A developer on your team writes flawless code but goes silent in the brainstorm and "doesn't sell themselves" in reviews. A designer spots the one flaw in the deck that everyone else missed, then misses the social cue that the meeting has moved on. Neither is a performance problem. Both are what happens when a brain that works differently meets a workplace that only rewards one kind of working.

The quick version

  • Neurodiversity is the simple fact that human brains vary, in attention, sensory processing, social style and thinking. Conditions like autism, ADHD and dyslexia are part of that natural range, not just deficits to be fixed.
  • The shift that matters for managers is from a medical model ("what's wrong with this person?") to a social model ("what in our environment is getting in their way?").
  • The business case is real but easy to overstate: companies running structured neurodiversity programmes report strong retention and quality gains, though the headline figures come from self-selected programmes, not controlled trials.
  • You don't diagnose anyone. You change the process, how you interview, run meetings, give instructions and manage sensory load, so the change helps everyone, not just the person who asked.

The idea in depth: where the word came from, and why the framing changes what you do

"Neurodiversity" is a recent word for an old reality. It was popularised by Australian sociologist Judy Singer, who used it in her 1998 honours thesis at the University of Technology Sydney and built it by analogy to biodiversity, the argument being that, just as an ecosystem is more resilient for its variety, a population of minds might be too (Singer, NeuroDiversity: The Birth of an Idea; note that some autistic scholars now credit the early online autistic community alongside Singer, a useful reminder that even origin stories get contested). The term covers neurodevelopmental differences such as autism, ADHD, dyslexia and dyspraxia. The estimate you'll hear most often is that roughly 15–20% of people are neurodivergent in some form; treat that as a broad order-of-magnitude figure rather than a precise statistic, because prevalence numbers vary widely by condition and by how you count, adult ADHD prevalence alone is estimated anywhere from around 3.5% in a DSM-5 field trial to far higher in clinical samples.

The more useful idea for a manager isn't the vocabulary, it's the model of disability underneath it. The medical model locates the problem inside the person: a condition to diagnose and correct. The social model of disability, born of the 1970s disability-rights movement, locates it in the environment: a person is disabled less by their difference than by stairs where a ramp could be, or a hiring process that screens for confident small talk when the job is deep analysis.

A person isn't disabled by their difference so much as by stairs where a ramp could have been.

So the move is: stop asking "how do I help this person cope with our process?" and start asking "what in our process is doing the disabling?" It's a small reframe with large consequences. If a strong engineer freezes in a whiteboard interview, the medical-model instinct is to coach them on interviews. The social-model move is to question whether a live whiteboard predicts the actual job at all, and to offer a take-home task instead. One fix targets the person; the other fixes the filter, and quietly improves your read on every candidate who interviews badly under pressure. The honest limitation here: the social model can overcorrect. Some conditions carry real distress and support needs that no amount of redesigned environment removes, and pretending otherwise helps no one. The model is a lens for finding removable barriers, not a claim that every difficulty is the building's fault.

flowchart TD
  A(["A capable person
struggles in a situation"]) --> B(["Medical model:
'What is wrong
with them?'"])
  A --> C(["Social model:
'What in the environment
is getting in their way?'"])
  B --> D(["Fix the person
(coach, accommodate,
work around)"])
  C --> E(["Fix the barrier
(redesign the process
or environment)"])
  E --> F(["Change helps everyone,
not just the
person who asked"])
The same struggle, two framings, and only one of them tends to improve the system for the next person. Leaders Loop

The idea in depth: the strengths case, and where the evidence is thinner than the slogans

The commercial argument for neurodiversity is genuine, and it's been made carefully at the top of the management literature. In "Neurodiversity as a Competitive Advantage" (Harvard Business Review, May–June 2017), Robert Austin and Gary Pisano documented how firms including SAP, Hewlett Packard Enterprise and Microsoft reformed their hiring to reach neurodivergent talent, replacing high-pressure interviews with practical work trials, and reported gains in productivity, quality and innovation. SAP's Autism at Work programme, the most-cited example, reported a retention rate of around 90% by 2019, well above typical corporate baselines, though that figure is a single firm reporting on its own programme, not an independent benchmark. For the broader case that neuroinclusion builds real organisational capability, see MIT Sloan Management Review.

So the move is: treat the redesigned hiring process, not "hiring autistic people", as the transferable lesson. The thing that worked wasn't a charitable instinct; it was swapping a poor predictor (interview charisma) for a good one (a sample of the actual work). That change is yours to make on Monday for any role, and it benefits every candidate whose strength is the work rather than the performance of getting the work.

Now the honest limitation, because the strengths narrative is where this topic most often tips into spin. Those retention and quality figures come overwhelmingly from self-selected corporate programmes reporting their own results, not from controlled studies, and not from the average team. A 2024 systematic review of autism and employment in the Journal of Management (Ezerins and colleagues) is blunt: the field is young, the evidence patchy, and the upbeat "autism advantage" framing risks an unfair expectation, that a neurodivergent hire must be a savant coder to be worth accommodating. The defensible claim isn't "neurodivergent people are secretly superhuman." It's "you are almost certainly losing capable people to processes that test the wrong thing."

There's a second, subtler finding worth carrying into every awkward one-to-one. Autistic researcher Damian Milton's "double empathy problem" (Disability & Society, 2012) argues that the communication gap between autistic and non-autistic people is mutual, not a one-way deficit in the autistic person, but two groups with different styles each struggling to read the other. The manager's takeaway is humbling and practical: when a conversation with a neurodivergent colleague feels like hard work, half of that friction is yours. Which means half of the fix is too.

flowchart LR
  A(["Awareness:
brains vary"]) --> B(["Audit the process:
does it test the
actual job?"])
  B --> C(["Change the default:
work samples, written
instructions, agendas"])
  C --> D(["Offer adjustments
to anyone,
no disclosure required"])
  D --> E(["Better defaults
lift the whole team"])
Neurodiversity awareness becomes useful only when it moves from a mindset to a change in the defaults everyone works under. Leaders Loop

A worked example: the interview that almost lost a great hire

Consider a data team hiring an analyst. (The figures here are illustrative, not drawn from a specific company.) Two finalists. Priya is warm, quick on her feet, and breezes through the panel interview, riffing confidently on hypotheticals. Sam gives short, literal answers, doesn't make much eye contact, and when asked "tell me about a time you influenced stakeholders," pauses for an uncomfortable ten seconds before giving a precise but flat reply. The panel's gut read: Priya's the obvious yes; Sam "lacks presence."

The hiring manager, having read about the social model, pauses on that gut call and asks a sharper question: does the interview measure the job? The role is 80% rigorous analysis and 20% writing up findings, almost no live, high-pressure improvisation. So she adds one step: a take-home task using real (anonymised) data, scored blind against a rubric. Sam's submission is the cleanest the team has seen in a year, he catches a sampling error baked into the brief that no other candidate noticed. Priya's is competent but ordinary.

Sam is hired. The accommodations cost nothing: an agenda before each meeting so he can prepare rather than improvise, instructions in writing as well as out loud, a quieter desk away from the walkway. Within two quarters his teammates are quietly adopting the written agendas too, because everyone prepares better with notice. That's the tell of a good neurodiversity move: it starts as an adjustment for one person and ends as a better default for the team. The point isn't Sam's talent, it's that the original process would have filtered him out for a skill the job didn't need, and the fix was a change to the filter, available to any manager, no diagnosis required.

Frequently asked questions

Should I ask someone if they're neurodivergent?

No, not directly, and not as a manager fishing for a label. Diagnosis is private medical information, and many people are undiagnosed or choose not to disclose. What you can do is make it safe and routine for anyone to request what helps them work well ("Is there anything about how we run meetings or assign work that I could change to help you do your best work?"). Universal questions get you the adjustments without forcing anyone to out themselves.

Isn't this just lowering the bar?

It's the opposite, it's removing a bar that was never measuring the thing you care about. Swapping a charisma-heavy interview for a work sample raises the accuracy of your hiring, it doesn't lower the standard for the work. The standard for the job stays exactly where it was; what changes is that you stop also testing for unrelated traits like interview confidence or comfort with open-plan noise.

What if accommodating one person feels unfair to the rest?

The strongest neurodiversity adjustments aren't special treatment for one person, they're better defaults available to all. Sending agendas in advance, giving instructions in writing, offering a quiet space, judging output not optics: none of these are exclusive. This is the principle of universal design, build the ramp and the people pushing prams use it too. If an adjustment genuinely can only apply to one person, that's fine and legal in most jurisdictions; but reach for the team-wide version first.

I'm not an expert, won't I get it wrong?

You'll get some things wrong, and that's recoverable. You don't need clinical expertise; you need curiosity and a willingness to ask "what would help?" and then actually do it. The double-empathy point applies: assume the friction is mutual, stay specific and literal, and don't expect someone to read between your lines. Awareness here is a posture, not a credential.

Where does HR or legal come in?

Formal "reasonable adjustments" (UK) or "reasonable accommodations" (US/Australia) can carry legal weight, and the specifics vary by jurisdiction, so loop in HR or a qualified professional for anything contractual or disclosed-and-documented. But most of the day-to-day moves in this piece are just good management you can make without any process at all.

Related in the Toolkit

Where to go next