The Safety of No Objections

In February 2026, I built an app more as an experiment than anything else.

I had spent years telling higher ed clients that their websites all looked alike. Over time, that grew to include sounding alike, as I worked with more brands and the websites where those brands had to express themselves. Eventually I realized that, like many things in life, showing would be more compelling than telling. So I decided to see if I could build something that would do just that.

I built the Blanding audit. Paste in a .edu, get a score for how distinctive your language is.

I seeded the leaderboard with 30 or 40 schools to test the analysis, refine the app, and give it context, then shared it on LinkedIn. Then it took off. Turns out, people are curious about these things. They ran their own schools. Then their competitors. Then their design schools, honors colleges, extension programs, research initiatives.

By late August 2026, the rolling leaderboard held over 850 higher-ed web properties spread across more than 600 institutional web domains. It is self-selected, not a census, and the pages were captured at different points in time.

As the numbers increased and the data pool grew, I started to realize the lack of distinctive language was a symptom of a much larger set of issues.

Fair warning: this post wanders through AI search, university committees, and a drawing teacher named Betty Edwards before it gets home.

The entry points are shifting

For most of my career, a website's words were one ingredient in an experience that also included design, motion, interaction, and the feeling of a place coming through a screen.

Research has found that people can assess a web page's visual appeal in as little as 50 milliseconds. The look and feel register almost instantly.

This matters because up until recently, a website was usually the front door to a brand experience. You had to welcome visitors with the right visual experience, and then keep them with the overall experience, which quickly extends into language, navigation, etc.

But AI search (or GEO or AIO or EIEIO) has added a new kind of gate to the entire neighborhood, and this one is all about the language.

Today, a prospective student may form an initial impression of a school without ever visiting its website. In a survey of more than 5,000 high school students fielded in late 2025 and released in 2026, EAB found that 46% were using AI during their college search. Nearly one in five had removed a college from consideration based on information surfaced through AI-generated search results.

They may remove you from their list before they ever reach the website you spent hundreds of thousands redesigning. So what do we do about this new gatekeeper? Turns out, we just have to be more "us." Which isn't a bad thing. But more importantly, we have to answer the real questions our audiences have. They don't care about the mission statement.

These are not keyword searches. They are judgments.

Which schools fit me?

Where can I get in?

Which one is worth the debt?

Where will I be happy?

Among the students using AI in EAB's study, 62% were using it to find schools that were a good fit.

Encoura, grouping the prompts students reported using, found the questions clustering around six areas: college selection and comparison, majors, career outcomes, cost and aid, admissions criteria, and campus life.

And the language students use is often almost painfully direct.

One Halda analysis of 528 queries submitted through the AI search tool on Siena University's website included: "My grades aren't good enough?" "How do I find a roommate?" "How much does the MBA cost?"

That is a very different job than the one most university websites were built to do.

We built websites around the information we have. Offices. Divisions. Programs. Initiatives. A student is not trying to understand the organizational chart. They are trying to make a decision.

AI collapses ten tabs into one answer. It compares. It summarizes. It makes inferences. And, sometimes, it confidently gets things wrong.

But even when the answer is accurate, the school has handed an outside system the job of deciding which facts matter and what they add up to.

The machine may find the story anyway

For a while, I assumed bland language would make a school essentially invisible to AI. If there was nothing distinctive on the website, what could the machine possibly have to say? Would it be forced to rely on outside information sourced from wherever it could find it, if the most trusted resource was lacking in detail and specificity?

So in July, I tested one small version of that assumption.

I asked Perplexity the same question about four liberal arts colleges: "What makes this school different from other liberal arts colleges? Be specific about what distinguishes it."

Four schools. One prompt. One platform. This was not a study, and it does not establish anything about AI search at scale.

All four schools received specific, recognizable answers, including the schools with the lowest Blanding scores in that small sample.

At the time, DePauw's homepage led with "We dream. And we dive in." Perplexity led with its three-school structure, named Honors and Fellows programs, and a residential campus where students live all four years.

The riches were DePauw's all along. Perplexity found them. The homepage did not lead with them.

And I'm not sure it had to.

That is both the relief and the warning.

A bland homepage does not make a school invisible. AI will look beyond it. It can pull from program pages, course catalogs, rankings, journalism, Wikipedia, Reddit, and other people's descriptions of the institution.

The school will still appear. But it may have less control over which facts become the story, how those facts are interpreted, and what a student comes to believe they mean.

AI can find what makes you different. But it cannot decide what you stand for.

That is still the brand's job.

AI doesn't read your homepage. It reads your institution. Which is why this isn't really about homepages, or AI search tactics, or even language. Blanding is the symptom. The disease is a brand strategy too safe to give anyone, human or machine, something to hold onto.

Why the words at the top will always disappoint you

This is the part where I'm supposed to tell universities to fix their taglines, their hero statement, their "About Us" copy. I'm not going to, because this cannot be fixed at these levels, and because the tagline, hero statement, and About Us copy were never the battle.

The battle is the underlying brand strategy: how the school sees itself, and how it understands its audiences and the value it provides to them. And if that isn't developed through a framework of distinction, of specificity, and of proof, then anything can be stacked on top of it, with little judgment applied. When you don't have a clear system within which everything else must be organized, the easiest things to keep stacking are the things no one will object to.

If the website is the front door, it must answer to all visitors. Admissions, advancement, athletics, alumni, trustees, faculty senate, the brand council. Every audience gets a seat, and every sentence gets sanded until nobody objects. Sure, we say we are prioritizing audiences, and we direct certain audiences to "their section" before we start speaking directly to them. This all is fine. But it's what we scatter around the front door that bothers me because it's all mostly a bunch of nothingness…

What survives is the language of no objections.

"A vibrant community committed to excellence."

Nobody fought for that sentence. Nobody fought against it either. It is the definition of a safe sentence.

Across the dataset, I kept noticing where language escaped the sanding, and those sites scored much higher.

MIT has a website about concrete. In the late-August snapshot, it was the highest-scoring property in the rolling Blanding dataset. Number one out of more than 850. Nobody sands a concrete site the way a university homepage gets sanded. It answers to people who care about concrete, so it just says what it knows.

Some of the most specific language appeared at the edges. My hunch is that fewer audiences are looking for it, therefore it is governed by fewer stakeholders. The edges can be sharper because they are… edges.

The edges of an institution sound like someone. The center sounds like everyone, aka no one.

The puddle

For a while I thought that was the whole story. Committees flatten language, the flattening now costs more because machines read words first, the end.

But the more properties people fed into the tool, the more I noticed the scores doing something odd. They cluster. The overwhelming majority of properties land in a narrow band in the middle, and almost nothing breaks the ceiling.

I expected a spread and got a puddle.

That didn't look like a few hundred isolated writing decisions. It looked like a system and an equilibrium that the system rewards. But where does an equilibrium like that come from?

A drawing lesson

The best explanation I've found for all of this comes from a drawing teacher.

I've been rereading Betty Edwards' Drawing on the Right Side of the Brain. Her central observation is that beginners don't fail to draw the face in front of them. They succeed at drawing something else: the symbol of a face they've carried since childhood.

The almond eye. The nose like a golf tee. The concept, not the observation.

Her famous fix is to turn the reference picture upside down. In Edwards' explanation, the symbolic mind has a harder time naming what it sees. Instead of drawing "an eye," the student begins drawing shapes, angles, distances, and relationships.

Students often produce a drawing that surprises them in a single exercise.

Not because the hand improved. Because the seeing did.

Blanding is symbol-drawing in language.

"World-class faculty. Vibrant community. Hands-on learning."

That is not a description of any actual college. It's the almond-shaped eye. It's what an institution writes when it reaches for the concept of a college instead of looking at the one it actually is.

Which means the pattern I've been measuring all year was never just a writing problem. Copy alone can't cure it.

It's a seeing problem.

Edwards makes a larger claim in that book that I love: learning to draw is really learning to see, and learning to see changes how you meet the whole world.

I'd make the same claim about brands.

An artist, over a body of work, isn't showing you her subjects. She's showing you how she sees, and underneath that, how she sees herself. A brand is doing the same thing whether it means to or not.

Audiences don't only receive your tagline. They receive your way of seeing, or the absence of one.

An institution that has never looked hard at itself has nothing specific to show, so it reaches for the category symbol instead. Sometimes sameness is a form of institutional self-protection. Familiar category symbols feel safer than a specific claim someone can challenge.

And I can vouch for Edwards' trick as a painter. I often look at an in-progress painting in the mirror, or upside down, so I can see the painting from a different or fresh perspective, evaluating rhythm, values, composition outside the expectations or conventions my brain may have already been imposing on the process.

That's how so many institutions end up drawing the same face from memory.

Three questions your message has to survive

So how do you know if your brand has done the seeing? Try three questions on your primary message.

Does it stand for anything?

What does it stand against?

And is there proof only you can claim?

The second question is the one that separates the living from the embalmed. It gives the message a boundary. The third keeps it honest.

A message that can't fail can't work.

Skidmore's motto is "Creative Thought Matters." It's about two decades old now, and the detail I find most telling is that the student paper's editorial board once asked, "Does Creative Thought Matter at Skidmore?" They weren't rejecting the motto. They were demanding proof: stop stamping it on every administrative program and contest, the piece argued, and show us where creative thought actually lives in the curriculum. That is what a living message makes possible. It gives people a standard, and it gives the institution a job: show us.

Now try the same test on the genre. In my dataset, homepage after homepage leads with some version of the same claim. "Leading the Future." "Power your future." "Your Future Awaits."

Fine. Show me how.

And to be fair, some of those schools do: further down the page, in the programs and the numbers and the stories. But the sentence travels alone.

And alone, it's another almond eye. A category symbol. A claim anyone can make and nobody can check.

A message has to be defensible. It has to be shown. It has to be provable.

Clichés rarely are.

One is a standard. The others are wallpaper.

Some of the strongest examples in my dataset do not stop at a claim. They point to something named.

Reed calls its required senior project "The Reed Thesis." Hendrix has "The Odyssey Program."

You cannot swap those between schools.

A named program is a sentence somebody fought for.

Reed drew Reed. The generic sites drew "a college."

Distinct compared to whom?

One more turn of the lens.

"Only you can claim" is not measured against the whole world. It's measured against your set.

A student is not deciding between Reed and every property in the Blanding dataset. They are comparing Reed with the much smaller set of schools that made the same shortlist.

Distinctiveness is local.

Which suggests a homework assignment, and it's the linguistic version of Edwards' upside-down drawing.

Swap your homepage headline with your closest rival's. Read yours on their site and theirs on yours. If nobody in your building would notice, your symbolic mind has been doing the writing.

This works in any industry, by the way. Higher ed is where I measured it because the pressures are unusually visible: many stakeholders, tight budgets, and long institutional memories.

Run the same swap on two hospital systems, two regional banks, two nonprofits, two B2B platforms.

I'd bet the puddle is everywhere.

Learning to see

The original tool is free at blandingaudit.com. It only speaks .edu for now, so if that's your world, run your school. If it isn't, you already have the swap test.

Either way, the score matters less than the exercise.

Skip the tagline argument at your next brand meeting and ask the harder question: How do we see ourselves?

Not what do we say. What do we see when we look, and would anyone recognize us from the description?

Because that's what you're showing everyone, in every sentence, whether you mean to or not.