Take the Dashboard Tile Challenge! ‌‌‌‌‌‌‌‌‌📋‌

[Michal_Gebacki]

Welcome back, Dynatrace Community!

Welcome to the last Community Challenge in its current shape, as preparing an essential overhaul and exciting changes in the formula from June! But before it happens, let's enjoy the edition we've prepared for this month, which is targeting a topic well-known for many of Dynatrace Community users: Dashboards!

As we all know, friendly design, readability, and accessibility are the key fundaments of a good dashboard. Share with us the best and the worst examples of Dashboard tiles you're familiar with! It could be a single example in both categories, but feel free to share with us more of these! Let's try to find a pattern - what are the essential characteristics of a good dashboard tile?  👀

Benefits of taking Community Challenges!

👉 Every participant receives a "Dashboard Tile"

👉 You will also get +100 bonus points for extra activity

👉 Engage with others and have fun!

[MaximilianoML]

Nice topic, as usually!

For me, a good dashboard tile should make things obvious at a glance. It should not make the user stop and decode what they are looking at.

I usually look for three simple things:

1. What is happening?
2. Is it expected or abnormal?
3. Where should I look next?

When a tile answers those questions clearly, it is doing its job well.

On the other hand, the worst tiles are often the ones that show data without really telling a story. A big number without context, a chart that is too busy, or a vague title can make a dashboard harder to use instead of more helpful.

A dashboard should guide the eye and reduce effort. The best tiles feel almost natural to read.

I’ve already built several dashboards, and there is always plenty of room to miss something. Most of the time, we do not get everything right on the first try. There is almost always something to adjust, improve, or rethink, and that is part of the process. The important thing is to stay aware of it and keep learning with every iteration.

[ChadTurner]

🎯Dashboard Design: It’s All About Intent

Dashboard design really comes down to two things: who the audience is and what action you want them to take.

A great example of this is the Dynatrace UFO. It’s incredibly simple—just a rotating set of colors—but it’s highly effective:

• 💚Green = No issues
• 🟡 Yellow = Infrastructure problem
• 🟣 Purple = Service problem
• 🔴Red = Application problem

Why does it work? Because at a glance—whether you’re up close or across the room—you immediately understand the state of your environment. Its purpose is clear: grab your attention and direct it to what matters most, right now.

---

🔍Apply That Simplicity to Your Dashboards

You can bring that same philosophy into your own dashboards:

• Let Dynatrace drive color automatically via Davis AI
• Or define custom thresholds that shift tiles from green → yellow → red
• Use color intentionally to signal health, not just decorate visuals

But beyond color, the content matters even more.

---

📊Focus on What Actually Matters

A dashboard filled with:

• 40 overlapping lines
• CPU, memory, disk metrics
  …often creates more confusion than clarity.

Instead, shift your focus to:

• KPIs that reflect business impact
  • Number of calls
  • Error rates
  • User experience metrics (LCP, VCT)
  • Active user counts
• SLO-driven views that tie performance to outcomes

The goal is simple:
👉Someone should be able to glance at your dashboard and immediately know:

• Everything is healthy (green)
• or
• Something needs attention (red) — go investigate

---

🧭 Build, Refine, Simplify

The reality is: no two dashboards are the same.

What works for one team might not work for another—and that’s okay.

Start with your vision:

1. Build what you think you need
2. Use it in real scenarios
3. Refine it
4. Remove what you thought you needed but never use

Over time, you’ll naturally evolve toward a clean, purpose-driven dashboard that delivers value at a glance.

---

🔥Final Thought

The best dashboards don’t try to show everything—
They show the right things, at the right time, in the simplest way possible.


(Yes it wrote this, Copilot helped me turn my novel on dashboards, into something readable 😂)

[MaximilianoML]

I agree with you both! 

A funny thing I noticed now that AI-written text is everywhere: the icons, the perfectly polished structure, and especially the sudden rise of the “—”.

Most people do not even know how to type one, yet somehow it is showing up in every other post. 🤣

[ChadTurner]

Sometimes I fall into the rabbit hole when typing about Dynatrace, and I always just toss it into copilot to "provide clarity" to what I wrote, then i supply the reader with my thesis and - a statement "I know I might have gone to deep into the weeds, so i asked copilot to simplify it" 😂

[AntonioSousa]

Interesting topic!
I hate the simple traffic lights tiles that everyone of my clients want. When I tell them that they have to define exactly what a green/yellow/red really means, is when the fun begins! But it gets bad when a problem happens and it's green, or when it's red but no one is complaining 🤣

One of the best dashboards in the last years is definitely the Christmas tree that I referenced here. Besides being totally made by a client of ours, it really conveys a quick visibility at the service level, while being appealing, especially in the Christmas season:

In the new dashboards, I really love the "bubble map" tiles. There are specific for conveying geographical information. I have them in several clients, so they can see metric information displayed on a map. The following example shows data layered across several geographical locations in Portugal, for a quick perception of what is happening:

[DanielS]

Agree that the Xmas tree that you shared is awesome!!!

[victoria]

@ChadTurner did a really great job with points about good dashboard tiles.. but I have notes on some bad ones.

• Definitely I have looked at tiles before and just thought "what does this even represent?" That's pretty bad. Even if the tiles are representing something I am unfamiliar with, I should have an idea at a quick glance of what it represents at least.
• Tiles showing TOO MUCH. You can usually do it in multiple tiles and make it easier to understand. When there is too much to look at, you don't know what to focus on and don't really take anything in.

Could find some examples but not sure I want to roast anyone like that 😂 Maybe I'll have to make my own dashboard of "examples of bad tiles"

[dannemca]

I like the way you can define different ways of view your data in the same widget.

The basic Golden Signals, for instance. Easy to catch the total of request with failure vs the total of transactions and its response time in the same graph.

Here we can see the drop on response time during the failed transactions. Being fast is not always a good thing.

[RohitBisht]

There are 2 types in my experience:
1. Good ones:

• Having a structure, key KPI on the main one and drill down dashboard links.
• Trends and comparisons
• Easy to be used as a template, replicate and versioning

2. Bad ones:
           -> Random tiles that feel like brainstorming session and are everywhere on the dashboard. easy to blow up and risk of becoming unusable after a while.

[sujit_k_singh]

A good dashboard tile should be understood in under 3 seconds. If someone needs to "study" a tile, it's too complex — split it up.

[tracegazer]

Hey everyone!

The worst tiles I've come across

1. The infinite table

Tables with 12-15 columns where 80% of the information is noise. The user scrolls horizontally, loses context from the first columns, and ends up understanding nothing. Bonus points when the query returns 500+ rows with no pagination or clear sorting.

2. The orphan number

A Single Value tile showing "4,532" in giant font. Is that good? Is that bad? Compared to what? No thresholds, no trend line, no clear unit. The most useless tile out there because it requires the user to already know the context before looking at it.

3. The spaghetti chart

A time series graph with 10+ stacked series. Each line has a color almost identical to the one next to it, the legend takes up more space than the chart itself, and nobody can follow any individual series. It's decoration, not observability.

4. The decorative Markdown

Markdown tiles with logos, emojis, and lengthy text that take up a third of the dashboard without adding any operational value. They look great in the demo, but during an incident at 3 AM, nobody reads them.

---

The best tiles I've built

1. Single Value with full context

A number with clear thresholds (green/yellow/red), a visible unit, and a background sparkline showing the trend. In under 2 seconds you know: what the value is, whether it's normal, and where it's heading. Example: P95 response time with thresholds at 500ms and 2s — green, yellow, red. No thinking required.

2. Honeycomb for health landscape

When monitoring the state of 30-50 hosts or services, nothing beats the honeycomb. Each cell is an entity, the color indicates status, and the density lets you see the entire environment in a single tile. If you spot one red dot among 49 green ones, you already know where to look.

3. Focused table (4-5 columns max)

The best tables I've built have: entity name, main metric, status (with color), and a drill-down link. That's it. Sorted by severity descending so the important stuff is always at the top.

4. Markdown as structural separator

A short one-line title that divides the dashboard into logical sections: "Infrastructure", "Application Performance", "Business KPIs". It doesn't decorate, it organizes. The dashboard reads like a newspaper: top to bottom, section by section.

---

The pattern I've found

A good tile answers one single question in under 5 seconds. If the user needs to interpret, mentally cross-reference data, or scroll to understand what they're seeing, the tile has failed.

Three rules I always apply:

• One question, one tile. If a tile tries to answer two things, split it into two.
• Color = status. If it doesn't have thresholds or visual health indicators, it's a pretty chart but it's not observability.
• Less is more. Every extra column, series, or element competes for attention. If it's not essential, remove it.

[DanielS]

Personally I love the new gauge tile for single values, it allows to see the value but also the indicator to have a volume idea. With a quick shot you can see if you have the business running.

[Patrick_H]

For me charts are the best and the worst at the same time. They are so flexible/versatile you can basically show any information with them, but on the other sied you also can overdo it to have them so crouded/overloaded that you can't see the big picture anymore from too much data. So it heavily depends on the user configuring it in which direction this goes. 📊

[VenkataSainath]

SLOs, SLIs, and SLAs are key components within Dynatrace dashboards that help define and monitor the availability and performance of applications and services. Additionally, leveraging Synthetic monitoring, service-level insights, infrastructure metrics, and problem detection enables a more comprehensive understanding of overall application health.

We can further enhance observability by incorporating Business Events (BIZEVENTS) and tracking critical error events, which provide valuable context and help improve monitoring and analysis.