We had two companies ask us to build them better dashboards, but by the end of our scoping sessions, we agreed that the dashboard didn’t solve the problem.

“We push out a bunch of reports, many of them custom, and I don’t even know if the team reads them…And as soon as that happens we move from like data and into following the room’s loudest voice.”

- VP, Food & Beverage Company

“We know where to pull the data from, but we don't know how to make it more actionable and alive… Just because you're scanning a whole bunch of information doesn't mean that you're analyzing it.”

- CEO, Waste Management Company

Both companies spent lots of time building their initial dashboards and reporting mechanisms. The time, energy, and cost meant that they didn’t want to throw it away, but they also knew they needed better outputs.

Most dashboards we see are built around inputs. An analyst pulls the data, a manager checks the sources, a user downloads the files, and then an engineer patches the system. The cycle is endless, always demanding more time from multiple people.

By the time the numbers are right, the team might think the job is done. However, it’s the outputs that matter. We think humans should own the insights, but AI should be the one flagging the data to facilitate the analysis.

For example, a technology company we met didn’t have time to meet with all of their 100+ existing customers each year to renew or upsell. They had a bunch of data to track customer relationships but it was stuck in a dashboard. What they needed was an automated alert when a customer’s KPIs dropped below a certain threshold. The business development team could then react accordingly.

We’ve heard concerns that employees will get lazy and accept whatever the machine says. That’s fair, but that’s why analysis still falls on the employee. When the outputs are working well, they come as tasks assigned to an employee to make a decision.

Based on what we’ve seen work, here’s what the best outputs seem to all have:

1. One click: the employee should know exactly what piece of data to look at without having to open multiple tabs.

2. Rarity: typically 1-2x a week is fine, but if the alert sounds daily then you get into a “boy who cried wolf” situation

3. Solo owner: avoid the bystander effect by singling out one person to do the job.

Here’s what to do next:

• If you’re in a meeting, what are the five pieces of business data that you’d want to get interrupted with? Don’t include stuff like a call from your kid’s daycare or lunch ETA. Get your team to do the same and then compare your lists. Figure out who should own what alert and then how to get them firing from your dashboard.

Some people get weirdly defensive about their dashboards, so we need to hit this note once more. Yes, it’s a great foundation, so keep it. Now build a feature that sits on top of it as if someone is monitoring every number.