How to Make a Decision Tree: Step-by-Step Creation Guide

Knowing what a decision tree is and actually building one that looks professional are two different skills. Most guides cover the theory -- node types, expected value formulas, fold-back logic -- and then hand you a blank PowerPoint slide. The gap between "I understand how decision trees work" and "I can build one that a steering committee will take seriously" is where most people get stuck.

This guide closes that gap. It covers the hands-on construction of decision trees across four tools -- PowerPoint, Excel, Google Sheets, and specialized software -- using a single worked scenario so you can compare outputs directly. For the analytical methodology behind decision trees, see our Decision Tree Examples and Decision Tree Analysis guides. For the broader frameworks toolkit, start with the Strategic Frameworks Guide.

The Worked Scenario: Build vs. Partner#

Every construction example in this guide uses the same decision so you can see how each tool handles it differently.

Scenario: A mid-market fintech company must decide whether to build a payments integration in-house or partner with an established processor. The build option costs $1.8M upfront with a 12-month timeline. The partner option costs $600K with a 4-month timeline.

Tree structure (4 nodes, 2 stages):

• Root decision node: Build In-House vs. Partner
• Build path leads to a chance node: On-Time Delivery (P = 0.55, payoff = +$4.1M) vs. Delayed 6 Months (P = 0.45, payoff = +$0.9M)
• Partner path leads to a chance node: Favorable Terms Renewed (P = 0.70, payoff = +$2.6M) vs. Terms Worsen at Renewal (P = 0.30, payoff = +$1.1M)

Expected values: Build EV = (0.55 x $4.1M) + (0.45 x $0.9M) = $2.66M. Partner EV = (0.70 x $2.6M) + (0.30 x $1.1M) = $2.15M. Recommendation: Build, with sensitivity on the on-time delivery probability.

We will now build this exact tree in each tool.

How to Make a Decision Tree in PowerPoint: Full Walkthrough#

PowerPoint is the standard delivery format for steering committees. The challenge is getting precise, professional-looking trees without spending an hour on alignment.

Slide Setup#

Start with a blank slide layout -- remove all title and content placeholders. On a standard 13.333 x 7.5 inch widescreen slide, leave a 0.8-inch top margin for the title and 0.4-inch bottom margin for the source line. The tree occupies the middle 5.3 inches of vertical space.

Node Dimensions and Placement#

Use these exact dimensions for consulting-grade output:

• Decision nodes (squares): 1.2 x 0.7 inches, rounded corners at 0.05 inches
• Chance nodes (circles): 0.65 inch diameter
• Terminal nodes (triangles): 0.5 x 0.5 inches, pointing right
• Branch lines: 1.5pt weight for standard branches, 3pt weight for the recommended path

Place the root decision node at x = 0.6 inches from the left edge, vertically centered at y = 3.75 inches. Space subsequent node columns at 3.2-inch horizontal intervals. This puts Stage 1 chance nodes at x = 3.8 inches and terminal nodes at x = 7.0 inches -- leaving room for payoff labels on the right.

Step-by-Step Construction#

1. Place the root decision node. Go to Insert, then Shapes, and select a rounded rectangle. Draw it at the specified dimensions. Type "Build vs. Partner" inside. Set the font to 9pt Calibri Bold, centered. Fill the node with a muted blue (#4472C4) and set text color to white.

2. Draw two branches from the root. Use Insert, then Shapes, then Lines, and select Elbow Connector. Draw one branch angling 25 degrees upward to the "Build" chance node position and one angling 25 degrees downward to the "Partner" chance node position. Consistent branch angles are critical -- use PowerPoint's guides (View, then Guides) set at y = 1.9 inches and y = 5.6 inches for the two branch endpoints.

3. Add branch labels. Place a text box on each branch line. The upper branch reads "Build In-House ($1.8M)" and the lower reads "Partner ($600K)." Use 8pt Calibri, no fill, no border. Position labels 0.2 inches above the branch line, centered horizontally between the two nodes they connect.

4. Place the chance nodes. Draw two circles at 0.65-inch diameter. Fill them with light orange (#ED7D31). Position them at the endpoints of your branch lines: approximately x = 3.8 inches, y = 1.9 inches (Build) and x = 3.8 inches, y = 5.6 inches (Partner).

5. Draw second-stage branches and terminal nodes. From each chance node, draw two more elbow connectors -- one angling slightly up, one slightly down. Place terminal triangles (0.5 x 0.5 inches, filled with green #70AD47 for positive payoffs) at x = 7.0 inches. Add probability labels on each branch ("P = 0.55", "P = 0.45") and payoff labels next to each terminal node ("+$4.1M", "+$0.9M", "+$2.6M", "+$1.1M"). Use 8pt for probabilities and 10pt Bold for payoffs.

6. Highlight the recommended path. Select the "Build In-House" branch and the "On-Time Delivery" branch. Change their line weight to 3pt and color to dark blue (#2F5496). Add a callout box near the root node reading "Recommended: Build (EV = $2.66M)" in 10pt Bold.

7. Group and finalize. Select all shapes with Ctrl+A, then Group (right-click, then Group). This locks connector attachment points so nothing shifts during presentation mode. Add a slide title at the top: "Build vs. Partner: Decision Tree Analysis" in 18pt.

For a pre-formatted starting point that handles node sizing and connector routing automatically, the decision tree template includes placeholders for a 2-stage tree. Deckary's alignment shortcuts also help keep nodes evenly spaced when building manually.

How to Make a Decision Tree in Excel: Formula-Driven Approach#

Excel is the right tool for the analysis phase -- running scenarios, adjusting probabilities, and stress-testing assumptions before you build the presentation version.

Spreadsheet Layout#

Structure the tree using columns with spacer columns between stages:

• Column A: Root decision labels ("Build In-House", "Partner")
• Column B: Empty spacer for clarity
• Column C: Probability values (0.55, 0.45, 0.70, 0.30)
• Column D: Outcome descriptions ("On-Time Delivery", "Delayed 6 Months")
• Column E: Payoff values ($4,100,000, $900,000, $2,600,000, $1,100,000)
• Column F: Weighted payoff formulas
• Column G: Branch EVs

Cell References and Formulas#

Set up the Build In-House branch in rows 3-4 and Partner in rows 7-8:

Cell F3 (weighted payoff, Build/On-Time): =C3*E3 returns $2,255,000

Cell F4 (weighted payoff, Build/Delayed): =C4*E4 returns $405,000

Cell G3 (Build EV): =SUM(F3:F4) returns $2,660,000

Cell F7 (weighted payoff, Partner/Favorable): =C7*E7 returns $1,820,000

Cell F8 (weighted payoff, Partner/Worsen): =C8*E8 returns $330,000

Cell G7 (Partner EV): =SUM(F7:F8) returns $2,150,000

Cell G11 (Recommendation): =IF(G3>G7,"Build In-House","Partner") returns "Build In-House"

Adding Sensitivity Analysis#

The real power of the Excel approach is scenario testing. Add a sensitivity table in columns I through K:

Row 1 headers: "On-Time Probability" (I1), "Build EV" (J1), "Partner EV" (K1)

In column I, list probabilities from 0.30 to 0.80 in 0.05 increments. In column J, use =I2*$E$3+(1-I2)*$E$4 to recalculate the Build EV at each probability. Column K remains constant at $2,150,000 since the partner path is unaffected.

The crossover point -- where Build EV equals Partner EV -- is the critical threshold. With these payoffs, the Build path drops below Partner when the on-time probability falls below 0.42. This single number is often the most valuable output of the entire exercise.

Apply conditional formatting to highlight cells in column J where the Build EV drops below the Partner EV. This visual flag makes the sensitivity threshold immediately obvious.

How to Make a Decision Tree in Google Sheets#

Google Sheets uses the same formula layout as Excel with one key advantage: real-time collaboration. Create the tree in a shared sheet, assign probability cells to different team members for independent estimation (use protected ranges so each person edits only their assigned cells), then reveal all estimates simultaneously to avoid anchoring bias.

One useful addition: the SPARKLINE function creates inline mini-charts. Add =SPARKLINE(J2:J12, {"charttype","bar";"max",3000000}) in the sensitivity table to give stakeholders a visual scan of how the Build EV changes across probability assumptions.

Specialized Software: When Tools Matter#

For trees with more than two stages or where Monte Carlo simulation is needed, specialized tools justify their cost.

When to upgrade from PowerPoint and Excel: If your tree exceeds three stages or you need continuous probability distributions rather than discrete branches, specialized software saves time. For most strategy work with two or three stages, the Excel-for-analysis and PowerPoint-for-presentation workflow remains most efficient. Tools like Deckary provide pre-built templates that handle the formatting layer.

From Napkin to Boardroom: One Scenario in Three Stages#

Here is how the Build vs. Partner scenario evolves from rough thinking to polished deliverable.

Stage 1: The Napkin Sketch (5 minutes)#

A strategy lead draws two branches on a whiteboard: "Build" and "Partner." Under each, rough outcomes and gut-feel probabilities. No formatting, no precise numbers -- just the logical skeleton. The value at this stage is alignment: does the team agree these are the two options? Someone asks "what about a hybrid approach?" -- and that structural gap gets surfaced before anyone spends time on precision.

Stage 2: The Excel Model (2 hours)#

The analyst builds the napkin sketch into the spreadsheet structure described above. Probabilities get anchored to base rates: the 0.55 on-time delivery estimate comes from the company's last four integration projects (three on time, one delayed). The sensitivity table reveals a crossover threshold of 0.42. The team runs three scenarios -- optimistic (0.65), base case (0.55), pessimistic (0.40) -- and documents the reasoning behind each.

Stage 3: The PowerPoint Deliverable (45-60 minutes)#

The final tree uses the exact dimensions, color coding, and annotation standards from the PowerPoint walkthrough above. The recommended path is highlighted. A callout box shows the EV. A footnote cites the source of probability estimates. The audience traces the logic in under 30 seconds. The sensitivity threshold appears in the recommendation text: "Build is optimal unless on-time delivery probability falls below 42%."

Total time from napkin to boardroom: roughly 4 hours. PowerPoint construction drops to 15-20 minutes when starting from a template.

Consulting-Grade Formatting Standards#

These specifications distinguish a professional deliverable from a first draft.

Color System#

Assign colors by node function, not by outcome:

• Decision nodes: Muted blue (#4472C4) with white text -- signals a choice the audience controls
• Chance nodes: Light orange (#ED7D31) with white text -- signals uncertainty outside the audience's control
• Terminal nodes, positive payoff: Green (#70AD47) with white text
• Terminal nodes, negative payoff: Red (#C00000) with white text
• Terminal nodes, break-even: Gray (#808080) with white text
• Recommended path branches: Dark blue (#2F5496) at 3pt weight
• Non-recommended branches: Medium gray (#A6A6A6) at 1.5pt weight

This system makes the tree scannable -- an executive can identify every decision point, every uncertainty, and the recommended path without reading a single label.

Spacing Ratios#

Maintain consistent proportions regardless of tree size:

• Horizontal gap between node columns: 2.5 to 3.5 inches, consistent across all stages
• Vertical gap between sibling branches: At minimum 1.2 inches center-to-center, to prevent label overlap
• Branch label offset: 0.2 inches above the branch line, horizontally centered between the two connected nodes
• Payoff label placement: 0.3 inches to the right of terminal nodes, right-aligned so dollar signs stack vertically
• Probability label placement: Directly on the branch line, positioned at the midpoint between the chance node and the terminal node

Layout Rules by Tree Size#

2-stage trees (4-6 terminal nodes): Single slide, landscape. Root at x = 0.6 inches, terminals ending by x = 10 inches, leaving 3 inches for payoff labels and EV annotations.

3-stage trees (8-12 terminal nodes): Tight on a single slide. Use 8pt labels, reduce horizontal spacing to 2.5 inches between columns. If labels overlap, split across two slides -- a structure overview and a calculation detail slide.

Trees with more than 12 terminal nodes: Summarize the top-level decision on one slide with branch EVs as single numbers, then create detail slides for each major branch showing full sub-trees.

Annotation Placement#

Place the EV callout box adjacent to the root decision node. Add a source line in the bottom-left corner (7pt gray text, citing probability sources). List the one or two most critical assumptions in the bottom-right corner at 7pt.

Common Construction Mistakes#

Curved connectors. They look elegant but create ambiguity when branches cross. Always use straight or elbow connectors -- less visually interesting, but unambiguous.

Inconsistent node sizes. Varying dimensions implies varying importance. Copy and paste one template node rather than drawing each individually to ensure every node of the same type is identical.

Labels inside nodes instead of on branches. Probabilities belong on branch lines, payoffs belong next to terminal nodes. Overloading nodes with text forces tiny fonts and reduces scannability.

Missing the "do nothing" branch. Every tree should include the status quo path, even if its payoff is zero. Without it, the audience cannot tell whether the recommendation is genuinely better than inaction.

No grouping before presenting. Ungrouped shapes shift in Slide Show mode and connectors detach from nodes. Always select all objects and group them as the final step.

Key Takeaways#

Building a decision tree is a three-phase process: sketch the structure, model the math, then format the deliverable. Each phase uses a different tool optimized for that task.

• PowerPoint is for presentation, not analysis -- use exact node dimensions (1.2 x 0.7 inch decision nodes, 0.65 inch chance nodes) and consistent 3.2-inch column spacing for professional results
• Excel is for stress-testing -- build the sensitivity table first, find the probability threshold that flips the recommendation, then document the reasoning
• Color-code by node function -- blue for decisions you control, orange for uncertainty, green and red for outcomes -- so the tree is scannable without reading labels
• Keep trees to 2-3 stages per slide -- anything beyond 12 terminal nodes needs to be split across multiple slides
• The napkin-to-boardroom progression takes roughly 4 hours -- most of that time goes into the Excel model, not the PowerPoint formatting

For worked examples with full probability calculations, see decision tree examples. To understand the analytical methodology in depth, see the decision tree analysis guide. To compare decision trees with other evaluation frameworks, see the decision matrix guide. For a ready-to-use PowerPoint layout, start with the decision tree template.