Building for Neurodiversity

The number tends to arrive before the person: thirty per cent.

Neurodiverse teams are said to be thirty per cent more productive. The claim has travelled through conference decks, recruitment campaigns and consultancy articles with the efficiency of a statistic nobody wants to slow down and examine. It sounds precise. It also makes a useful slide.

Its origin is narrower and more interesting. In 2017, Robert Austin and Gary Pisano wrote about an HPE programme that had placed more than thirty people in software testing roles at Australia's Department of Human Services. Preliminary results suggested that the neurodiverse testing teams were thirty per cent more productive than other teams.

Every qualifier earns its place. Preliminary results. One programme. One employer. One kind of work. A sample of roughly thirty people.

The observation deserves attention at that scale. It suggests that a carefully designed programme can reveal performance which conventional hiring misses. It cannot carry a universal claim about neurodivergent employees, much less a general promise about creative teams. Asking it to do so turns a promising case into marketing theatre.

The case for neuroinclusion grows stronger once that borrowed multiplier stops doing all the work.

Start with the gap that can be measured

The employment record is less flattering and better established. Office for National Statistics data for the year ending June 2021 recorded an employment rate of 29.0% among disabled people whose main impairment was autism. The House of Commons Library reported that the rate reached 34.0% in 2024/25, compared with 55.3% for disabled people overall and 82.0% for non-disabled people.

The Buckland Review adds a sharper detail. Autistic graduates were twice as likely to be unemployed fifteen months after graduation as non-disabled graduates, with 36% in full-time work. The same review reported that autistic people faced the largest pay gap of all disability groups, largely because they remained under-represented in senior, better-paid roles.

These figures establish a persistent access problem. They cannot tell an agency why a particular candidate disappears between application and offer. That requires looking at the process.

Creative companies are fond of describing talent as their only product. Their recruitment methods often give talent a surprisingly small role.

One measurable filter sits at the interview table

A standard interview asks a candidate to perform several jobs at once. Interpret an ambiguous question. Read the room. Maintain the expected level of eye contact. Decide how much confidence looks confident. Produce a concise example at speed. Make the panel feel that an eight-hour airport delay together would be enjoyable.

Most of these tasks sit beside the work being hired for.

Research by Noah Sasson and colleagues shows how quickly social style can overtake substance. In one of three linked studies, 214 observers rated forty adults, half autistic and half non-autistic, after seeing ten-second samples in five formats: audio, video, audio-visual, a still image and a written transcript. Autistic adults received less favourable ratings across every format except the transcript. When only the content of the speech remained, the difference disappeared.

The study examined social impressions rather than job interviews, so its relevance needs discipline. It shows that judgements about approachability and likeability can form within seconds from delivery cues while leaving the substance unchanged. An unstructured interview gives those judgements plenty of room to influence a hiring decision, then calls the result culture fit.

Katie Maras and colleagues tested a practical response in a 2021 study of fifty adults, twenty-five autistic and twenty-five non-autistic. Employment experts, unaware of participants' diagnoses, first rated answers to standard interview questions. Six months later, the researchers ran a second interview using adapted questions with clearer context, broken-down prompts and requests for specific examples. Ratings improved for both groups and improved most for the autistic candidates, reducing the performance gap.

That result matters because a more accessible format produced better information about everyone. The standard interview had been presented as a neutral instrument. The adapted version revealed how much noise the instrument contained.

Test the work, not fluency in the ritual

The Buckland Review reached a similar practical position through evidence gathered from autistic people, employers and support organisations. Its recommendations include clearer job descriptions, questions supplied in advance, practical assignments and work samples that resemble the role.

For a creative agency, this should be familiar territory. A copywriter can respond to a brief. A designer can explain a visual decision. A strategist can turn contradictory evidence into a point of view. A producer can show how a plan changes when the budget does. These tasks expose judgement, craft and reasoning. A broad question about personal weaknesses mainly exposes who has rehearsed the expected answer.

Structure begins with deciding which evidence matters before the candidate enters the room. Conversation still has a role, alongside criteria tied to the work. Work samples should have a clear scope and reasonable time demand. Reviewers should score the same dimensions. Candidates should know the format and what good performance looks like.

The work sample can introduce its own distortion if nobody designs it properly. An open-ended task rewards the candidate with the most spare time. A vague brief rewards familiarity with the agency's private habits. A request for a polished campaign can become several days of unpaid labour disguised as assessment. None of that provides cleaner evidence.

A useful task resembles a real decision from the role, fits inside a stated time limit and gives every candidate the same essential context. The scoring should separate the parts the agency actually cares about. A strategist might be assessed on the quality of the question, use of evidence and clarity of recommendation. A designer might be assessed on hierarchy, craft and the reasoning behind a choice. Finish should count only where finish belongs to the job.

The format may also need more than one route. Someone could explain a decision in writing, through annotated work or in a short conversation, where the role permits it. Different routes do not lower the standard. They remove irrelevant barriers around the evidence.

The result is a selection process that can explain its own decision. Agencies usually demand that level of rigour from a media recommendation. Talent deserves at least the same courtesy.

The portfolio owns the conditions

An individual agency can improve an interview. A network can change the system around it.

Small agencies hire irregularly, often against an urgent account need. The team reaches for the familiar interview because it already exists and everybody is busy. A parent group has enough hiring volume to build shared work-sample tasks, train assessors and compare where candidates leave the funnel. It can also keep strong candidates visible across several agencies when the first vacancy proves to be the wrong fit.

This is where standardisation becomes useful. Each market will have different employment data, disclosure norms and legal duties. The network can still apply common process measures: which roles used a work sample, whether questions were provided in advance, how long each stage took, what percentage of candidates withdrew and where scoring patterns diverged between assessors.

Common standards do not require identical tasks in London, Johannesburg and Toronto. They require comparable evidence. Each agency can use a locally relevant brief while the group agrees on scope, time demand, assessment dimensions and adjustment options. That creates enough consistency to see whether one office is selecting for the work and another is selecting for familiarity with its preferred social style.

It also gives the group somewhere to put institutional memory. Assessment tasks can be reviewed after each hiring round. Scoring guidance can improve when assessors interpret a criterion differently. Candidates who meet the standard can remain visible across the portfolio, subject to their consent, instead of restarting from zero when another agency has the better vacancy.

Those measures travel better than a headline employment figure collected under one country's definitions. They also point towards action. If candidates repeatedly exit after a group exercise, investigate the exercise. If one assessor consistently scores communication lower than the rest, examine what communication means in that role. If adjustments depend on disclosure, make the useful parts of those adjustments standard practice.

The network has another advantage. It can distribute people across real teams rather than placing them in a showcase programme. A separate initiative may photograph well and still leave the main operating model untouched. Inclusion becomes credible when neurodivergent employees can enter ordinary roles, move between agencies, win promotions and access support without becoming proof of the programme's virtue.

The work changes after the hire

Recruitment is only the first test. A person can survive the interview and still enter an environment built around verbal briefs, open-plan concentration, spontaneous feedback and decisions recorded mainly in the memories of whoever attended the meeting.

The CIPD's 2024 Neuroinclusion at Work research surveyed more than 1,000 employed adults, including 790 neurodivergent respondents. Only 52% described their organisation as open and supportive enough for people to discuss neurodiversity. At team level, the figure was 54%. Almost a third had not told a line manager or HR about their neurodivergence, with fear of assumptions, stigma and career damage among the reasons given.

That is an operating issue. Support which depends on confident disclosure will miss people who have good reason to remain private. Agencies can reduce that dependency by improving the defaults.

A written brief should define the problem, audience, constraints and decision required. Meeting agendas should state what the room needs to resolve. Teams should accept written, visual and verbal contributions. Feedback should address the work directly. Priority changes should be visible. Quiet space, remote options and sensory adjustments should sit inside normal resourcing rather than arrive as favours negotiated during a difficult week.

None of this assigns a standard set of strengths to a diagnosis. Birkbeck's Neurodiversity at Work research found that employers recognised qualities including creativity, hyperfocus, innovative thinking and detailed processing among neurodivergent workers. The report also stresses variation between individuals. A strength becomes useful when the role needs it and the environment allows it to appear.

Progression deserves the same attention as entry. The Buckland Review's pay-gap finding points towards a workplace that may hire autistic people without moving enough of them into senior roles. Creative companies should inspect what promotion quietly rewards. Visibility in the office, confidence in unplanned meetings, availability for late social events and a manager's sense that somebody is leadership material can all become unofficial criteria. They are rarely written down, which makes them difficult to challenge.

Promotion criteria should describe the contribution required at the next level. Creative leadership can appear through judgement, craft, mentoring, reliable decisions, client trust and the ability to improve a team's work. Line managers should document feedback, give examples and agree what progress would look like. Important opportunities should not circulate mainly through the people who happen to be present when a senior leader has an idea.

This matters beyond fairness to one employee. An agency that admits different thinkers at junior level and filters for one communication style at senior level gradually restores the sameness it claimed to value. The hiring campaign remains visible. The loss of perspective happens quietly.

The same rule applies to creative output. Neurodivergent colleagues may notice inaccessible language, sensory overload or audience assumptions that others miss. Their presence does not replace research, accessibility expertise or paid consultation. One employee cannot represent every customer who shares a label.

Measure what the organisation controls

The thirty per cent claim survives because it promises an outcome without asking many questions about the system producing it. Better evidence leads somewhere less convenient and more useful.

It asks the organisation to examine who reaches the interview, what the interview rewards, how work is briefed, where decisions live, who advances and which adjustments remain available without disclosure. These are choices a creative network controls.

British employment data has recorded the same broad exclusion for years while agencies have described talent as scarce. The shortage deserves a closer diagnosis. Some of the people the industry says it cannot find may already be standing outside processes designed to recognise someone else.

‍