DEV Community

Reviot
Reviot

Posted on

Designing a dashboard a 70-year-old business owner can actually use

Most dashboards are built by people who love dashboards. Charts, filters, tabs, a dozen KPIs above the fold. That works if your users are analysts. It doesn't work if your user runs a bakery, a dental clinic or a family hotel and has five minutes between customers.

That's the user we design for at Reviot, a reputation intelligence tool that shows business owners what's being said about them online. Our internal test is simple: could a 70-year-old owner open it and understand what's going on in under a minute?

Here are the principles we're applying. None of them are new, but they're easy to forget once you start adding features.

1. Answer the question before showing the data

Someone opening a reputation tool has one question: "Is anything wrong?" So the first thing on screen should be the answer, in words, not a chart they have to interpret.

  • Bad: a line chart of sentiment over the last 30 days.
  • Better: "3 new mentions this week. 1 needs your attention."

The chart can still exist. It just doesn't go first.

2. Compact by default, expandable on demand

Progressive disclosure is the whole game. Every item starts as one line: what happened, where, and whether it's good or bad. If the user wants more, they tap and it expands with the full detail.

You don't need a library for this. The native <details> element gets you surprisingly far:

<details class="mention">
  <summary>
    <span class="badge badge--negative">Needs attention</span>
    New 2-star review on Google Maps
  </summary>
  <p>"Waited 40 minutes and nobody apologised..."</p>
  <p>Posted 2 hours ago on Google Maps</p>
</details>
Enter fullscreen mode Exit fullscreen mode
.mention summary {
  cursor: pointer;
  padding: 12px 16px;
  font-size: 1.05rem;
}

.mention[open] summary {
  font-weight: 600;
}

.badge--negative {
  background: #fde2e1;
  color: #9b1c1c;
  border-radius: 999px;
  padding: 2px 10px;
}
Enter fullscreen mode Exit fullscreen mode

It's keyboard accessible, works without JavaScript and screen readers understand it out of the box.

3. Words over numbers, plain words over jargon

"Sentiment score: -0.42" means nothing to most people. "Negative" does. Same with "SERP", "share of voice" or "mention velocity". If a term needs a tooltip to make sense, it probably shouldn't be on the main screen.

4. Colour means one thing

Use colour only for status: good, neutral, needs attention. Not for branding, not for categories. When red always means "look at this", users learn it in seconds.

5. Let AI write the summary, but make it sound human

A lot of what we show is summarised by AI. The trap is text that reads like a machine-generated report: rigid, full of headings, the same structure every time. We aim for the opposite: short, natural sentences that sound like a colleague telling you what happened, with formatting only where it actually helps.

6. Bring the information to them

The best dashboard for a busy owner is the one they don't have to open. A short briefing at a time they choose, and alerts only when something really matters, beat a beautiful dashboard nobody checks.

Wrapping up

Designing for non-technical users isn't about dumbing things down. It's about deciding what matters, saying it plainly and keeping everything else one click away. Power users will still find the detail. Everyone else will finally understand what they're looking at.

If you've designed for a similar audience, I'd love to hear what worked for you in the comments.

Top comments (1)

Collapse
 
nix_ilands_bf7622f8dc1d30 profile image
Nix Ilands • • Edited

Two things I'd add, from the other end of the same problem.

The <details> pattern in #2 does more than progressive disclosure. When every item is one line, you can count the red badges in about a second, so "is anything wrong?" gets answered by five rows instead of one sentence. Cheap, and it survives a user with no patience left.

I'd push on #1 a little. Answering the question before the data is right, but for a shop the expensive loss happens one screen earlier than the dashboard: on the card that links to it. On a marketplace post of mine, 14 views, 3 past three seconds, 0 past ten. Eleven people left before any screen that had an answer on it. Where the first line sits matters, and so does what the link preview says, because that's the screen most people actually see.

Last one: the 70-year-old test is mostly type size. 16px, no grey on grey, tap targets that survive a shaky hand. Words over numbers is right, and it won't help anyone who can't read the words.