The Wrong Field on Your Persona: Designing for Every Education Level

Almost every persona profile I've ever been handed has a line near the top that says something like Education: Bachelor's degree. It sits between the age range and the job title, looking very research-y, and I genuinely can't remember it changing a single design decision. Ever. It's the parsley on the plate of persona fields.

That's not because education doesn't matter. It matters a lot. The trouble is that "education level" is a demographic label standing in for at least four different things, and a persona that records the label instead of the things gives the team nothing to design against. So in this post I want to pull those four apart and look at what each one actually asks of an interface.

Start with who's really reading

Before we get to personas, it's worth resetting the default reader, because the one most teams carry around in their heads is wrong.

In December 2024 the National Center for Education Statistics published the US results of the 2023 PIAAC, the OECD's Survey of Adult Skills and the largest direct assessment of adult literacy there is. It found that 28% of US adults perform at Level 1 or below in literacy, up from 19% in 2017. Those aren't people who can't read. Level 1 readers can read short, familiar texts and pull a single piece of information out of them. What they struggle with is the exact thing most product copy demands: holding several pieces of information at once, comparing them, and drawing a conclusion from the combination.

Health literacy, measured separately, is worse. The 2003 National Assessment of Adult Literacy, which is still the only large-scale US assessment of it, found that only 12% of adults had proficient health literacy, and more than a third were at Basic or Below Basic.

Neither number lines up neatly with degrees. Proficiency and qualifications are related, sure, but a degree in one field doesn't make anyone a confident reader of a denial letter, a mortgage disclosure or a permissions screen in another. That gap is really what this whole post is about.

Education is standing in for four different things

When a team writes "low education" or "highly educated" on a persona, they're usually reaching for one of these, and often all of them at once:

The four vary independently. A surgeon can be an expert reader, a domain expert in medicine, a nervous user of your billing portal, and exhausted at the end of a shift, all at the same time.

That last row is the one personas miss most often, and it trumps the other three. Your education protects you a lot less than you'd think when you're reading a hospital discharge notice on your phone at eleven at night.

Plain language isn't the "low-literacy version"

The natural instinct with a mixed audience is to write two versions: a simple one for the people you're worried about, and a "proper" one for everyone else. The research says you only ever needed the first one.

Nielsen Norman Group tested this directly with professional audiences and published the result in 2017 as Plain Language Is for Everyone, Even Experts. Highly educated readers (doctors, lawyers, specialists) preferred plain, concise text for exactly the same reason everyone else does: it lets them get the information and get on with their day. Their summary is still my favorite one-line argument for plain language: no one has ever complained that a text was too easy to understand.

Christopher Trudeau's 2012 study of legal communication, The Public Speaks, came at it from the other side and found the same thing. Across 376 respondents, around 80% preferred plain English to legal language, and the more specialist the reader, the stronger the preference. I love this finding. The people you'd assume want the dense version are the ones most annoyed by it, because they already know what it would say if it would just get to the point.

So plain language is the default for every persona, not a special accommodation for some of them. What changes across the range isn't how clearly you write. It's which words you can take for granted.

But experts need their own words

This is where "just write simply" goes sideways. Plain language doesn't mean scrubbing out every technical term. Jakob Nielsen made the counter-case in 2014 in Use Specialized Language for Specialized Audiences: when your readers are insiders, their terminology is the plain language. Swapping it for a layperson's paraphrase makes the text longer, vaguer, and a little bit insulting.

I ran into this on an irrigation console I designed for commercial and municipal water managers. The person reading its alerts is deciding whether to send a truck out this morning. An alert that says "Check the solenoid and the decoder address before the next window" is plain language for that reader: short, direct, and actionable. Rewrite it for a general audience ("there may be a problem with a valve component") and it reads more easily while being completely useless, because it strips out exactly what the crew needs to know. And the worst option of all, "An error occurred", manages to fail every reader at every education level. It's almost impressive.

The rule I work by is that jargon isn't the same as technical vocabulary. Jargon is vocabulary the reader doesn't share. The same word can be precise for one persona and a brick wall for another, which is why the persona needs to record which terms each person actually owns.

How someone's day is going beats everything else

The clearest example of this in my own work is a site for an independent healthcare navigator. Its readers are older adults and adult children in the middle of a crisis: a hospital has just said a parent is ready for discharge, or a denial letter has shown up with a deadline on it. Some of these readers are highly educated, and it barely makes a difference. At eleven at night, scared, on a phone, a reader with a doctorate and a reader who left school at sixteen need almost exactly the same page.

That drove a handful of decisions that would otherwise look like plain old styling.

Eighteen pixels is the reading floor. It's set once and never shrunk on any paragraph. Labels and fine print can go smaller, but anything a reader actually has to read can't.

Every price carries its unit, every time: flat fee, per hour prepaid in blocks, per month for three scheduled hours. If a rate can be misread as a flat total, someone tired enough will misread it.

The key sentence gets repeated wherever it's needed. "I advise. You act." sits above the rates as well as on the boundaries page, because it has to be read by someone who's about to pay. It's four words, and it carries the most important thing the site has to say.

And the phone number lives in the header, not the footer. The one action the whole site is asking for shouldn't be waiting at the bottom of a long scroll.

None of that is "dumbed down." Those pages are as rigorous as anything I've built, and every statistic carries its source. They just don't ask for more attention than the reader has left.

One product, both ends of the range

The trickier case is a single product that has to serve a total beginner and a daily expert at the same time, on the same screens. My answer is layering instead of a "simple mode": a front door that asks nothing of the beginner, with depth behind it that the expert can reach without being slowed down.

PropertyReach is a property-data platform used by first-time investors and by wholesalers who basically live in it all day. Behind the search box sit more than 130 filters. A first-timer doesn't know what half of them mean and shouldn't have to. So the dashboard offers preset lists (pre-foreclosures, high-equity, vacant, tax-delinquent) one click away, which gets a newcomer a useful result before they've opened a single filter. The filters themselves live in a categorised rail with a running summary of everything that's applied, so a power user can go as deep as they like without losing track of what they're looking at. It's one interface doing both jobs, no mode switch required.

Platinum Loans splits the same way, by role. The borrower's application asks for one thing at a time (personal details, employment, financials, loan preferences, documents), because someone filling in a mortgage application has no reason to know what a lender will eventually want. The broker, meanwhile, opens every loan into the same structured tabs and has qualifiers to test a borrower against a product before building the file. Same data, two levels of assumed knowledge, and neither persona ever sees the other's view.

What the persona should say instead

If "Education: Bachelor's degree" is the wrong field, here's what I put in its place. Every line is something a designer can act on, and every line can be checked against research instead of assumed:

Persona · Adult child, parent in hospital

What replaces the education line

• Reads: on a phone, late at night, in short bursts between other calls. Skims for the answer, and only reads in full once they trust the page.
• Words they own: "insurance", "discharge", "appeal". Words they don't: "skilled nursing facility", "observation status", "asset limit". Define these inline, every time.
• Confidence: fine with a phone, frightened of forms that might commit them to something. Needs to know what happens after they press submit.
• State on arrival: a deadline is running, and they don't know how long they have. Put the deadline and the phone number before anything else.
• Cost of misreading: high. A missed appeal window can't be undone.

If you only add one line, make it cost of misreading. It tells the team how much clarity is worth on a given screen. A misread marketing tagline costs nothing. A misread dosage, deadline or payment amount costs a great deal, and that screen deserves extra care no matter who the persona is.

Recruit for it, and test understanding instead of approval

A persona is only as good as the research underneath it. Research recruited from a product's own mailing list, or from a panel of confident early adopters, will quietly leave out exactly the readers this post is about. Two habits fix most of that.

First, recruit on purpose. If the product will be used by people with limited literacy or limited confidence, go find some of them deliberately instead of hoping they wander in. Ask about what people do (how they handle paperwork, who they call for help) rather than asking them to rate their own reading, because very few people will answer that one candidly. I wouldn't either.

Second, test comprehension, not approval. Asking "was that clear?" mostly measures how polite your participant is. Instead, borrow teach-back from clinical practice: after someone reads a screen, ask them to explain in their own words what it means and what they'd do next. The gap between what the page says and what they tell you is your finding. It's also the fastest way I know to fill in the "words they don't own" line.

Readability scores are a smoke alarm, not a target

Flesch-Kincaid and its cousins are handy and easy to misuse. They count sentence length and syllables, which is a decent sign that a paragraph has gotten away from you. What they can't tell you is whether the reader knows what a word means. "Your claim was denied under the plan's timely-filing provision" scores as easy reading, and it's a total mystery to most of the people who receive it.

So I treat them like a smoke alarm. When one goes off, I go look. A high grade level on a screen that a stressed, general audience reads is worth rewriting. A low grade level proves nothing on its own.

The accessibility standard is smarter about this than it usually gets credit for. WCAG's Success Criterion 3.1.5, Reading Level, asks that when text needs reading ability beyond the lower secondary level (roughly seven to nine years of school), a supplemental version or supporting content is available. It's a AAA criterion, so few sites are formally held to it, but it gets the shape right. It doesn't demand that every sentence be rewritten for the lowest level. It asks that no reader be stuck with only the hard version.

Numbers have a reading level too

PIAAC measures numeracy alongside literacy, and US adults declined on both between 2017 and 2023. Numbers get less love than words in content design, and they're often where the most expensive misreadings happen.

A few rules have held up across every audience I've designed for. Put the unit right next to the number: "$85" is ambiguous, while "$85 an hour, after the first twenty miles" isn't. Use counts where you can, because "1 in 10 people" is understood more reliably than "10%", and far more reliably than a relative risk. Gerd Gigerenzer's research on natural frequencies is the reference here, and it holds for doctors as well as patients. And do the math for the reader. If the decision depends on a total, a difference or a date, show the total, the difference or the date. Don't make them combine two figures to get a third, since that's exactly the task Level 1 readers find hard.

If you remember one thing

Delete the education line, or at least stop letting it stand in for anything. In its place, write down how people read, which words they own, how confident they feel, the state they arrive in, and what a misreading would cost them. Write plainly for everyone, since experts prefer it too (the more specialist, the more they do). Use the reader's own vocabulary without apologizing for it, and define everyone else's inline. Design for the worst plausible moment, not the average reader on a good day. Layer depth behind a simple front door instead of splitting off a "simple mode." And check understanding with teach-back, treating readability scores as a nudge to look rather than a pass mark.

Your future self, and some very tired reader at eleven at night, will thank you.

Sources

• National Center for Education Statistics, Highlights of the 2023 U.S. PIAAC Results (December 2024): nces.ed.gov
• National Center for Education Statistics, The Health Literacy of America's Adults: Results from the 2003 National Assessment of Adult Literacy (2006): nces.ed.gov
• Hoa Loranger, "Plain Language Is for Everyone, Even Experts," Nielsen Norman Group (2017): nngroup.com
• Jakob Nielsen, "Use Specialized Language for Specialized Audiences," Nielsen Norman Group (2014): nngroup.com
• Christopher R. Trudeau, "The Public Speaks: An Empirical Study of Legal Communication," The Scribes Journal of Legal Writing 14 (2012): SSRN
• W3C, Understanding Success Criterion 3.1.5: Reading Level, WCAG 2.2: w3.org
• Gerd Gigerenzer and colleagues on natural frequencies and risk communication, including "Helping Doctors and Patients Make Sense of Health Statistics," Psychological Science in the Public Interest 8 (2007)