A dataset is a collection of dimensions and measures from your semantic layer that you import into Excel — build it once, reuse it anytime.
Astrato names each one for you (Dataset1, Dataset2, …), or you can give it your own name.
Pick your data, and pull it into the sheet whenever you need it. To pull it in, click a cell and import it — or reference it in a formula, e.g. =ASTRATO.TABLE("Sales").
Create a dataset
Before you start: you’re connected to Astrato (see Connect to Astrato).
Open the Astrato panel and go to Create your first dataset.
Pick a semantic layer.
A semantic layer is your organization’s governed Astrato data model — the single set of dimensions, measures, and fields that also powers your dashboards and reports. You pick it first because it decides which dimensions, measures, and fields you’ll have to choose from. No semantic layer chosen, nothing to pull.
Tip: Not sure which semantic layer to pick? Open the dashboard in Astrato and check its data source / semantic layer in the dashboard’s settings, then choose the same one here so your Excel numbers match the dashboard.
Result: when you pick a semantic layer, the Data tab opens automatically. This is where you pick the data you want to pull into your workbook.
Picking your data
Which do I use?
Dimension — the “by what”. Attributes you group or break results down by: region, month, product, customer. Reach for these when you want rows or categories.
Measure — the “how much”. The numbers, already calculated in the semantic layer: revenue, order count, average deal size. Use when you want an aggregate, a total, or a calculation.
Field — a raw column straight from a table. Use only when you need underlying detail that the dimensions and measures don’t already expose.
Filter — narrows what you pull back (e.g. this year only, EMEA only, top 100 rows). Use it to keep results relevant and small.
Tip: Most datasets are a few dimensions plus one or more measures. Add fields only for raw detail, and filters to keep the result focused.
For this example workbook, we picked Age Groups as a dimension, and #customers, %Gross Profit, and Customer LTV as measures.
Dimension settings
Click the arrow at the end of a dimension label to open its settings menu.
From here you can:
Display the dimension with the display toggle.
Set the custom Format type.
Allow it to include Null values.
Use a Custom label for the dimension name in your workbook.
Measure settings
Click the arrow at the end of a measure label to open its settings menu.
From here, you can:
Give it a Custom label for the measure name in your workbook.
Change the value Format.
Allow the measure to show Null values.
Apply nested calculations for the measure.
Apply advanced calculations for the measure.
Sort measures
You can easily sort measures from the Data tab:
Rearrange the order in which sorting is applied to the dataset.
Sort the measure value from lowest to highest and vice versa.
Limiting
You can also limit the data you see in the dataset you pull into your workbook.
From this menu, you can:
Hide zero values from your table.
Show Top N dimension values (e.g. Top 20 + Others summarised). This displays only a specified number (N) of items based on their ranking. In the Top N sub-menu: choose whether to show other dimensions, set their position, and give them a custom label.
Rank measures by Ascending or Descending order and sort the chart by rank.
Two related limits live nearby:
Query timeout — the maximum time a query is allowed to run. Valid values are 120–600 seconds.
Visualization limit — cap the rows returned by sorted value. Maximum permitted value is 100,000.
Filters
In the Filters tab that sits next to the Data tab, you can apply different filters to your dataset.
To add a filter:
Select field — pick the dimension you want to filter by.
Pick the operator, which is by default set to =Equals.
Choose the value: change the drop-down icon to filter the whole dataset and pick the value from the list, or filter a variable (not created in this view).
Click Apply.
You can also apply Result filters, which filter the totals after they’ve been calculated. Like the filters above, you apply them to the whole dataset or to a variable by switching the drop-down icon.
Time travel (Snowflake only)
Time Travel lets a dataset return your data as it was at an earlier point in time, rather than as it is now — handy for historical comparisons or checking figures as of a past date.
Important: Time Travel is a Snowflake feature. It only works when your data is on Snowflake (on tables, not views), and only as far back as your Snowflake retention allows (up to 90 days).
In the dataset’s Filters panel, choose a Time Travel mode:
None — off (default). The dataset uses current, live data.
Date — pick a fixed date and time; the dataset returns the data as it stood at that exact moment.
Offset — set a relative time back, such as
-1d(one day ago) or-1h(one hour ago). This shifts as time moves — “1 day ago” is recalculated on each refresh.
Changing a dataset’s connection
If you have more than one connection, the dataset editor shows a connection dropdown at the top, and each dataset remembers which one it uses.
Important: Switching a dataset to a different connection resets which semantic layer it points to, along with the dimensions, measures, and fields you’d selected. This is on purpose: data models don’t carry across environments, so the Add-in starts you fresh rather than mixing them up. Your semantic layers themselves aren’t touched — they live in Astrato and are read-only. Only the dataset’s selection is cleared.
Save or start over
Once you’ve picked your dimensions and measures and added a filter (if needed):
Accept (the check) saves your changes.
Cancel closes the editor and puts everything back the way it was.
Result: once saved, the dataset appears in your list. Click the table icon next to it to drop its results into the sheet at your selected cell.
Full details on inserting: Insert a dataset as a table.
Refresh modes: how often a dataset refreshes
Each dataset has a refresh mode — you decide how live it is. This is the canonical explanation; other articles link here.
Note: Cached data is the result of your last refresh, stored inside the workbook. That’s why the numbers still show when you’re offline, or when you send the file to someone who doesn’t have your connection — they’re reading the cached values, not re-querying Astrato.
Refresh manual — only ever show the cached data. The dataset never re-queries Astrato on its own; what’s in the sheet is what you get. Handy when you share a workbook and want recipients to see your figures without using your connection.
Refresh when no data exist — fetch fresh data only when a formula needs a value that isn’t already cached. This happens automatically the first time a formula runs, or when you use a new filter, variable, or parameter combination it hasn’t seen. If a matching result is already cached, it’s reused instead of re-fetched. You can always force a fetch with the refresh icon.
Refresh always — fetch fresh data whenever the sheet recalculates. Excel recalculates whenever something a formula depends on changes: you edit a referenced cell (for example, one driving an Astrato variable), you press F9, or the workbook recalculates on open. With Refresh always, each of those can pull fresh data from Astrato.
Refresh a dataset
Hit the refresh icon in the dataset (or in the list). If you’ve used the same dataset a few different ways in one workbook — different filters, variables, or parameters — refresh brings them all up to date at once.
Inspect a dataset
When you create a dataset, you can inspect it in an expanded view. Click the drop-down arrow next to the dataset name to expand it.
From the expanded view you have quick links to your Semantic Layer and the Query Log in your data warehouse.
Unlink a dataset table
When you insert a dataset as an Excel table, the Add-in keeps it linked to the dataset so it can refresh and stay in sync. From the expanded dataset view, you can Unlink it.
Selecting Unlink breaks that link: the data already in your sheet stays put as a normal Excel table, but the Add-in stops refreshing or managing it. Your dataset isn’t affected, and nothing is deleted.
Use it when you want to freeze the current figures — a month-end snapshot, a file you’re sending to someone without Astrato, or a table you want to edit by hand without it being overwritten on the next refresh.
Note: The Add-in renames the table to Unlinked_<name> when you unlink it.