1.40x (rich benefits) of base, or 1.2-1.3x as a typical figure. For international hires one guide uses 1.3 to 1.6x. SHRM puts contingency fees at 20 to 25 percent of first-year cash pay and retained search at about 33 percent, and specialised verticals such as security command 25-30% even on contingency.

Here is one seat slipping one quarter (derived arithmetic):

InputValue
Base salary (example)$180,000
Loaded cost at 1.3x$234,000/yr, or $19,500/mo
Burn deferred by a one-quarter slip$58,500
Contingency fee at 20-25%$36,000 to $45,000, one-time

A reprice works in the opposite direction. Raising that $180,000 base by 10% adds $18,000 x 1.3 = $23,400 a year, or $1,950 a month, and makes any percentage-based fee larger. Retained search at about 33% of a $180,000 base comes to about $59,400. That is a lower bound, because the fee is charged on first-year cash pay, which can be higher than base.

Then set the restated plan against three reference points:

  • The deck's own runway claim. A common investor checklist item is that the round should fund roughly 18 to 24 months.
  • The seed-to-Series A gap. Carta's median was 1.9 years across 9,843 US rounds in one quarterly read and 616 days in another. Both are longer than the 18 months many plans assume. Carta publishes this quarterly, so re-check it before you cite it.
  • Graduation odds. None of the seed cohorts in a cited Carta series has reached a 30% graduation rate to Series A at the two-year mark. A plan that loses a quarter on its key seats is drawing down a margin that was already thin.

No published figure exists for how much runway should remain after the last critical hire starts. Choose a buffer, state it in the memo, and apply it to every deal the same way.

Restated hiring plan, summary for the investment memo
RESTATED HIRING PLAN: [Company], [Round]
Seats in plan: [n]   On plan: [n]   Slips: [n]   Reprice: [n]   Not fillable: [n]

Deck runway: [x] months      Restated runway: [y] months
Deck milestone date: [month]   Restated milestone date: [month]
Restated milestone vs Carta seed-to-A median: [months of margin]
Buffer after last critical hire (my assumption): [months]

Seats that drive the milestone and are not on plan:
- [Seat ID] [role, location]: [verdict], [reason], [model change]

Fees added: [total]      Loaded multiplier used: [x]
Assumptions not established publicly: start lag [days], pool discount [%]
Questions for the founder before close:
1. Which not-fillable seats can change location or level?
2. Which reprice seats have offers or candidates in hand?

Paste under the use-of-funds section; every number should trace back to a worksheet block.

How the seat read goes wrong

Most failed seat reads fail the same way: a number that looked like evidence turns out to measure something else. These seven failure modes cover nearly every bad verdict I have seen, and each comes with a quick check.

  1. Inflated pool count. Title matching sweeps in founders and people in adjacent titles. The US senior ML sample included a "Founder & Principal AI Engineer", who is not a candidate for an employee seat. Check: read 25 sampled profiles, strike out the non-candidates, and apply that strike rate to the whole count.
  2. Wrong benchmark year. SHRM editions report different medians (about 44 days in one, 39 in the next), and secondary sources disagree on which edition says which. Mixing them understates time-to-fill. Check: confirm the report year on shrm.org and use a single edition across every seat.
  3. Wide posted ranges. A wide band can hide the real rate. In Ontario only 16% of postings marked as salary-transparent quoted exact pay. Check: use midpoints only from jurisdictions with a good-faith range rule, such as California, and mark pay unmeasured elsewhere.
  4. Employer concentration read as supply. One large employer can dominate a pool, as Meta does with 6 of 25 profiles in the UK senior ML sample. Check: count anyone at an employer holding a fifth or more of the sample as competing demand, not available supply.
  5. Blended reply rates. The 21.3% blended reply rate overstates engineering conversion more than fourfold compared with the 4.64% engineering rate. Check: use the rate for the seat's role family in every capacity assumption.
  6. Missing start lag. A model that treats time-to-fill as the start date leaves out notice periods, checks and onboarding. Check: because no published median exists, write the lag next to each seat and make it visible in the memo.
  7. Leadership seats treated like IC seats. There are 237 US ML leaders in Refolk's index, and 64% of the sample sits in the Bay Area. A remote or non-Bay ML leadership seat may not be fillable on schedule at all. Check: score leadership seats against the leadership pool and the 90-day tail, never against the IC pool or the median.
A pool is not supply until you remove the people already being paid to stay where they are.

Two quieter errors are worth naming. First, outbound candidates accept offers at 73% against 81% for inbound, so a plan built entirely on outbound sourcing should expect more late-stage losses. Second, only 7% of nonexecutive positions were filled internally in SHRM's data, so a founder saying "we'll promote into it" is a claim you should test, not a reason to skip the seat.

Before you sign off, and how to keep the read current

A seat read is finished when every seat has a verdict backed by measured numbers and the restated runway sits next to the deck's figure in the memo. Run this checklist before the investment committee meeting.

Seat read sign-off

  • The deck's hire count and the model's seat list match, row for row.
  • Every seat has a metro-adjusted pool count and a written discount, with a 25-profile sample read by a person.
  • Every seat with an employer holding a fifth or more of its sample has that share counted as demand, not supply.
  • Every pay gap uses midpoints from good-faith range jurisdictions, or is marked unmeasured.
  • Every time estimate names its benchmark source and edition, the senior premium, and a written start lag.
  • Leadership seats are scored against the leadership pool, not the IC pool.
  • Every seat has one verdict, a secondary verdict where relevant, and a one-line reason.
  • The restated model books cost from the realistic hire month, uses one stated loaded multiplier, and includes fee lines.
  • The memo shows deck runway, restated runway, restated milestone dates, and the margin against the seed-to-Series A median.
  • Every assumption that is not publicly established (start lag, pool discount, post-hire buffer) is labelled as an assumption.

The inputs change over time, so re-check how each one is produced rather than trusting a stored value:

  • Time-to-fill: SHRM publishes new benchmarking editions. Re-pull the median and confirm which edition you are citing.
  • Pay transparency: states keep adding posting rules, and trackers disagree about which apply. Check a current tracker for the seat's jurisdiction before you treat a midpoint as good-faith.
  • Seed-to-Series A gap: Carta reports quarterly. Use the latest median when you restate runway.
  • Pool and demand: counts move as people change jobs and companies open requisitions. After the round closes, re-run the same seat queries at each board meeting. Comparing the new verdicts with the original read tells you which seats are drifting before the plan misses its dates.

Keep the worksheets. The second time you see the same profile in a different deal, you will already know the pool, the main bidders and the realistic start, and the seat read takes minutes instead of days.

Questions practitioners ask

How do you diligence a startup hiring plan beyond the runway math?

Test each seat against the labour market, not just the budget. For every planned hire, measure the reachable pool in the stated location, count the competing open postings, compare the budgeted base with posted pay midpoints, and estimate a realistic signed start date. Give each seat a verdict (on plan, slips, reprice or not fillable), then move start months and add fees in the model to get a restated runway you can compare with the deck.

How long should a financial model assume it takes to hire a senior engineer?

Plan for about 60 to 90 days to offer acceptance, then add a written lag before the first day. An agency source puts the median for tech roles at about 48 days, with senior roles taking roughly 20% longer, and nearly 40% of senior roles run past 90 days. No published median exists for the time from offer acceptance to start, so state your notice-period assumption next to each seat.

How big does the talent pool need to be for a startup hiring plan seat?

No published standard exists, so anchor on conversion. The published outbound benchmark is 2.5% sourced-to-hired, which means about 40 sourced candidates per hire. My working cut lines are 400 or more discounted profiles in the stated location for a comfortable seat and under 160 for a thin one. Read a 25-profile sample first, because title matching pulls in founders and adjacent roles.

Should the founder's recruiting ability change the verdict on a seat?

No. Keep it separate. The seat read grades whether the market can supply the plan at the stated price and date, which is what an ordinary, competent hiring process would achieve. A founder with an unusual network can beat those numbers, but judge that in its own assessment and treat it as upside. If a seat only works because of founder pull, say so explicitly in the memo.

What do I do when pay is not posted in the seat's country?

Mark the pay gap as unmeasured. Do not borrow a midpoint from another market. Disclosure varies widely: secondary figures citing Indeed Hiring Lab put UK postings with pay at 58% and German postings at 12%. Where disclosure is that thin, ask the founder for offer letters or accepted offers for comparable seats, and keep the seat at amber on pay until you have evidence.

Does a hire slipping a quarter extend the startup's runway?

On the cash line, yes. On the milestone line, no. A one-quarter slip on a $180,000 seat at a 1.3x loaded multiplier defers $58,500 of burn. But the metric that seat was meant to move also arrives a quarter later, and Carta's median gap from seed to Series A is already 1.9 years. Restate both the cash-out month and the milestone dates.

Try it on the search you came here for

Stop building boolean strings. Just describe the person.

Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.

  1. 01Describe them

    One plain sentence. Role, city, stack, stage, whatever matters to you.

  2. 02I read the web live

    GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.

  3. 03You read the shortlist

    Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.

  • No boolean, no filters, no seat to buy. One box.
  • Read at search time, so a profile updated yesterday counts today.
  • Every step visible as it runs, every name with its reason.

500 free credits on sign-up. No card, no demo call. See real searches.

Read next