Creating a Custom Dashboard with Advanced Analytics

Custom dashboards let you build your own views of your Robin data: the metrics your organization cares about, arranged the way you want them, using the same data that powers Robin's pre-built dashboards.

This guide takes you from an empty dashboard to your first chart. No prior experience with reporting tools is assumed.

Custom dashboards let you build your own views of your Robin data: the metrics your organization cares about, arranged the way you want them, using the same data that powers Robin's pre-built dashboards.

This guide takes you from an empty dashboard to your first chart. No prior experience with reporting tools is assumed.

What you can build

A custom dashboard holds as many charts, tables, and numbers as you want, across as many pages as you need. You choose which Robin data tables to pull from, how to filter them, and how to display the result.

• Combine desk, meeting room, presence, and custom resource data in one place
• Build any of 17 chart types, plus tables and pivot tables
• Add filters your viewers can change themselves, such as a building picker or date range
• Join two tables together, for example employee records with office attendance
• Keep a dashboard private, or share it with your team
• Schedule it to arrive by email on a recurring basis

For the full list of tables and fields, see Robin's Data Dictionary for Advanced Analytics. Read it before you build, because it explains what each table actually counts.

Decide what to build first

The most common mistake is opening an empty dashboard and starting to drag fields around. Answer three questions first and the build takes minutes instead of hours.

1. What question are you answering?

Write it as a sentence someone would actually ask. "How full are our meeting rooms during business hours?" is a question you can build. "Room data" is not.

2. What is the denominator?

Any percentage needs a bottom half, and the choice changes the answer. "Desk utilization" could mean bookings divided by all desks, or by only bookable desks, or by desks in open buildings. Those give very different numbers from the same data.

Decide which you mean before you build, then write it in the description so nobody has to guess later.

3. Who is going to read it?

An executive summary and a facilities working document are different dashboards. If you find yourself building both at once, build two.

Create your dashboard

There are two ways to start, and they are not equivalent.

Start with Save as if you can. Copying a pre-built dashboard and removing what you do not need is usually faster than building from nothing, and you inherit filters and formatting that already work.

Either route opens the same window.

• My documents is private. Only you can see it.
• Shared is visible to others in your organization with analytics access.

Check this before you click Save. My documents is selected by default. If you are building for your leadership team and leave it in My documents, nobody else can open it, and Robin will not warn you.

Find your way around

The toolbar in the middle is how you add things: Data, Charts, Controls, UI, Layout.

The panel on the right shows settings for whatever is selected. With nothing selected, it shows settings for the page.

The bar along the bottom holds your pages, added with the plus button. On the right sits the Publish button, which matters more than it looks. See Publish before you share.

Set up your data page

Do this before you build a single chart. It takes a few minutes and it is the difference between a dashboard that stays manageable and one that becomes impossible to change.

The idea: load each Robin table you need onto one page, once. Then build every chart from those loaded tables rather than going back to the source each time.

Why it matters

• One place to change things. Filter your data page to the last 90 days and every chart built from it follows. Otherwise you are editing the same filter on eight separate charts.
• Your numbers agree with each other. Charts built from one table cannot drift apart. Charts built separately from the same source can, and will, and you will not notice until somebody adds up your bars.
• Changes flow downward. Add a calculation to your data page table and everything built from it can use it immediately.

Step 1: Add a page and name it Data

Click the plus button in the bottom bar. Rename it by clicking the caret on its tab and choosing Rename.

Step 2: Add a table

On your Data page, click Data in the toolbar, then Table.

The Select source window opens on the Suggested tab, which is empty the first time. Nothing is wrong; it lists sources you have used before.

Click through this navigation flow:

Data sources > Workspaces > Prebuilt > Datasets

Step 3: Click into the table, then click it again

Clicking a table name here does not add it. It opens the table, which contains one item with the same name. Click that item to actually add it.

This trips up almost everyone the first time. If you clicked a table and nothing seemed to happen, look for the matching name and click it.

Your table then loads onto the page.

Repeat for each table your dashboard needs.

Step 4: Hide the page

Click the caret on the Data page's tab and choose Hide page. A hidden page shows a crossed-out eye on its tab while you are editing, and does not appear at all for viewers.

Four things to know about hiding:

• It is not a security control. The data is still in the dashboard. Hiding tidies the view; it does not restrict access.
• Anyone who can edit the dashboard still sees hidden pages.
• At least one page has to stay visible.
• If someone copies your dashboard with Save as, every hidden page becomes visible in their copy.

Add pages and components

With your data page in place, everything else you build is a component placed on a visible page. This section covers adding pages, adding components, and pointing them at your data.

Add a page

Click the plus button in the bottom bar. To rename it, click the caret on its tab and choose Rename.

Two reasons to add a page:

• Length. A single page that scrolls past about two screens is easier to read split in two.
• Theme. When several components answer related questions, give them their own page. A page each for desks, meeting rooms, and attendance reads better than one page holding all three, and it lets someone go straight to the part they care about.

Add a component

Five buttons on the toolbar, each opening a menu of component types:

Point it at your data page

Now build your charts on a visible page. Each time you add one, point it at the table on your Data page instead of going back to Datasets.

1. Go to your visible page
2. Add the element you want, from Data or Charts
3. In the Select source pop-up, click the Elements tab, next to Suggested and Data sources
4. Elements lists what already exists in this dashboard, grouped by page. Open Data and pick your table

Each entry has a Preview option so you can confirm you have the right table before committing.

This is the whole point of the data page. Choosing from Elements links your new chart to that table rather than to the original dataset. Sigma calls the result a child element: the table is the parent, your chart is the child, and changes to the parent flow down.

Go back to Datasets for every chart instead, and each one becomes an island with its own filters and its own version of the truth. That is the dashboard nobody can safely change six months later.

If you build on the data page, move it rather than copying it

Hovering over a table shows a plus icon labelled Create child element, which builds a chart from that table right where you are standing. That is a perfectly good way to work, including on your data page.

What matters is how you get the finished chart onto a visible page.

To move a component: hover over it, open the more menu (three dots), choose Move to, and pick the destination page.

So either build the chart on the visible page and source it from the Elements tab, or build it on the data page and move it across. Both leave you with one table feeding everything. Copying is the only route that quietly breaks that.

Add filters people can use

Filters come from the Controls button on the toolbar. They are worth knowing about because a viewer who can narrow the dashboard themselves asks you for far fewer one-off reports.

There are sixteen types, and most people only ever find the first one. They fall into five groups.

Top N is the quiet useful one. It trims a chart to the largest handful of categories, which rescues any chart with a long tail, pie and donut especially.

Once you add the filter to the page, select the filter, then you select the "Properties" tab on the right. 

You will need to choose the Value source, which allows you to select the dashboard that contains your column for the filter. 

Next, click into "Source column". This is where you select the column to apply to the filter. 

Finally, you will navigate to the "Targets" tab to select which components you want the filter applied to.  You need to choose them manually as you may not want to apply a filter to specific components.

Putting the same filter on more than one page

This is the one place where the obvious action is the wrong one. If you copy or duplicate a filter onto another page, you get a second, independent filter. A viewer sets one, moves to the other page, and finds it still at its default. The two pages then disagree, and the viewer trusts neither.

Use Create a synced copy instead:

1. Hover over the filter
2. Open the more menu (three dots)
3. Choose Create a synced copy
4. Use Move to to send it to the page you want

Synced copies share one identity. Change the value on any of them and every copy updates, along with everything they filter. There is no original and no hierarchy, so deleting one does not affect the rest.

Build the chart on the page where you want it, and pick its source from the Elements tab. That is the only route that leaves you with one table feeding everything.

Choose the right chart

Ask what shape your answer is, then pick from this table. Each of these is covered step by step in the next section.

If your answer is not one of those, the question usually needs narrowing before you pick a chart.

The rest of the chart types

The eight above cover almost every workplace question. Robin offers eleven more, listed here so you know they exist and what they are for. Sigma's documentation covers each in detail.

Charts to be careful with

• Pie and donut stop working past about five categories. Either switch to a sorted bar chart, or cut the list down: add a filter to the component, change its type to Top N, and show only the largest few.
• Gauge needs a real target. Inventing one to fill the dial is worse than showing a plain number.
• Stacked area becomes unreadable past three series. Use a line chart.
• Treemap is marked Beta. Avoid it for anything other people rely on.
• Maps only earn their space across many cities. For one campus, a bar chart by building says more.

Looking for a heatmap?

There is no heatmap chart type, which surprises people because Robin's own dashboards use the look. "Expected occupancy next 7 days" on Desk Insights is a pivot table with conditional formatting. Building it is covered under Pivot table below.

Build each visualization

Every one of these starts the same way and fills in the same kind of slots, so the first one is the slow one and the rest go quickly.

The pattern. Add the component, choose your source from the Elements tab of the Select source pop-up, then fill its properties in the panel on the right. Two buttons do almost all the work:

Once a field is in place, its dropdown gives you Set aggregate to switch between sum, average and count, Truncate date to roll a timestamp up to day, week or month, and Format to control how it displays.

They are listed here roughly easiest first.

Table

The plainest component, and the one to reach for when somebody wants to see the actual records rather than a summary.

1. Add the component from the toolbar, then in the Select source pop-up click the Elements tab and choose the table on your data page.
2. Use the Columns list to add or remove fields.
3. Click a column heading dropdown to sort, or to hide the column.
4. To summarize instead of listing, add a field to Groupings and the table collapses into groups with subtotals.

Tables are the honest fallback. If a chart is not landing with your audience, a well-sorted table often communicates better than a bad chart.

KPI

One number, large. The right choice for the headline figure at the top of a dashboard.

1. Add the component from the toolbar, then in the Select source window click the Elements tab and choose the table on your data page.
2. In Value, click + Add calculation and choose the field you want to measure.
3. Open that field's dropdown and use Set aggregate to pick sum, average, or count.
4. Optionally, in Timeline, click + Add column and choose a date field to draw a trend line under the number.

A KPI with no timeframe on it invites the question "since when?". Either add a timeline or say the period in the title.

Bar

Comparing a measure across categories, and the safest default when you are not sure what to use.

1. Add the component from the toolbar, then in the Select source window click the Elements tab and choose the table on your data page.
2. In X-axis, click + Add column and choose your category, such as Building.
3. In Y-axis, click + Add calculation and choose your measure. Row count is there if you just want to count records.
4. Sort by value using the column dropdown.
5. Under Chart type, switch between vertical and horizontal, and choose no stacking, stacked, or stacked 100%.

Sort by value, not alphabetically. A ranked bar chart answers "who is highest" instantly; an alphabetical one makes the reader do the work. Switch to horizontal when category names are long.

Line

Change over time. Use this whenever your X axis is a date.

1. Add the component from the toolbar, then in the Select source window click the Elements tab and choose the table on your data page.
2. In X-axis, click + Add column and choose a date field.
3. Open its dropdown and use Truncate date to set the grain: day, week, or month.
4. In Y-axis, click + Add calculation and choose your measure.
5. To draw one line per category, set Color to By category and choose the field.

Truncating to week smooths out the weekday and weekend sawtooth that dominates most workplace data and hides the real trend.

Pivot Table

Two dimensions crossed with a measure in the cells. The most versatile component available, and the basis for the heatmap look.

1. Add the component from the toolbar, then in the Select source window click the Elements tab and choose the table on your data page.
2. In Pivot rows, add the field you want down the side, such as Building. Add a second, such as Floor, to nest it underneath.
3. In Pivot columns, add the field you want across the top, usually a date.
4. In Values, add the measure that fills the cells.
5. For the heatmap look, go to the Format tab, choose Conditional formatting, click + Add rule, pick your column, and apply a color scale.

Conditional formatting also offers a single color or data bars. Color scale is what produces the graduated heatmap.

Combo

Bars and a line on one chart with separate scales. The clearest way to show a count and a rate together.

1. Add the component from the toolbar, then in the Select source window click the Elements tab and choose the table on your data page.
2. Set the X-axis to your category or date, as with a bar chart.
3. In Y-axis, add your first measure. This becomes the bars.
4. Add a second measure to Y-axis. This becomes the line.
5. Use each series' dropdown to control which is drawn as bars and which as a line, and which axis it belongs to.

Bookings as bars with check-in rate as a line is the single most persuasive chart in workplace analytics, because it shows reservations climbing while actual use stays flat.

Scatter

One dot per thing, positioned by two measures. Use it to find outliers rather than to report totals.

1. Add the component from the toolbar, then in the Select source window click the Elements tab and choose the table on your data page.
2. In X-axis, add your first measure with + Add calculation.
3. In Y-axis, add your second measure.
4. In Color or the grouping property, add the field that defines one dot, such as Space.

Room capacity on one axis and average attendees on the other, one dot per room, is how you find four-person meetings booked into twenty-person rooms. The dots far below the diagonal are your oversized rooms.showing all of its property slots.

Donut or Pie

Parts of a whole. Only worth using when you have a handful of categories and the split is the point.

1. Add the component from the toolbar, then in the Select source window click the Elements tab and choose the table on your data page.
2. In the category property, click + Add column and choose the field to split by.
3. In the value property, click + Add calculation and choose your measure, or Row count.

Past about five slices this becomes unreadable. If you find yourself squinting at the legend, rebuild it as a sorted bar chart.

Write a description for every element

Every chart and number has a description field. Use it. A percentage with no explanation will be questioned, and answering that question takes longer than writing the sentence would have.

Say what the number counts, over what period, and what is left out. Write it for someone who has never seen the underlying data.

Publish before you share

Your edits save automatically, but they save as a draft. A draft is visible to you and anyone else who can edit the dashboard.

Click Publish, at the bottom right, to push your work out to everyone you have shared it with.

This catches people out. You build a dashboard, tell your team to look, and they see something other than what you just built. Building it and shipping it are two separate actions.

Robin keeps a version history of everything you publish, so you can return to an earlier version if a change does not work out.

A checklist before you share

• Your tables are on a hidden data page, and every chart takes its source from the Elements tab rather than from Datasets
• Every number has a description saying what it counts
• You know the denominator of every percentage, and it is written down
• Bar charts are sorted by value, not alphabetically
• Your date range does not end mid-week or mid-month, which makes the last point look like a collapse
• No table is exposing meeting titles, personal email addresses, or anything else your viewers should not see
• The dashboard is in Shared, not My documents, if other people need it
• You clicked Publish
• You opened it as a viewer would and checked it looks the way you expect

Worked example: average room booking per open hour

This builds one number end to end. Follow it once and the pattern applies to anything else you build.

The question: during the hours our offices are open, how much of each hour is a meeting room actually reserved?

The data: Hourly Space Utilization has one row per room per hour, and a Minutes Booked field between 0 and 60. So the average of Minutes Booked across open hours answers the question directly. A result of 25 means the average room is booked 25 minutes out of every open hour.

Step 1: Filter your data page to open hours

On your Data page, select the Hourly Space Utilization table and add a filter on Office Open, set to true.

This step is the whole ballgame. Every room has 24 rows per day, including three in the morning. Leave the closed hours in and you are averaging across a day that is roughly three times too long, and a room booked 25 minutes per open hour will report about 8. Because you are filtering the data page, every chart you build afterwards inherits it.

Also filter out disabled rooms, which still generate rows: add a filter on Space Disabled Time and keep only rows where it is empty.

Step 2: Add the KPI

Go to your visible page. Click Charts, then KPI. In the Select source window, click the Elements tab and choose the Hourly Space Utilization table from your Data page.

Step 3: Set the value

In the Value property on the right, click + Add calculation and choose Minutes Booked.

Sigma aggregates it automatically, and the default is a sum. A sum of minutes across millions of rows is a meaningless number, so change it: hover over the column name, click the dropdown arrow, choose Set aggregate, and pick Average.

You should now see a number between 0 and 60.

Step 4: Sanity check it before you go further

Ask whether the number is plausible before you make it pretty. A result near 60 means every room is booked solid, which is almost certainly a filter problem. A result near 0 means your Office Open filter did not apply, or you are averaging over the wrong rows.

Cross-check it against Space Insights. If your custom number and Robin's pre-built number disagree, they are using different filters, and you should understand why before anyone else sees it.

Step 5: Add a trend

In the Timeline property, click + Add column and choose Time Local. The KPI now shows the latest period with a trend line behind it.

Use Time Local rather than Date. Time Local is the hour in the building's own time zone, so a dashboard covering several regions still lines up.

Step 6: Name it and describe it

Give it a title someone outside your team would understand. "Average minutes booked per open hour" is better than "Utilization."

Then write the description:

Average minutes a meeting room was reserved during each hour the office was open. Out of 60. Excludes disabled rooms and hours the office was closed.

That sentence is what stops someone asking you about the number in six weeks.

Step 7: Publish

Click Publish. Until you do, your team sees the dashboard without this on it.

Turning it into a percentage

Dividing by 60 turns the same number into a percentage of each open hour. That needs a calculation rather than a built-in aggregate: in the Value property, choose + Add new column instead of picking a column, and write the formula there.

Decide which version you want before you build it. Both are defensible, they are not interchangeable, and the description has to match whichever you pick.

Robin's Custom Dashboards are built using Sigma Embedded Analytics. Use the following links for more information:

• Custom dashboards (Workbooks) overview - use the table of contents on the left for more guides. 
• Function index- Browse this function index by category to learn more about the specific types of functions available to you.