Ataytis Assurance documentation.

The visual reads a table, not a drawing. Whether your items live in a Visio file, a slide, a spreadsheet, a database or another tool, they have to become rows in Power BI first. You do not need to know any of the analysis methods to do this. Every table needs three things:

• An ID for each item. Anything unique to it: 1, 2, 3, a code such as LIM-004, or a name. The visual calls this the Node ID, and each item a node.
• The ID of the item it connects to. In a hierarchy that is the item it belongs to, the one above it, which is what a Parent ID column holds. The visual calls this the Connected node. Leave it blank for an item at the top, or one that is not connected to anything.
• A name to show. The text drawn on each item, which the visual calls the Label.

Everything else is optional. A limit, the latest value, a status, an owner, a date or a link goes in a column of its own, and each can be shown when the item is selected, written beside it or used to colour it. A Node type column, which says what kind of item each row is, matters only when you want the symbols and rules of a particular method: Fault Tree Analysis, Reliability Block Diagram and Markov State Diagram all draw without one.

The diagram type is chosen in the editor, not in the data. If most of your node types belong to a different type, the Validation tab says which one.

A risk team keeps its limits in a table. Each limit has a number, the number of the limit it rolls up into, a threshold, the latest reported value and whether it is breached. It is not a fault tree or any other method, but it is a hierarchy, and a hierarchy is all the visual needs.

Limit 9 rolls up into nothing, so it is drawn on its own beside the tree.

1. Put the columns in the visual

   Drag Limit ID, Parent Limit ID, Limit Name, Owner and Breach Status into Data fields. Drag Limit and Latest Value into Values and set each to Don't summarize.

2. Choose Fault Tree Analysis

   Open the editor with More options (…) → Edit and select Fault Tree Analysis on the left. It draws a hierarchy top to bottom. With no Node Type column every item is a plain box and only loops and out-of-range values are checked, so no fault tree rules get in the way, and the editor reads Validation clear.

3. Map three columns

   On the Data Mapping tab, set Node ID to Limit ID, Connected node to Parent Limit ID and Label to Limit Name. Columns called ID, Parent ID and Label would map themselves.

4. Colour the breaches

   Under Appearance → Conditional formatting, choose Add rule: When Breach Status, Is Equal to, Value Breached, Colour Node fill, and a red. Turn on Labels → Legend → Show legend so readers know what the colour means.

5. Show the numbers

   Under Labels → Value annotations, add rows for Limit and Latest Value to write them beside each box. Select a limit on the report and the Node Inspector lists the rest of its row, such as its owner.

The same steps work for any list of items with an ID and the ID of the item they belong to: KPIs rolling up into a scorecard, cost centres, controls grouped under objectives, or parts in a product structure. If you keep every reported value over time, put the reporting date on a slicer so the diagram shows one date at a time.

A diagram that exists only as a picture, such as a Visio or PowerPoint drawing, is typed out as a table once. Keep the drawing open beside Excel, or beside Enter data in Power BI, while you work.

1. Give every box a Node ID

   Short and unique, such as 1, 2 and 3, or TOP, G1 and P1. Writing the IDs onto a printout or into the shapes makes the rest easier.

2. Write one row for each box

   Its Node ID and its text, in columns called Node ID and Label, and its node type in a Node Type column when the diagram follows one of the methods.

3. Add the lines

   In a Connected Node ID column, write the Node ID at the other end of each line, on the row of the box the line belongs to. A box with two lines needs two rows: copy its row and change Connected Node ID. A box with no node to name, such as the top of a fault tree or the output of a reliability block diagram, keeps its row with Connected Node ID blank.

4. Put each line on the end the diagram type expects

   A fault tree or event tree row names the node before it, such as a basic event naming the gate above it. A bow-tie, reliability block diagram, goal structure or Markov row names the node after it. If your table is the other way round, change Reads as on the Data Mapping tab instead of retyping it.

5. Add your values

   Probability, status, owner and any other detail, in columns of their own. A node's values can go on one of its rows or on all of them, as long as its rows do not give different values.

A fault tree with a top event TOP over an OR gate G1, and two basic events under G1, is the four rows in Build your first diagram.

Most analysis and diagramming tools can export to Excel or CSV. Look at what one row of the export describes. If it is not already one row per connection, reshape it in Power Query, which opens when you select Transform Data while loading it.

A state that nothing leaves, such as an absorbing state, has no row left after a matrix is unpivoted. Append a list of the states, so every state keeps a row of its own.

• Trim the ID columns with Transform → Format → Trim. An ID with a stray space does not join.
• Where a spreadsheet writes a value once and leaves the cells below it empty, as merged cells do, select the column and choose Transform → Fill → Down.
• Give number columns a number type, such as Decimal Number. A number stored as text is not bound to a numeric role.
• Store percentages as fractions, 0.96 rather than 96.
• Keep the node type wording your tool uses. Wording the visual does not recognise is mapped once under Node type values on the Data Mapping tab.

1. Get the data

   In Power BI Desktop, select Home → Get data and choose where your data lives. Any source Power BI connects to works, including Excel, Text/CSV, SharePoint lists, SQL Server and other databases, Dataverse and web sources: the visual only sees the finished table, so what matters is its shape, not where it comes from. A small table can be typed straight in with Home → Enter data.

2. Reshape it if it needs it

   In the Navigator, tick the table and select Transform Data instead of Load. Make the changes above, then select Close & Apply.

3. Add the visual and fill the wells

   Put IDs, types, labels and other text in Data fields. Put numbers in Values and set each one to Don't summarize.

4. Choose the diagram type and check the mapping

   Select the visual, open More options (…) and choose Edit. Pick the diagram type on the left, check that each column is mapped to the right role on the Data Mapping tab, and read the Validation tab.

When the source changes, refresh the report and the diagram is rebuilt from the new rows. As long as the column names stay the same, nothing needs mapping again.

Each diagram type's Preparing data page lists every column it reads, which row each value goes on, and a complete worked example.