Counting Who Applied, Not Who Got In

What the number hides

Most organisations that collect disability workforce data do so at the point a candidate accepts a job. The figure lands in an annual report, sits beside a headline percentage and is used — usually charitably — to assess how the organisation is doing on disability inclusion. The problem is structural: if your data collection begins at offer, you are measuring the candidates who made it through the whole process. You are not measuring what happened before that.

Application rates, withdrawal rates, progression through screening stages, drop-off between one interview and the next — none of this is visible in a single end-point number. And it is precisely in those earlier stages that many of the most significant barriers operate.

Consider what an application form requires before a disabled person has spoken to anyone at the organisation. It may demand information entered in a format that does not work with a screen reader. It may have no mechanism for requesting an adjustment to the process itself — or the mechanism exists but is buried, or asks the applicant to disclose a condition rather than simply request a change. It may have an arbitrary word limit that disadvantages people who communicate differently, or time out during completion. By the time the organisation counts who got through to interview, the people filtered out by these mechanics are already gone, and they are entirely invisible in the data.

Where the barriers concentrate

The gaps between stages tell a different story from the end-point total. If disabled applicants apply at broadly the same rate as non-disabled applicants but are offered positions at a lower rate, that is a signal worth investigating. If the gap opens specifically between application and first interview, the selection process itself is the likely site of the problem. If it opens between first and second interview, the format of those interviews — unstructured, heavily time-pressured, dependent on particular communication styles — is worth scrutiny. If withdrawal rates are higher among disabled applicants at any stage, the experience of moving through the process may be the cause.

None of this analysis is possible without stage-by-stage data. And collecting it requires both a monitoring question asked early — at the point of application, not appointment — and a sufficiently large and consistent dataset to draw any meaningful conclusion. Organisations with low overall recruitment volumes will need to aggregate data across multiple cycles before patterns emerge. That is not a reason to avoid collecting it; it is a reason to start now rather than later.

The Equality and Human Rights Commission, in its guidance on workforce equality monitoring, recommends collecting data at multiple points in the employment lifecycle, including recruitment. The Government's Disability Confident scheme asks employers to commit to inclusive recruitment, but does not mandate stage-level data collection or reporting, which limits what can be inferred from participation in the scheme alone. Voluntary frameworks, however well intentioned, tend to measure commitment rather than outcome.

What earlier data would reveal — and what it demands

Collecting data at application stage immediately raises a question about declaration rates. Workforce monitoring depends on people choosing to share information, and disabled people have good reasons to be cautious: evidence that disclosed disability status influences selection decisions, uncertainty about how data is stored or used, and a reasonable assessment that adjustments will be easier to secure once inside an organisation than during recruitment. Low declaration rates are not a data quality problem to be engineered away; they reflect a trust problem that earlier, better data collection cannot solve on its own.

What it can do is reframe the question being asked internally. An organisation that discovers its application-to-interview conversion rate is materially lower for candidates who declared a disability has a specific, actionable problem to investigate. An organisation that only counts who was hired has no mechanism to find this at all.

Practically, stage-level monitoring requires some investment in HR systems: tracking needs to be linked across stages of a single recruitment process without exposing individual identity, which means either candidate reference numbers stripped of names or aggregate reporting at cohort level. Neither is technically complex. What they require is deliberate design — someone choosing to build it in — rather than a default reliance on whatever the applicant tracking system happens to export.

The information most worth having is not whether your workforce contains a certain percentage of disabled employees. It is whether your recruitment process is doing the filtering for you before the numbers reach the spreadsheet. A hiring rate that looks respectable is not evidence of an accessible process; it is compatible with a process that has already removed a large proportion of disabled candidates before the point at which anyone started counting. The percentage is the residue of the process, not a report on it.

Organisations serious about this will audit their application form for accessibility, make adjustment requests easy and visible from page one, track declaration and progression rates by stage, and treat unexplained gaps in conversion as a prompt for investigation rather than a data artefact. The number that should matter is not how many disabled people were hired. It is how many applied, and what happened next.