Skip to content
OpenNow accepting new client engagements
All posts
Playbook3 min read

Admin dashboards ops teams actually use: five rules for custom web apps

Most internal dashboards get opened twice and abandoned. Five rules we follow so the admin becomes the place your operations team actually works.

mrkd teamOctober 12, 2026

Every company has a dashboard graveyard. Someone built it, it was impressive in the demo, and three months later the team is back in spreadsheets.

Dashboards fail for predictable reasons. Here are the five rules we follow so they don't.

1. Start from a decision, not a dataset

The wrong first question is "what data do we have?" The right one is "what does this person decide every day, and what do they need to see to decide it?"

An operations lead deciding which bookings to chase needs a different screen than a finance lead checking cash position. We design each view around the decision its user makes, then work backwards to the data. Reports and exploratory charts come second.

2. Put the action next to the number

A dashboard that shows a problem and then sends you to another system to fix it is a dashboard people stop opening.

If a number is red, the thing that turns it green should be one click away: approve the request, reassign the ticket, chase the invoice, flag the order. This is where dashboards and internal tools start to merge — and where they become genuinely useful.

3. Fewer numbers, more context

Ten KPIs on one screen means none of them gets attention. We aim for three to five headline numbers per role, each with the context that makes it meaningful: versus last week, versus target, versus the same period last year.

A number without context is trivia. A number with a comparison is a signal.

4. Permissions are a feature, not an afterthought

The moment more than one team uses a dashboard, permissions get complicated. Regional managers see their region. Finance sees amounts; support doesn't. Contractors see their own work.

Off-the-shelf BI tools often struggle here, which is one of the most common reasons we end up building custom. We design role-based access from the first prototype, along with audit logs for anything that changes data.

5. Ship a prototype to real users in two weeks

The only reliable test of a dashboard is the people who'll use it, with real data, doing real work. Not a design review.

We put a working prototype — connected to real data, even if only one source — in front of users in about two weeks. What they ignore tells you as much as what they use. Then we lock scope and build the rest.

Buy or build?

Not every team needs a custom dashboard. Tools like Metabase, Looker Studio, or Retool are great starting points, and we'll tell you when one is enough.

Custom tends to win when:

  • Workflows are complex or specific to how your business runs.
  • Permissions are nuanced, or people outside the company use it.
  • Non-technical users need a polished, fast experience every day.
  • The dashboard needs to take action, not just report.

We wrote more about this trade-off in when to buy, when to build.

The test

A good admin dashboard passes one test: three months after launch, it's the first tab your operations team opens in the morning.

If you want one that passes, see our web apps service, or send us a short brief.

Let's build something

Need a partner like us?

Tell us about your project. We respond within one business day.