Almost every hiring process I have watched closely has a number it is proud of and a number it never produces. The proud number is applications received. The number nobody generates is how many of those applicants could have said yes to the job if it had been offered to them. The gap between those two figures is where most conversations about access to opportunity quietly fail.
"Talent is distributed evenly and opportunity is not" has become the kind of sentence people say at the end of a discussion rather than the beginning. It arrives as a sigh. But read as an operating claim rather than a sentiment, it is unusually demanding. If you actually believe talent is everywhere and access is not, then every place you choose to recruit is a statement about where you think talent lives, and most of those places were chosen for convenience rather than conviction.
A funnel is a volume problem; a door is a permission problem
A funnel widens when more people enter at the top. A door widens when more people can walk through it without having to be someone other than themselves. These are different projects, they cost different amounts, and only one of them is easy to report.
Widening the funnel is cheap, visible and immediately measurable. Post in more places. Attend more events. Rewrite the job advertisement in warmer language. Applications go up and the slide looks good. But if the filters downstream are unchanged, a wider funnel produces exactly one new output: more rejections. You have manufactured disappointment and labelled it outreach. Worse, you have taught a group of people that they applied, were considered and were found wanting, when in fact they were measured against something that was never about the work.
The door is the set of conditions a person must already satisfy to be hireable by you. It includes your screening criteria, the cost of being evaluated, and the terms of the offer itself. Those three things are entirely within an employer's control, which is why they are the only honest place to have this argument. I cannot change who grew up near which industry. I can change what I require, what I charge candidates in time and money, and what a first day assumes about their bank balance.
Most filters measure prior access, not the work
The test I apply to any hiring requirement is blunt: does this describe the work, or does it describe the people who have already been allowed to do the work? Both can be true at once. The question is whether the requirement is the cheapest available way to learn what you need to know.
The usual suspects are familiar enough that we stop seeing them:
- A degree requirement for a role where nobody can name which course content matters.
- Continuous employment, which filters on the absence of caregiving, illness or a bad year rather than on capability.
- A prior title, which measures whether a previous employer was generous with titles.
- "Polish," which is a near-perfect proxy for how many rooms like yours somebody has already sat in.
None of these are useless. They are correlated with something. But correlation is what you fall back on when you have not built a direct measurement, and a direct measurement is almost always available for the price of two well-designed hours. When I argue for hiring for judgment rather than for output, this is part of the reason. Judgment can be observed by giving somebody a real, messy decision and listening to how they reason through it. Credentials cannot be observed at all; they can only be trusted. And now that producing a polished application costs a candidate almost nothing, the temptation to lean harder on credentials as a sorting device is about to get much stronger. That would be exactly the wrong move. Infinite applications make direct assessment more necessary, not less.
Referrals deserve the same scrutiny. A referral is a genuinely useful signal about how someone behaves over time, and I would not give them up. But a referral programme is also the purest funnel-narrowing instrument ever invented, because it reproduces the network of the people you already employ. The fix is not to ban referrals. It is to price them correctly: treat a referral as evidence about character, never as a substitute for the evaluation everyone else has to pass.
Who pays the cost of being evaluated
Employers treat candidate time as free, because it does not appear on any budget line. It is not free, and it is not evenly expensive. Five rounds of interviews scheduled during business hours, an unpaid exercise that takes a weekend, travel that gets reimbursed six weeks later — all of that is affordable to a person with savings, flexible work and no dependents, and close to impossible for somebody without those things. Length of process is a wealth filter wearing business attire.
This is the cheapest door-widening available to any organisation, and it requires no change in values whatsoever. Compress the rounds. Offer slots outside the working day. Pay for any exercise that takes more than an hour or two. Tell candidates in advance exactly what each stage will ask, because the only people harmed by that transparency are the ones who were going to win on familiarity rather than substance. Every one of those moves shifts cost from the candidate onto the employer, which is precisely why they improve access to opportunity — the employer can absorb it.
An offer someone cannot afford to accept is not an offer
Access that ends at the offer letter is access to a lottery ticket. Roles routinely assume things nobody writes down: that you can relocate on your own money, bridge several weeks before the first payment, supply your own equipment, or take an internship that pays in experience. Each assumption is a second filter applied after the hiring decision has been made, and it filters on exactly the thing the whole exercise was supposed to ignore. If social mobility careers are the stated aim, the start-date terms matter as much as the salary band.
Then there is the rationing that happens after people are inside. Who gets the visible project. Who is in the room when the client is unhappy. Who is given something slightly beyond their current competence and covered while they learn it. That allocation is where talent pipeline diversity either becomes real or quietly reverses, and it is usually handled informally by whoever feels comfortable asking. I have written before about what mentorship actually consists of, and the short version applies here: advice is abundant and sponsorship is scarce. Being told you are promising costs the senior person nothing. Being handed the assignment costs them risk.
The filter nobody applies to you
The hardest barrier is the one that operates before any application exists. People remove themselves from consideration for opportunities they have never seen anyone like themselves hold. No employer rejected them, so no employer counts it.
Serving on a board at the FIU Honors College has made that pattern unmistakable to me. The students are not short of ability or ambition. What varies is whether anyone has ever shown them the specific sequence of steps between where they are and a role they would be good at — which firms exist, what the work actually involves on a Tuesday, who to write to. That cannot be fixed by rewriting a job posting. It is fixed by showing up where those students are, repeatedly, and describing the path in concrete terms. It is slow, it does not report well, and it is the part of this that actually moves. I have found that students are more instructive than executives on precisely this point, because they will tell you plainly what they do not understand about an industry instead of pretending.
The narrow promise worth keeping
The honest version of an equal opportunity business is much narrower than the language most organisations use, and far more credible for being narrow. I cannot promise that the world is fair at the point someone reaches me. I can promise that my requirements describe the job, that being assessed does not cost more than a candidate can bear, and that an offer is one a person without a financial cushion can accept.
So the metric I would actually watch is not applications. It is the number of filters in a hiring process that the hiring manager can defend purely in terms of the work. Take one requirement per role per cycle and make it justify itself. Most will not. Each one you remove widens the door by an amount you can genuinely claim — which is more than any of us can say about widening the funnel.