What IQ Do You Need to Be a Software Engineer?
None, and this is the one high-band profession with no credential gate at all, which changes the shape of who ends up in it.
No IQ requirement, and no requirement of any other kind either. Software has no licensing body, no protected title and no mandatory qualification. Nobody can stop you calling yourself a software engineer, and nobody checks.
Occupational data places software developers in a band with a median around 115 to 120, grouped with accountants and managers. That figure is worth less here than in any other profession in this series, for reasons worth spelling out.
The only profession with no gate
Medicine has medical school and registration. Law has qualification and admission. Engineering has accredited degrees and chartership. Nursing has a degree and a regulator. Even piloting has licensing and medical certification.
Software has none of that. Self-taught practitioners work at every level of the field including the most senior, bootcamp graduates sit alongside computer science PhDs, and a large share of hiring never asks what anyone studied.
This matters for reading the occupational figure, because every other profession’s median is inflated by its filters. Medicine’s band partly reflects who can afford a long training pipeline. Law’s partly reflects who could access a training contract. Software has no equivalent squeeze, which means its distribution is wider and less socioeconomically filtered than the others, and the median is a weaker description of the people in it.
One job title, enormous range
“Software engineer” covers work with genuinely different cognitive demands:
- Maintaining internal business tools and integrations
- Building consumer product features against a design
- Distributed systems, compilers, query planners, cryptography
Those are not variations on a theme. They demand different things, and one occupational median describes all of them. Worse, much of the occupational literature predates the current shape of the industry entirely, so the figure describes a profession that has since changed size and composition several times over.
Are technical interviews IQ tests?
Closer than most hiring processes, and this is where software’s missing credential gate reappears in another form.
With no qualification to filter on, the industry built its own filter: the algorithmic technical interview. Those resemble reasoning tests in three ways. They are timed. They present novel problems rather than testing recall. And they assume no company-specific knowledge.
Three things separate them from an actual cognitive test:
- They are heavily coachable. An entire preparation industry exists and it works, which is exactly the property IQ tests are designed to resist.
- They test a narrow slice. Algorithmic puzzles are a small part of most software work and a famously poor proxy for the rest.
- They are not normed. There is no standardisation sample and no percentile, just a hiring bar that varies by company and by interviewer.
So the honest read is that a strong reasoning profile helps with technical interviews and preparation helps more, which is a different thing from what the interview claims to measure.
Curious where you sit? Our 50-question assessment reports verbal, numerical, spatial and memory reasoning separately. It is not a technical interview and will not predict one.
What the work actually demands
The distinctive cognitive load in software is working memory: holding a model of a system in your head while changing one part of it, and tracking how that change propagates through everything else.
That maps onto the working memory index more than onto verbal comprehension or crystallised knowledge. It is also why interruption is so costly in this work in a way it is not in some other professions: the cost is rebuilding the model, not resuming the task.
Which makes the software profile different rather than higher. Law loads on verbal, engineering on spatial, medicine on retained knowledge, software on working memory. A single composite averages exactly that distinction away, as covered in our guide to what a composite hides.
Related reading
Frequently asked questions
What IQ do you need to be a software engineer?
None, and software is the clearest case of that in any high-band profession, because it has no licensing body, no protected title and no mandatory qualification. Occupational data places software developers in a band with a median around 115 to 120, grouped with accountants and managers.
What is the average IQ of a software engineer?
Roughly 115 to 120 in the occupational literature, though that figure is less meaningful here than for most professions. The job title covers everything from maintaining internal business tools to compiler and distributed systems work, and the occupational data predates most of that range existing at scale.
Do you need a computer science degree?
No. Software is the only profession in this series where self-taught practitioners work at every level including the most senior, and where hiring routinely ignores whether a candidate has a relevant degree. That absence of a formal gate is the single most distinctive thing about the field.
Are technical interviews IQ tests?
Not formally, but algorithmic interviews resemble reasoning tests more closely than most hiring processes do. They are timed, they present novel problems, and they assume no company-specific knowledge. The differences are that they are heavily coachable, they test a narrow slice of the work, and preparing for them is itself a substantial and well-documented industry.
Is programming harder than other professional work?
Not straightforwardly. What is distinctive is the working memory load: holding a system model in mind while changing part of it, and tracking how the change propagates. That is closer to what a working memory index measures than to verbal or crystallised knowledge, which makes the cognitive profile different rather than higher.
The bottom line
No IQ is needed, and uniquely in this series, no credential is either. That absence is the most interesting fact about the field: with no licensing filter, the distribution is wider and less socioeconomically selected than medicine or law, which makes the 115 to 120 median a weaker description of who is actually doing the work.
What the industry built instead is the algorithmic interview, which resembles a reasoning test while being coachable, narrow and unnormed. And what the work really loads on is working memory, which is a different profile from the other professions rather than a higher one.
See your profile by domain
Our 50-question test reports verbal, numerical, spatial and memory reasoning separately, not just a single number.
Start the free testSources
- Gottfredson, L. S. (1997). Why g matters: The complexity of everyday life. Intelligence, 24(1), 79-132.
- Hauser, R. M. (2002). Meritocracy, cognitive ability, and the sources of occupational success. CDE Working Paper 98-07, University of Wisconsin-Madison.
- Sackett, P. R., Zhang, C., Berry, C. M., & Lievens, F. (2022). Revisiting meta-analytic estimates of validity in personnel selection. Journal of Applied Psychology, 107(11), 2040-2068.
- Wechsler, D. (2008). Wechsler Adult Intelligence Scale-Fourth Edition (WAIS-IV). San Antonio, TX: Pearson.