Futureful for Employers · the US labor market, indexed to the county
We index a country so you can fill one seat.
Job boards, applicant-tracking feeds and company career sites stream into one analyzed index, geocoded to the county and deduplicated to named employers. Then it narrows to your question: what this role pays where you hire, who else is posting for it, and why your seat is still open.
Pre-launch · the engine is built · founding partners shape v1
Demo access is PIN-gated · real market data · no account needed
Scope · the universe we index toward, not a live-queryable count
The fifth month
The seat is not the cost. The fifth month is.
Overtime that stopped being temporary. A line that runs one person short. A manager who has stopped asking because asking has not worked. Somewhere in month three the req stopped being a hiring problem and became an argument you have every two weeks with the same three people and the same three theories, and nobody in the room has the one thing that would end it.
The answer exists. It is sitting in the postings your competitors published this month, in a federal wage file, and in your own funnel. Nobody has handed it to you in a form you could take into the meeting.
The index
The surface area.
Every figure here is a ceiling: the universe we index toward, not a live-queryable count. We publish the ceiling because it is the honest way to state scope, and because it only ever goes up.
Every figure above is read from one canon module, and a build gate compares that module against a stored floor on every single deploy. A figure that decreases, or a figure that is quietly deleted, fails the build. That is why the scope you are reading is the scope we can still defend next quarter.
The narrowing
Scale is worthless until it narrows.
An index of the whole country answers nothing on its own. The work is the descent: from every indexed source, down to one occupation, one metro, and the wage band you have to defend on Thursday.
- 01
Every indexed source
Job boards, applicant-tracking feeds and company career sites, analyzed into one index and deduplicated to real named employers.
- 02
One occupation
Classified to a standard occupation code, so the same job posted under six different titles counts once instead of six times.
- 03
One metro
The published federal wage band for that occupation in that metro: 25th percentile, median, 75th, 90th, printed with the source table and its vintage, next to the maximum you are actually posting.
Scope today, stated plainly: the federal wage file is national. It carries 393 metro areas and a further 135 nonmetropolitan areas, so a rural county is not automatically a dead end, and 25 of those metros are Californian. What is still partial is the county to metro crosswalk, which resolves 989 counties across 48 states. A county outside that set is told BLS does not publish a wage for your metro and is shown nothing. There is a national row in that file and we will not quietly put it in a slot labelled with your city. Widening the crosswalk is a data task on our side, not a feature you have to buy, and founding partners set the load order.
- 04
The competing set
How many active postings across the whole feed are for the same occupation, and how that count changes when you widen from your county to your metro. Counted from the index, never estimated, each half carrying its own coverage disclosure.
- 05
Your own posting
Your posted maximum against that band, so you can tell whether the seat is open because of the market or because of you.
Any employer.
Any sector. Any size.
There is no sector gate and no headcount tier in this product. A dental group, a distribution center, a school district, a bank and a health system open the same console, in all 50 states. What differs is the question you arrive with.
The role that will not fill
Open five months. Find out whether it is the rate, the market, or your own posting, read against what is actually being posted around you.
Hiring in more than one market
Every market you hire in on one screen, because what a role pays in one of them tells you nothing about the next.
Pricing a role you have never hired
A new function, a new site, a first hire in a discipline nobody on your team has managed. Start from the real local rate instead of a guess.
The shortage you cannot verify
Find out whether the scarcity is real in your market or lives in your job description, read against the postings competing with yours.
The employer who would train
The require-versus-train lever reads against real registered-apprenticeship counts in your market and the published return on training, instead of against a hunch in a meeting.
The seat that keeps reopening
Understand why: the local rate, the employers competing for the same people, and what your own posting is doing to your fill.
In the console today
Everything below is a route you can open right now.
We shipped this list wrong once, in public, for about forty minutes: two entries described data we hold rather than a screen a customer has. Now nothing enters this list unless it maps to a route in the employer console that renders it today, checked against the codebase before it ships. Everything we have not built has its own section further down, and it says so on every card.
Market console
/employer/intelligenceThe wage band, the local employment count, the postings competing with you and the four levers you control, per role, in the market you hire in. Every panel carries a Decompose link that opens the source table, the join and the vintage that produced the number.
The four levers
/employer/intelligencePay. Radius, which widens and never narrows, because a geographic filter pointed at a person is a proxy and we will not ship one. Require versus train, read against real registered-apprenticeship counts and the DOL-commissioned finding of a $1.44 median return per $1 invested. Certifications, showing what competing postings actually state, with the stating denominator printed beside it.
Post a role
/employer/postDescribe the job in your own words and get a structured posting back, entering the same index as every ingested job. Write with AI drafts only from facts already on your form, refuses to run until you have given it enough to work with, and strips any to-be-determined filler line before you ever see it. You edit it. Nothing publishes without you.
One pipeline
/employer/pipelineApplied, shared, interview requested, hired, passed. Stage, flag, search, sort and page all live in the URL, so the board you are looking at is a link you can send to the person who signs. Saved views, a peek drawer, and health tiles that count what is actually happening: awaiting action with its aging delta, shared-to-interview share, and median days to first action.
Discovery, gated by reciprocity
/employer/discoverYou see people who applied to, saved, or were shortlisted for your own postings. Nobody else. The gate lives in the query and is re-proved server-side on every contact attempt, so it is not a setting and it cannot be bought.
An evidence ledger, never a score
/employer/discoverEvery person renders as a list of decomposable facts with their counts and their dates. There is no number anywhere on the card about the human being it describes.
Interview kit
/employer/discoverA role-tailored interview guide, drafted from that candidate's own real facts and saved to their record. Adults only, never a minor.
Your own funnel, measured
/employer/analyticsInterest rate, week-over-week deltas, the oldest wait in days, per-stage conversion and median days to first action, computed from your pipeline and nobody else's. Marking a hire writes a real placement row with the start date and the hourly wage.
Shortlists, with team and messages beside them
/employer/shortlistsEveryone you expressed interest in, with a contactable-adults read. Teammates invited at /employer/team, scoped server-side to your own organization, and one shared inbox for every role at /employer/messages. Minors are guardian-gated, never contactable and never in an export.
The named competitive set
/employer/shedEvery employer posting your role in your market, by name, with the staffing firms flagged, how long each has been hiring for it and how recently they reposted. Exports with its own provenance. It is in the console rail and it exports with its own provenance.
Adjacent occupations, on the employer side
/employer/widen-reqPick one of your live postings and see the occupations the U.S. Department of Labor rates as closest to it in your metro, the skill differences it publishes, how many people already do that work near you and what they are paid. 1,016 federal occupations, 891 of them rated for adjacency, and the rest say so when you pick them rather than returning a silent blank. Your roles are read from your own live postings, and the ones we could not classify to a federal occupation with enough confidence are counted and said out loud instead of quietly dropped.
Local training supply
/employer/supply-formationPick a role and see where its people are actually trained near you: the institutions within fifty miles teaching a feeder program, the federal CIP program families that feed the occupation, how far the nearest one is, and how many people in the metro already do the work. Every figure names the tables it came from, IPEDS HD2023 schools crossed with C2023_A school programs through the CIP-to-SOC crosswalk. What is not there yet is completions per year and a program office to call, and this row will say so until they are.
The Monday email
/employer/analyticsOne scheduled brief a week on what moved in your funnel, sent Monday morning, deduplicated so a quiet week does not mail you and no person is mailed twice. It reads the same figures this console shows you.
The line we do not cross
We never score a person.
No fit percentage. No matching score. No readiness rating. No ranked list of human beings. Not as a setting, not on an enterprise plan, not later.
Enforced in the build, not promised in the marketing.
A gate scans this codebase for any number attached to a person and fails the build if it finds one, so it runs on every promote and nothing ships past it. It reads its banned list from the same rules file the policy lives in, so the rule and its enforcement cannot drift apart.
We shipped a number like this once, and then we measured it.
On the job-seeker side of this product there used to be a percentage that read as a person's fit for an occupation. An audit took it apart and found it was mostly a language model's opinion, resting on an instrument whose top-ranked result did not survive its own published margin of error. It was not a conservative number or a rough one. It was not a measurement at all. We deleted it, and then we made bringing it back fail the build.
So there is nothing to audit.
There is no bias audit to commission here, because there is no automated employment decision tool. You read the evidence and you decide, and you can still explain how you decided, because nothing in the middle of it was a number nobody can account for.
What a person renders as, instead.
An evidence ledger. What they did, when they did it, which of your postings they did it against, and which stated credentials are on file, each with its own count and its own date. It is slower to read than a digit and it is the only version you can defend out loud, to a hiring manager, to a candidate, or to a regulator.
And the reciprocity gate sits underneath all of it: you only ever see a person who reached toward you first. There is no database of strangers here to rank in the first place.
No slot exists
Evidence discipline
A number you can be questioned on.
The enterprise feeds blend and model, then hand you a figure with no parentage. We cite and stamp. When you are defending a pay band, the only figure worth having is the one that survives being asked where it came from. Open any line below for the mechanism.
Every number carries its sourceSource instrument, as-of date and denominator, printed where the number lives. On screen and on paper.Cited<T> · source · asOf · denominator · coverage
Every quantity the market console returns is wrapped in a type that will not compile without its instrument, its vintage and the universe it was computed over. A cell that cannot produce all of them does not render at all, which is why the panel can always answer where a number came from.
Every panel decomposesA link on the panel opens the source table, the join and the vintage that produced the figure you are looking at.five decompose endpoints, one per panel
Wage, supply, competing postings, apprenticeships and certification demand each answer with the instrument, the exact predicate, the denominator and a deterministic path to contest the number. No person data crosses that endpoint, and we never hand-edit a published government statistic: if you believe one is wrong upstream, we escalate it and tell you what we did.
Floors, never estimatesPositive detections read as at least N, labelled on the number itself rather than in a legend.excluded and disclosed, never imputed
Only a minority of active postings state a county, so every county-scoped count in the console is an at-least and is labelled that way on the number itself rather than in a legend. A floor you can see is worth more than a total you cannot check.
Evidence that replaysA rate you cited in March reprints identically in June. What you took to your CFO does not quietly change under you.a vintaged instrument, not a rolling model
The wage band is a published federal statistic with a release name attached, not a blended model that drifts every time we retrain something. That is the difference between a figure you can defend in a room and a figure you have to re-defend every quarter.
Small numbers stay honestA cohort too small to report suppresses rather than misleads, and the suppression is complementary so the hidden cell cannot be back-computed from the margins.a reporting floor, and the hidden count is omitted too
No composite scores, no invented precision, and no number attached to a human being. A count we could not read renders as unavailable, never as a confident zero. Suppression is the plumbing working correctly, not a gap we are waiting to fill.
A count we could not read is unavailable, never zero. That one rule is why nothing on this page is rounder than the thing it describes.
Why this compounds
The other side of this market is not somebody else's product.
An applicant tracking system holds your pipeline and cannot see the market.
A labor-market data vendor holds the market and never sees a hire.
A job board holds the seekers and sells the same shortlist to everyone on your street.
We run both. Futureful is free forever for job seekers, nationwide, and that is not charity: it is why one index has a demand side and a supply side inside it, and why a role you post here becomes real inventory on a surface real people already use to find work.
- 01
A person uses the free product. Real signal about what people can do and where they are.
- 02
An employer posts a role. Real inventory on the free side, never a fabricated listing.
- 03
Both sides sharpen the index: the occupation lexicon, the wage band and the competing set get more defensible every week the system runs.
- 04
The evidence leaves the building. What you hand your CFO carries our source line into rooms that have never heard of us.
A competitor can license the same public data tomorrow. What they cannot do is stand up the other side of the market, because they have no reason for a job seeker to ever open their product.
What we will not claim yet: a candidate pool.
Not built · shaped by founding partners
What we build next is decided in the room, not by us.
None of these exists today. These are the instruments the first cohort is arguing about, in the order they are arguing for them. Every card says so, and it will keep saying so until the route ships.
County coverage for the wage band
The wage file is already national, at 393 metro areas plus 135 nonmetropolitan ones. What is partial is the county to metro crosswalk, which resolves 989 counties across 48 states today, so a county outside it gets the honest empty state instead of a band. Widening that crosswalk is what makes the band answer for every county you hire in, and it is the first thing founding partner money buys.
The agency spread
What staffing firms post for this work in your market, how much of it they hold, and the measured spread against direct employers. We would publish measured posted rates only, never a bill rate or a markup we did not observe.
The ramp calendar
Which month this work is actually posted in your county, so you open the req before the peak instead of during it.
The one-page exhibit
A dated, sourced, portrait, print-first document you hand across the table unaltered: the competing set, the band, who moved and when, with numbered source footnotes carrying their vintages and the timestamp it was generated.
A subject line that earns the open
The Monday email ships today and is in the row above. What is not built is the version that carries a dollar amount and a name in the subject line itself, so the value lands before the open rather than after it.
Pricing
The whole ladder is published, including the enterprise band.
The meter is the labor shed, never the location. A shed is one commute market, so two sites inside the same market are one shed and not two. There is no checkout on this site and nothing here can charge a card, so this is the pricing we intend to run and not a meter you are on today.
One labor shed. No card, no phone field, no sales call. Not a trial.
Unlimited seats. Monthly, cancel anytime. Volume is on the published ladder.
For the largest footprints. A band you can plan against, not a quote you have to extract.
What we refuse to sell
There is no plan, at any price, that ranks a human being.
No fit percentage, no matching score, no readiness rating, no ordered list of people.
No plan surfaces a minor.
No plan lowers our reporting floor.
No plan opens a person who has not reached out to you first.
Money never moves a safety floor here, and that is enforced in our build, not promised in our marketing.
Publishing the price and shipping a working free tier is not a discount. It is a different category of company. Futureful is pre-launch, so there is no checkout on this site and nothing here can charge a card: what is published is the list we have committed to, so a founding partner can plan against a real number instead of a promise.
See the full ladder, every tier, and what each one does not include
The Founding Partner Program
Become a founding partner.
Bring us the req that has been open five months. The first cohort gets an environment scoped to the markets they actually hire in, direct influence over which of the cards above we build first, founding terms locked before public pricing, and setup with a real person. We ask for your actual open roles, so v1 is shaped by the people doing the hiring. No fabricated results and no inflated promises, and we will tell you what we cannot answer for your market before you sign anything.
Live demo · real market data · PIN-gated · honest-empty where the index is thin
Nothing in Futureful scores, rates, ranks or orders a human being. There is no bias audit to commission, because there is no automated employment decision tool.