Datasets

The Datasets area lets you upload your own lists of properties, buildings or geometry and use them as layers on the map or as an input list for further work. Click Datasets in the left menu to open the panel.

The example on this page

The screenshots below show the same setup throughout: Placepoint Levering AS, a fictional home delivery company with depots in 22 cities. The data is synthetic, but the structure is the same as a production customer runs today, and it explains why each feature exists.

Dataset	Content	Row count
Stops	One point per delivery address, with the customer name, a note for the driver, weekday, start time, sequence number, vehicle and delivery zone	15,800
Depots	One point per depot, with name and address	22
Delivery zones	One area per route, with weekday, week number, colour and the vehicle registration number	316
Drive time 30 min	One area per depot: everything reached within a 30 minute drive	22

All four sit in the same project, and each is switched on and off on its own. The planner sees the whole operation on one map, and operations and management see the same.

Uploading a dataset

Click Upload dataset at the bottom of the panel to open the upload dialogue.

Drag the file into the field, or click Choose an Excel file. Then work through the wizard: choose the sheet, choose how the rows connect to the map, check the column types and give the dataset a name.

File requirements

Requirement	Value
Format	Excel, .xlsx. Other formats must be converted first
File size	Max 30 MB
Rows	Up to 5,000 rows per upload
Sheets	One sheet per upload. You choose the sheet in the wizard
Column names	The first row with content becomes the column names. The names must be unique
Row content	One row per property, address or point

The row limit is not fixed for everyone: the stops in the example are a single layer of 15,800 rows, with a higher limit set for that organisation.

Two ways to connect the rows

The wizard asks how the rows should find their way onto the map, and you choose one of two routes:

• Your own coordinates or geometry. The file contains the geometry itself, such as latitude and longitude, WKT or GeoJSON. The rows appear on the map without any lookup against Placepoint. The delivery zones and the drive time zones in the example come in this way: they are areas that do not exist in any public register.
• Link to Placepoint. The rows are matched against addresses or matrikkelenheter based on address or matrikkel columns. The rows then also carry the Placepoint data about the property. A list of delivery addresses linked this way gives you access to ownership, buildings and zoning for every address in the list.

You can use both on the same dataset: matching connects the row to the property, and your own geometry gives you your own drawing on the map.

The choice is also available in the settings of an existing dataset, under Connect data:

Linking to companies

Coming soon!

The same link for companies, so that organisation numbers in the file pull company data into the dataset. Until then, you link on address or matrikkelnummer, and get the companies from the Company panel.

Your own geometry in the file

Format	What a cell looks like
Latitude and longitude	Two numeric columns, 59.9139 and 10.7522
WKT	POINT(10.7522 59.9139), POLYGON((10.74 59.91,10.75 59.91,10.75 59.92,10.74 59.91))
GeoJSON	{"type":"Point","coordinates":[10.7522,59.9139]}

All coordinates are interpreted as WGS84, that is EPSG:4326, in decimal degrees. The upload does not reproject anything. If the data is in EUREF89 UTM zone 32 or 33 (EPSG:25832 and EPSG:25833), which is common for Norwegian map data, you must transform it to WGS84 before you upload it.

The drive time zones above are areas uploaded this way: one row per depot, with the zone as a WKT column in the spreadsheet. The zone answers the question behind every new site, how many households are reached within 30 minutes of the depot, and it does not exist in any register we can match against.

You point out the geometry columns yourself in the wizard. The suggestions come from the values in the cells, not from the column name, so a column called Y works just as well as one called breddegrad. Keep the decimals from the source: coordinates rounded to two decimals are not recognised as coordinates, and they are also several hundred metres off.

Matrikkel and address columns

In the step where you point out the address and matrikkel columns, you choose what each column contains. You must include either the postal code and postal town, or the municipality number.

Column type in the wizard	Content
Municipality code	Municipality number, four digits, for example 0301
Matrikkelnummer	The whole matrikkelnummer in one column
Registration number	Gårdsnummer
Subregistration number	Bruksnummer
Lease Number	Festenummer, 0 when the property is not leased
Story Number	Seksjonsnummer, 0 when the property is not divided into sections
Street, Number, Entrance, Postal code, City	The address split across separate columns

If you have both separate number columns and a matrikkelnummer column, the number columns are used. The matrikkelnummer column is interpreted loosely, so all of these spellings work:

0301 48/99/4
0301-48/99
gnr. 48, bnr. 99
Oslo gnr 48 snr 3

In the same way, addresses can sit in a single free-text column, Karl Johans gate 22, 0159 Oslo, instead of being split into street, number, postal code and postal town. Matching against matrikkel and address applies to Norway.

Excel pitfalls

• Leading zeros disappear. 0301 becomes 301 if the column is a number. Format the column as text before you paste, or leave the municipality number in the matrikkelnummer column.
• Decimal separators. Both 59,9139 and 59.9139 are read as numbers. A thousands separator makes the cell invalid as a number.
• Dates. 2026-09-03, 03.09.2026, 03/09/2026 and 2026/09/03 are recognised, with a time of HH:MM or HH:MM:SS if you include it.
• One sheet. Formulas that point to other sheets in the workbook are not included. Paste the values before you upload.

Converting from other formats

Shapefile, GeoPackage, GeoJSON, GML, KML and CSV must be converted to .xlsx with the geometry in a WKT column. With GDAL and LibreOffice this takes two steps: reproject to WGS84 and write the geometry as WKT, then convert to Excel:

# Hvilket koordinatsystem ligger filen i?
ogrinfo -so -al eiendommer.gpkg

# Reprojiser til WGS84 og skriv geometrien som en WKT-kolonne
ogr2ogr -f CSV eiendommer.csv eiendommer.gpkg \
  -t_srs EPSG:4326 -lco GEOMETRY=AS_WKT

# Punktdata: to koordinatkolonner i stedet for WKT. X er lengdegrad, Y er breddegrad
ogr2ogr -f CSV punkter.csv punkter.shp \
  -t_srs EPSG:4326 -lco GEOMETRY=AS_XY

# CSV til Excel
soffice --headless --convert-to xlsx eiendommer.csv

Without the command line, QGIS does the same: right-click the layer, choose Export, Save Features As, format Comma Separated Value, CRS EPSG:4326 and GEOMETRY set to AS_WKT. Open the CSV file in Excel or LibreOffice and save it as .xlsx.

For small jobs, geojson.io (draw or paste geometry, export GeoJSON) and mapshaper (simplify large polygons, convert between formats) are enough.

If you download from Geonorge, choose GeoJSON or GML rather than SOSI. SOSI has to be converted in a separate step, and all three formats contain the same thing.

You can drop columns you do not need before uploading. The row limit applies to rows, not columns, but a narrow dataset is faster to work with in the table.

When something fails

• The file is rejected at the file step if it is too large, has the wrong format, is corrupt, lacks usable sheets or has too many rows. The message says what was detected and what is supported.
• Errors at row and cell level are grouped in the list below the table: duplicate column names, missing or extra columns, values that do not fit the column type, duplicate keys and invalid geometry. Click an error to jump to the row or cell it applies to.
• Rows without a match stay in the dataset, but without a connection to a property or address. You can run the matching again from the settings. The most common causes are a missing municipality number, a truncated postal code or addresses that have since been renamed.

After upload, the dataset appears in the panel list with its name, row count and date.

If you want to skip the wizard, Claude can do the upload and the matrikkel matching for you through Placepoint Dataset MCP.

Showing the dataset on the map

Click the toggle next to the name to show the dataset on the map. The geometry is drawn as a separate layer over the background map, and the Legend at the bottom left shows what the colours mean.

Several datasets can be on at the same time, and they are drawn in the order they sit in the panel. Here the stops sit on top of the delivery zones, so a stop is read against the zone it belongs to. The areas come from the third dataset, the points from the first.

Hold the mouse pointer over a row and the toolbar of actions appears:

Icon	Action
Plus	Add geometry: draw a point, line or area straight into the dataset
Filter	Filter the rows in the dataset
Table	Open the dataset as a table
Palette	Style: colour, size, label, clustering and visibility
Cog	Dataset Settings: name, display values and connect data
Upload	Update the dataset with a new file
Toggle	Show or hide the dataset on the map

A single row in the panel

Click a point or an area on the map, and the row opens as a panel on the right with every column from the file editable. Move point moves the geometry on the map, Table opens the row in the table, and every change is saved with the time and user at the bottom of the panel.

Select several rows and the panel shows them together: shared values are edited once, values that differ are shown as mixed. That is how a whole street moves to another weekday in one operation.

The column set belongs to the dataset, so the workflow sits in the columns. The stops have Proposed stop and Proposed removal, each with its own note field: the driver reports it, the planner approves it in the panel or in the table, and Status is calculated from the other fields.

Over aerial imagery

The dataset is drawn over the background map you choose. With Images, the stops sit on the roof they actually belong to, and you see an address matched to the wrong place straight away.

Also where the distances are large

The map is the whole country, and the dataset comes with it. The four datasets cover 22 cities from Kristiansand to Kirkenes, and a delivery zone in Lofoten is read the same way as one in Trondheim.

Styling

The palette icon opens Style. The choices are saved on the dataset, so everyone in the project sees the same map layer.

Group	Option	What you control
Style	Global styling	One colour for the whole layer
Style	Categorical	One colour per unique value in a column
Style	Colormap	A scale over a numeric column
Style	Heatmap	Colour by point density, in 2D only
Size	Global size	Fixed point size
Size	Size range	Point size scaled by a numeric column
Label	Label	Text on the map from a column
Clustering	Clustering	Merge nearby points into clusters with a count
Visibility	Visibility	Whether the layer is always visible, regardless of zoom level

The stops in the example are styled Categorical on the Color column, which holds the colour of the route that serves the stop. Six values give six colours, and values without a match are collected in an Others group. This is what makes the map readable as a duty roster: two neighbouring houses in different colours are served by different vehicles.

Clustering is on, so nearby stops are merged into one circle with a count, until you zoom in. The legend shows both the colours and the cluster sizes.

If you need your own symbols, upload icons to the symbol library and use them instead of the default symbols. For numeric columns, the colormap shows a histogram of the values, so you can see where rows actually sit before you choose the breaks.

The table

The table icon opens the dataset at the bottom of the screen. You can drag the panel up to almost full height.

The table shows the rows as they came from your file, in the External tab. If the rows are linked to Placepoint, you get one more tab per match type: the addresses, the matrikkelenheter and the buildings the rows were matched to.

You can sort and move columns, Columns controls which ones are shown, and Show selected only shows only the rows you have selected on the map. The row count is at the bottom left, here 15,800. Click a row to open the details, where you can also change the values in your own columns: a new gate code, a stop that has to move to another weekday.

Filtering the rows

The filter icon filters on column values. You choose a column, an operator and a value, and you can add more filters with Add filter.

The column picker lists your own columns. For the planner, these are the questions asked most often: which stops are served on Tuesdays, which belong to one depot, which sit in one delivery zone. The filter is saved on the dataset, and the map layer and the table show the same selection, so the row count at the bottom of the table tells you how large the selection is.

Exporting

Export to Excel in the table downloads the rows. You choose between:

• The dataset as it looks now, with your column choices and filters.
• The full dataset as it was uploaded, with the original columns.

If you have selected rows on the map, only the selected rows are exported. That is how a driving list comes out: filter on depot and weekday, sort on sequence number, export the selection.

The dialogue shows how many rows each option gives. Files over 5,000 rows open slowly in Excel, and above 50,000 rows the export stops. How to split a large extract. The file is named after the dataset and today's date.

The columns are in the same order as in the table. The first option includes the columns as you have moved and hidden them there. The second gives all the columns in the order they had in the file you uploaded, with the row's id first. The order is the same every time you export, so a spreadsheet that reads columns by position can rely on it.

If you want a different fixed order, move the columns in the table before you export, or ask Claude to do it through Dataset MCP: "Move Postal code first in Stops". That order is saved on the dataset and applies in the table, in the details panel and in the export.

The rows can also be pulled into your own systems through the Placepoint API, so that a driving list lands in the app the drivers already use rather than in yet another system to log in to.

Drawing and editing geometry

The plus icon lets you draw a point, line or area straight into the dataset, without uploading a file. Existing geometry can be moved and changed in the same way. The rows you draw get the same columns as the rest of the dataset.

That is how a delivery zone is adjusted when a housing estate moves to another route: redraw the area, and the new stops belong to the new zone.

Updating the dataset with a new file

The upload icon starts the update wizard. It follows the same steps as a new upload: choose the file, choose the sheet, review the preview and finish. New rows are matched as in the first upload, and the old rows keep their connection.

Running matching again

The settings include Run matching, with two options:

• Only unmatched rows - try again on the rows that did not get a match.
• Full dataset - match all the rows again, for example after you have corrected an address column.

Every run is saved. View report shows the runs with the date, the number of rows from the dataset, matches per type, precision and how long the run took.

If you turn matching off in the settings, the matches you have are kept, but new rows are not connected.

Managing datasets

From the panel list you can see the row count and last updated date for each dataset, open a dataset to change the styling, rename it or delete it.

The cog icon on the row opens Dataset Settings:

Under General you set the name and description, and choose the display values: which columns identify a row on the map and in the panels. The stops use Name, which is the street address, so a click on the map answers with the address and not a row number.

A dataset can be locked. If it is set to read-only, the rows can be read and shown on the map, but not changed. If uploading is turned off, no one can upload a new file over it. This is useful for a dataset several people work against, or one that is kept up to date from Claude through Placepoint Dataset MCP.

Access

A dataset belongs to a project. Everyone with access to the project sees the datasets in it, with the same styling and filters. The settings show who has access to each dataset. The four datasets in the example are shared with the whole organisation through one project, so sales, operations and management see the same map.

Limits and rules per organisation

The default limit is 5,000 rows per upload, but it is not fixed for everyone. We can set a higher row limit on a single dataset when you have a larger list to work with. The stops in the example are one such dataset: 15,800 rows in one layer, with the table, filter and export as usual.

The same goes for the rules the upload uses: what should stop an upload, what should only be reported as a note, and how values are recognised as numbers, dates or geometry. The rules sit at four levels, from general to specific: the default setup, your organisation, the project and the individual dataset. The most specific level wins, and the rules are locked when the upload starts, so they do not change part-way through a job.

If you have a file format with its own conventions, for example dates written in a fixed way or code fields that should be read as text rather than numbers, your contact with us can have it set up for your organisation instead of each user having to correct the file first.

Your datasets in Placepoint AI

Placepoint AI is three MCP servers. Placepoint MCP gives Claude the registers and the map layers, Placepoint Dataset MCP gives Claude your projects and datasets, and Combify MCP gives Claude Swedish property. See Three servers, one conversation for what separates them.

Through Dataset MCP, Claude does the same things you do in the panel:

• Create a dataset and upload the file, with the column types set.
• Link the rows to an address or matrikkelnummer, run the matching again and read what it matched.
• Add, change and hide columns, and correct values on individual rows.
• Style the map layer, categorical on a column or as a colormap.
• Add municipality, grunnkrets (basic statistical unit) or postal code as columns, and sum or count rows per area.
• Calculate travel time from each point, and find the shortest route that visits them all.
• Export the selection.

For a setup like the delivery example, this is the difference between a spreadsheet job and a question: "Add grunnkrets as a column on Stops and count stops per krets", or "Which stops are in a flood zone?". The answer stays as a map layer in the project, not as a file on your desktop.

Claude only sees the projects your Placepoint user has access to. The Agent in Placepoint Fusion uses the same data from a panel in the platform.

See also

• Tags - an alternative for collecting and filtering individual properties without uploading a file.
• Map layers - thematic datasets from Placepoint and public registers.
• Placepoint AI - the same datasets from Claude.