The Grandma Test for Website Copy

There’s an old piece of advice that you don’t really understand something until you can explain it to your grandmother. I want to argue for a version of it as a practical editing tool for websites: read your homepage aloud to someone with no connection to your industry, and watch their face. Not because your grandmother is your customer. She almost certainly isn’t. The test works for a different reason, and understanding that reason is what makes it useful rather than gimmicky.

Here’s the reason. When you explain something to an expert, you can lean on shared vocabulary, and shared vocabulary hides gaps. If I tell a colleague we’re “leveraging orchestration to streamline deployment workflows,” she’ll nod, because each word is familiar to her and familiarity feels like comprehension. But if I say it to my grandmother, she’ll ask what orchestration means, and then what deployment means, and somewhere around the third question I’ll discover that I can’t actually say what the product does without the jargon — which means I was using the jargon as a substitute for knowing, not a compression of it. The grandmother isn’t testing her comprehension. She’s testing mine. Jargon is a loan, and she’s the audit.

The usual objection arrives immediately: our customers are experts, so why should copy be understandable to a non-expert? Two answers. First, plain language costs experts nothing. A database engineer reading “we back up your data and can restore it in minutes” loses no information compared to a jargon-laden equivalent; she just reads it faster. Nobody has ever bounced off a website for being too easy to understand. The asymmetry only cuts one way: jargon excludes non-experts while plain words exclude no one. Second, the “our visitors are experts” assumption is shakier than it looks. The person evaluating your product is often not the person who will use it. A founder researches tools for her team. A procurement person compares vendors. An investor checks out a portfolio company’s site. A candidate decides whether to take the interview. These people are smart and consequential and do not share your vocabulary. The expert user might tolerate your jargon. The expert user’s boss won’t.

There’s a mechanical way to run the test if no grandmother is available. Go through the copy and circle every word your target reader’s neighbor wouldn’t use at dinner. “Leverage.” “Seamless.” “Robust.” “End-to-end.” “Solutions.” Then, for each circled word, try to replace it with what you actually mean. This is where it gets interesting, because the circled words split into two groups. Some have a plain replacement that’s just as accurate — “leverage” almost always means “use,” and the sentence survives the swap. Those were pure decoration. But some words resist replacement, and when you try to say the plain version, you realize you’re not sure what the plain version is. Those words weren’t decorating a thought. They were replacing one. The grandma test’s real output isn’t a simpler page. It’s a list of things you haven’t figured out yet.

I’ve felt this on my own product. GazeSite scans websites and grades them across areas like accessibility and clarity, and early on my draft copy said it “audits your web presence across multiple quality dimensions.” Every word defensible, the whole thing gelatinous. When I imagined explaining it to someone’s grandmother, what came out was: it looks at your website like a stranger would and tells you what’s confusing, broken, or hard to read. That sentence ended up shaping the product’s clarity checks themselves, because “would a stranger get this?” turns out to be checkable in a way that “web presence quality” is not. Writing for the plainest listener didn’t dumb the idea down. It forced the idea to exist.

I should say what the test is not. It’s not a claim that all products can be fully explained to a layperson, or that technical depth should be hidden. Deep products need deep pages eventually. The test applies to the front door — the headline, the first paragraph, the words a stranger meets in their first few seconds — because the front door’s job is orientation, and orientation must work for people who don’t yet know anything. Past the front door, write for whoever you like. It’s also not a license for baby talk. “We make computers go zoom” fails a different way: it’s condescending, and grandmothers detect condescension faster than anyone. The target register is how a good doctor explains a diagnosis — plainly, respectfully, without a single wasted term of art, but without pretending the subject is simpler than it is.

The reason I keep coming back to this test is that it’s the cheapest usability study in existence. One reader, five minutes, no tooling. And it’s brutally hard to argue with. You can dismiss an analytics chart. You can rationalize a bounce rate. But when a person you love looks up from your homepage and asks, gently, “so what is it you do?” — there’s nowhere to hide. That question is being asked silently, thousands of times, by every stranger who lands on the page. Your grandmother is just the only one kind enough to ask it out loud.