Not a spec sheet — the story each setting is designed to tell: why it exists, when a business user would reach for it, and what you should expect to see when you flip it. Every figure below is a live simulation of the visual's behavior.
Most Power BI visuals answer “how much?” This one answers how is everything connected? It's for relationship data — who reports to whom, which accounts trade with which, what depends on what. Rows become nodes, relationships become links, and the shape reveals the hubs, the bridges that hold the network together, and the groups that cluster on their own. Below is every setting the visual has, grouped exactly like the in-visual gear, with what each one does.
The default: force layout, community colours, a highlighted hub and the bridges between clusters.Community hulls on — the clusters read at a glance.
The data you bind
Two fields draw a network; everything else adds meaning. Self-loops, parallel edges and disconnected groups are all first-class — the visual never errors on messy data.
Source node · required
The 'from' end of each relationship — a column of node identifiers. One row = one link.
Target node · required
The 'to' end of each relationship. Self-loops (source = target) and parallel edges are fully supported.
Edge weight
A numeric measure per link; drives link thickness and the 'relationship value' insight. Defaults to 1 when unbound. Also required for another visual to cross-highlight this one.
Node category
A category for each node; drives node colour and clustering. Target-only nodes take a neutral colour.
Node size
A numeric measure per node; drives 'Node value'. Unbound, node value falls back to total connected edge weight (or degree for unweighted data).
Node parent
A parent-node key. Bound, it unlocks the org-chart / tree layout and powers drill-down and expand/collapse.
Latitude / Longitude
Source-node coordinates. Bound together, they unlock the Geo-route layout. No external map tiles are fetched — the outline map is embedded.
Edge type
A category per relationship (e.g. flight / rail / ferry). Drives edge colour and line style, with a legend.
Time
A date or number per link for the time-animation axis — links appear over time.
Node image
A per-node base64 data: URI image, rendered in place of the marker. Only data: URIs render (external URLs are ignored, for certification).
Node icon
A semantic entity name per node (Person, Company, Server, Database…), shown as a monochrome vector in place of the shape. A bound image wins.
Tooltips
Business fields shown in the node hover card and detail panel; can replace or supplement the network metrics.
Layout
How the nodes are arranged. Choosing a mode clears a pinned layout so the new one can run.
Circular — every node on one tidy ring, crossings minimised.Concentric — the strongest hub in the centre, less-connected nodes on outer rings.
Grid — a stable square matrix, hubs first.Force (default) with community hulls — clusters pull apart on their own.
Mode
Force reveals clusters on its own; Concentric puts the strongest hub in the centre with lower-connectivity nodes on outer rings; Circular places every node on one ring in a crossing-minimising order; Grid arranges nodes on a stable square grid, hubs first; Tree uses a layered hierarchy (most-connected root when no Node parent is bound); Geo needs Latitude and Longitude.
Repulsion
How strongly nodes push apart in Force mode. No effect in Concentric or Tree.
Edge length
The preferred link distance in Force mode. No effect in Concentric or Tree.
Outline map
A faint embedded world map + lat/long grid behind the nodes, in Geo-route mode.
Pin layoutPro
Freezes and saves the current node positions so refreshes and reopenings reuse them. Dragging a node also pins the layout. Turn off — or switch mode — to recalculate.
Nodes
Diamond shapes with node borders — borders keep light fills visible.Hexagon nodes with the halo glow style.
Size
Size by
Degree (number of links), Node value (a measure, or total connected edge weight), Importance (the Analysis calculation), or Same size for a uniform radius.
Auto size to canvas
Fits node radii to the tile so nodes never swamp a small or dense view. Turn off to set an exact Min/Max.
Radius range
The exact min/max node radius in pixels (used when Auto size is off).
Size scale
Scales all node radii up or down by a percentage.
Avoid overlap
Nudges overlapping nodes apart after layout. Off for exact geo/XY positions.
Node gap
Extra edge-to-edge clearance between nodes without enlarging them; the layout shape is preserved and it uses the largest spacing that fits.
Merge duplicate nodes
On: one node per name holding both its incoming and outgoing links (folds case/whitespace variants). Off: split each name into its source and target vertices.
Style
Type
Halo draws a soft glow ring of the node's colour behind each marker; Flat is a plain marker.
Shape
Circle, Square, Diamond, Triangle, Hexagon or Donut. A semantic icon or bound image replaces the shape for that node.
Animate adjustments
Animates transitions such as zoom and layout changes — how updates move, not the final positions. Respects reduced-motion and is limited under Canvas rendering.
Initial growth animation
On first load and after Reset, nodes grow out from each component root and settle into their deterministic positions (~2.4s, reduced-motion aware).
Border
Draws an outline around every node — useful when node colours are light and would vanish against the background.
Border colour / width
The outline colour (blank = automatic contrast) and its thickness in pixels.
Patterns
Apply patternPro
All nodes uses one texture; By level (Pro) gives each hierarchy depth its own fill texture, using the Node-parent role or a derived tree.
Pattern
One of 12 fill textures — Solid, Dots, Circles, Diagonal, Hatch, Grid, Horizontal, Vertical, Checker, Diamonds, Zigzag, Waves. An icon or image replaces the patterned marker.
Per-level patterns (L0–L4+)
In By-level mode, a separate texture for the root and each hierarchy depth.
Icons
Apply iconPro
One icon for all nodes; By node type / By field value / By hierarchy level (all Pro) assign entity icons from the Node category, an Icon field, or the hierarchy depth.
Icon
The single entity icon that identifies what each node represents (a bound image still wins). Per-level icons available in By-level mode.
Image
Fit
Fill crops a bound node image to cover the circle; Fit shows the whole image inside it.
Border colour / width
An outline around image nodes (blank colour = match the node's own colour).
Parents
Emphasise parentsPro
Styles nodes that act as parents in a hierarchy. Bind a Node parent first, then enable the border, fill and size controls below.
Border / Fill / Parent size
Override the border, fill colour and radius of parent nodes only (blank fill keeps their normal colour; size in %).
On click
Click action
A node click does exactly one thing: Cross-filter only (standard report filtering), Show full info (Pro — opens the detail panel), Expand / collapse (Pro — folds a parent's descendants), or Drill-down (Pro — makes the clicked node the temporary root). The three Pro actions take over the click.
Start collapsed
With Expand / collapse, the graph opens with descendants folded to roots until the user expands them.
Colours
Colour by degree — a ramp from few links (light) to many (dark).Colour by hub — grey everything except the community connectors.
Mode
Colour by
Single (one colour), Category (a Node category field), Value (a numeric driver), Community (Pro — detected clusters), Component (Pro — disconnected subnetworks), or Level (the hierarchy). A custom gradient or a matching rule overrides this.
Colour driver
In Value mode, the number mapped along the palette — Node value, Degree, Value (edge weight), Importance, Bridges (betweenness) or Influence (PageRank).
Palette
The colour scheme. The list auto-filters: discrete modes show qualitative palettes; Value and Level-ramp show continuous scales.
Reverse colours / CVD-safe only
Flip the palette direction; or filter the list to schemes safe for common colour-vision deficiencies.
Node colour / opacity
The single (or fallback) node colour, and overall node opacity. Semantic icons inherit the resolved colour.
Gradient
Custom node gradientPro
Paint every node along your own Start → End ramp, positioned by the Colour driver. Overrides 'Colour by'.
Start / Midpoints / End
The gradient endpoints plus up to five intermediate colour stops for a multi-hue ramp.
Edges
Curved links with direction arrows.Source → target gradient links — each link fades between its endpoints' colours.
Show links
Draw the links. Off, you can still 'Show links on hover' so a node's connections appear only when you hover it.
Link colour
Automatic, a single colour, the source or target node's colour, or a source → target gradient — the endpoint modes are best for tracing dense links.
Arrows / Merge two-way links
Show direction arrows; and combine A→B and B→A into one double-headed link instead of two parallel ones.
Width / Scale width by weight
The width of the heaviest link; with scaling on, each link's width maps to its edge weight (lightest at 20% of Width).
Curvature
How strongly links bow. Tree defaults to maximum curvature, but any value you set wins — including 0 for straight links.
Link labels / Link label
Print a value on each link: Weight, Edge type, Weight % of total, Edge betweenness (≤ 2,000 nodes), or Source → target names.
Animate flowPro
Animate directional flow pulses along the links, with an adjustable speed.
Labels
Outer (name labels)
Show outer label / Content
Print a label beside each node: Name, Value, Name (value), or Inflow / Outflow / Total flow (Pro — directed edge-weight totals).
Display units / Decimals
Format numeric labels (Auto K/M/B, full number, or a fixed unit) and their decimal precision.
Max labels
Cap how many labels are drawn to keep dense graphs readable; when capped, the most-connected nodes win. 0 hides all.
Position / Wrap
Where the label sits (Auto/Right/Left/Top/Bottom) and whether long labels wrap.
Font / Size / Style / Colour
Standard type controls for the label.
BackgroundPro
A backing behind the label — Card, Highlight or Pill — with its own colour and padding.
Inner (in-node value)
Show inner labelPro
Print a value inside each node (SVG rendering only): Node value, Degree, Value or Importance, with its own units, font and colour. Text shrinks to fit the node.
Filter
Show
Reduce the graph to the Top N or Bottom N nodes (or All).
Action
Filter re-lays out just the ranked N; Highlight (Pro) keeps the whole network, accents the ranked N and dims the rest.
Rank by / N
Rank by Degree, Value (edge weight), Node value or Importance; N sets how many nodes to keep.
Highlight response
When another visual highlights this graph, the Table & Insight either Collapse to the selection ('showing N of M') or Keep full network (Enterprise) — the graph always dims the non-selected nodes.
Rules (conditional formatting)
Rules editorPro
Colour nodes by any number of rules — numeric (degree, centrality…) or text (name, category). Enabled rules keep colouring even when the editor panel is hidden, and override the base Colours mode where their conditions match.
Business fields (from the Tooltips well), Business + network (Pro), or Network metrics (Pro) — choose whether your fields replace or supplement the metrics.
View switch (Graph / Table / Insight)
Summary table
Adds a Table segment to the floating view switch — the node-metrics grid (degree, in/out, weighted, and opt-in centrality columns).
InsightsPro
Adds an Insight segment — a plain-English read-out of hubs, connectivity, bridges, communities and density.
Position
Which corner the floating Graph / Table / Insight switch sits in.
Legend
Category legend / Edge legend
Show the node-colour legend (categories, communities, components, levels or a numeric scale) and the edge legend (edge type / weight).
Position
Which corner the legend sits in (Auto avoids other overlays).
Enterprise
Scale and heavy-analysis features. Every control here needs the Enterprise tier.
Clusters
Show clustersEnterprise
Turn on cluster detection and the hull/label/grouping controls below.
Cluster by
Auto (detected communities), Category field, or Connected components.
Granularity / Min cluster size / Max clusters
Tune community detection: higher granularity = more, smaller clusters; fold clusters below a size, or keep only the N largest, into an 'Other' group.
Collapse to meta-nodes
Replace each cluster with one aggregate node, so the graph shows cluster-to-cluster relationships instead of every member.
Hull appearance
Draw a boundary around each cluster — Rounded or Convex, with padding, fill opacity, border width/opacity, and a per-cluster or single-tint colour.
Cluster labels
Name each cluster (optionally with member counts), with full font/size/style controls and Auto (match cluster) or Manual colour.
Group by cluster / Grouping strength
In Force layout, pull same-cluster nodes together so hulls stop overlapping.
Click hull to filter / Hover to emphasise
Click a hull to cross-filter its cluster; hover a hull or label to dim everything outside it.
Scale
RendererEnterprise
Auto uses SVG below a threshold and Canvas above it; SVG keeps the richest interaction, Canvas is faster for large graphs. Time animation needs SVG.
Canvas above (nodes)
The node count at which Auto switches from SVG to Canvas.
Max edges / Load beyond 30k rowsEnterprise
Retain every relationship by default, or set a performance cap; and request additional Power BI data windows past 30k rows.
MinimapEnterprise
A small overview map for navigating large graphs.
Analysis
ImportancePro
Compute a centrality score for size/colour/labels/rules/filtering: Degree, Bridges (betweenness), Reach (closeness) or Influence (PageRank).
Search box
A search control to jump straight to a node by name.
Explore modeEnterprise
Focus a selected node and its neighbourhood, expanding hop by hop. Changes the visible subgraph, not the data.
Path analysisEnterprise
Pick two nodes and highlight the shortest path between them — optionally weighted by edge values.
Time animationEnterprise
A time slider that reveals nodes and links across a bound Time field (SVG rendering only).
Show annotationsEnterprise
Alt/Option-click a node to add a note; annotated nodes get a badge. Choose the annotation style (Marker / Text / Text + arrow / All).
Deterministic & private by design
Determinism: positions come from a hash of your data, never Math.random — the graph never “moves” between refreshes, and a pinned layout is saved with the report. Privacy: every insight is computed inside the Power BI sandbox with no external calls and no telemetry — nothing leaves your tenant. That's also what makes it certifiable by construction.
Free, Pro, Enterprise
The free tier is genuinely useful — core layouts, colouring, labels, cross-filter, search, and a graph up to 500 nodes, with no watermark. Pro adds analytics and production polish (centrality, insights, rules, drill, flow, styling — the Pro settings above). Enterprise adds scale and heavy analysis (clustering + hulls, explore, path, temporal, geo, annotations, uncapped size — the Enterprise settings). Report viewers are always free — anyone reading a published report sees the full interactive graph.
The fastest way to understand these is to touch them — the core ones are live in the playground.