Executive Dashboard Design

Executive summary

Executive dashboard design is the work of deciding which numbers a leadership team will look at every week, defining each one so nobody argues about it, and building a page that makes the next decision obvious. The charting is the smallest part. The hard parts are agreeing on definitions, assigning an owner to every number, and deciding in advance what happens when a number moves.

A dashboard that works answers three questions in under two minutes: are we on track, where are we off, and who is doing something about it. A dashboard that fails shows 40 tiles, three definitions of "revenue," and no owner. Teams stop opening it, and the real review happens in a spreadsheet someone built privately.

This guide gives a method in five steps: choose the decisions the dashboard supports, select about twelve metrics, write a spec for each, set thresholds by method, and lay out the page for a fixed review. It pairs with KPI design: choosing 12 metrics that run the company, which covers what to measure, and with the operating cadence, which covers the meetings the dashboard feeds.

When to use this

Use this guide when:

• Leadership meetings spend the first half hour reconciling numbers instead of discussing them.
• Different functions report different figures for the same metric.
• The board or a lender asks for a view you currently assemble by hand each month.
• You are starting a new role or a new ownership period and want one page that shows the state of the business.
• A dashboard exists but nobody can say what decision it has ever changed.

Do not start with a tool. If the ledger does not reconcile or the operational data is unreliable, a good-looking dashboard will display unreliable numbers more convincingly. Fix the data foundations of the few metrics you need first, and then build the view.

Diagnostic questions

Answer these about your current reporting before you design anything.

1. Which three decisions did leadership make in the last quarter, and which numbers informed them?
2. Which metric has more than one definition in use today, and who uses each?
3. How many hours per week does someone spend assembling the numbers manually?
4. Which numbers are owned by a named person who can explain a movement within a day?
5. How old is the data when leadership sees it: hours, days, or weeks?
6. What threshold, if crossed, would change what the CEO does this week?
7. Which metrics have a target, and who set it and on what basis?
8. Which metrics tell you what already happened versus what will happen next?
9. What does the board or lender see that the executive team does not, and the reverse?
10. Which tile has been on the page for six months without anyone asking about it?
11. If a number is wrong, who finds out first: the owner, the CFO, or a customer?
12. Can someone outside the finance team read the page and say in a minute whether the business is healthy?

Read the pattern. Several "no" answers to questions 2, 4, and 11 mean the problem is definitions and ownership, not design. A long answer to question 3 means the problem is data plumbing. A weak answer to question 1 means the dashboard has no purpose yet, and the first step is to name the decisions.

The framework: decision, metric, owner, threshold, action

Treat each tile as a five-part contract.

A tile without all five is decoration. Drop it from the page or finish the contract.

The three layers of a good executive view

Outcomes show the result and arrive late. Drivers explain it. Early signals give you time to act. A dashboard of only outcomes tells you about problems after you can fix them.

Step-by-step process

1. Name the decisions (week 1). Owner: CEO. Output: a list of five to eight recurring decisions the dashboard must support, such as cash allocation, pricing action, hiring, and capacity.
2. Draft the metric list (week 1). Owner: CEO with CFO. Output: no more than twelve candidate metrics, each linked to a decision from step 1. If a metric supports no decision, cut it.
3. Write a metric spec for each (weeks 1-2). Owner: the metric owner with FP&A. Output: a completed spec (template below) with formula, source system, refresh timing, and owner.
4. Reconcile the definitions (weeks 2-3). Owner: CFO. Output: one agreed definition per metric, with disagreements resolved in writing. Re-state history using the agreed definition so the trend is comparable.
5. Set thresholds (week 3). Owner: metric owner and CEO. Output: a watch and an act threshold per metric, each tied to a method (see Metrics and KPIs).
6. Lay out the page (week 3). Owner: FP&A. Output: a one-page view with outcomes at the top, drivers in the middle, early signals at the bottom, trend lines with the last 12 periods, and a short owner comment field.
7. Build the data pipeline (weeks 3-6). Owner: finance systems or IT. Output: automated refresh from source systems for the metrics where it is feasible, with a documented manual process for the rest and a visible "last refreshed" stamp.
8. Run two parallel cycles (weeks 5-8). Owner: FP&A. Output: the dashboard figures compared to the existing manual pack for two periods, with differences explained and fixed.
9. Hold the weekly review (from week 7). Owner: CEO. Output: a 30-minute review where every red or amber tile has an owner explanation and a decision or a dated action.
10. Audit quarterly (every quarter). Owner: CFO. Output: a review of which tiles led to a decision, which did not, and what to retire or add.

Templates

Metric spec (copy into a spreadsheet)

Page layout

Avoid decorative gauges and 3D charts. Show the value, the target, the direction, and the status in a form that can be read at a glance and printed in black and white.

Weekly dashboard review agenda (30 minutes)

1. Headline: are we on track for the month, and what changed since last week (5 min).
2. Red tiles: owner explains the cause, the action, and a date (10 min).
3. Amber tiles: owner states whether it will recover without intervention (5 min).
4. Early signals: anything arriving in the next four weeks that needs a decision now (5 min).
5. Data issues: any number that looks wrong and who is fixing it (5 min).

Status rule table

Worked example (illustrative)

The numbers below are invented to show the method. They are not benchmarks.

A distributor builds a one-page dashboard for the March review. Three tiles are shown with the arithmetic.

Gross margin. Revenue was $4,800,000 and cost of goods sold was $3,300,000.

Gross margin = (4,800,000 − 3,300,000) ÷ 4,800,000 = 1,500,000 ÷ 4,800,000 = 31.25%.

The target is 32.0%. The gap is 0.75 points. The watch band is within 1.0 point of target, so the tile is Watch. The owner (head of pricing) comments that two customers received a one-time rebate and attaches the amounts.

On-time delivery. 1,860 order lines shipped on or before the promised date, out of 2,000 due.

OTD = 1,860 ÷ 2,000 = 93.0%. The target is 95.0%, a gap of 2.0 points. The watch band is within 3.0 points, so the tile is Watch. The method behind the target: the company's own best trailing six months was 94.6%, and the target was set at 95.0% after the operations lead confirmed the constraint was material shortages, which procurement is addressing.

Cash. The 13-week forecast shows a low point of $1.2M in week 9. The minimum liquidity floor, set by the revolver agreement and management policy, is $1.5M.

Headroom = 1.2M − 1.5M = −$0.3M. The tile is Act: the forecast breaches the floor. The CFO brings a decision to the review: slow a planned inventory purchase or accelerate the collection of two named past-due accounts.

Cost of building the view manually. If twelve metrics each take 20 minutes to pull, check, and paste into a pack, that is 12 × 20 = 240 minutes, or 4.0 hours per week, which is 4.0 × 52 = 208 hours per year of skilled finance time. Automating the feeds for the eight metrics sourced from one ERP removes 8 × 20 = 160 minutes per week, leaving 80 minutes (about 1.3 hours) for the four metrics that still need manual work and a check.

The review output is three decisions with owners and dates, not twelve descriptions of numbers.

Metrics and KPIs

The design itself needs measures, so you know whether the dashboard is earning its place.

Setting thresholds by method. Do not take a threshold from a blog. Use these anchors in order: a contractual floor (covenants, customer service agreements), your own best trailing six or twelve months, and then named public peers where a comparable listed company publishes the metric in its filings. Set the watch band by how much movement you can absorb before a decision would change, and the act band at the point where inaction costs more than intervention. Write the basis next to each threshold so the next person can challenge it.

Retire rule. If a tile has not changed a decision in two consecutive quarters and no lender or board requirement keeps it, remove it. A page that only grows becomes unread.

Common mistakes

1. Starting with the tool. Choosing a software package before agreeing the metrics produces a prettier version of the same argument.
2. Too many tiles. Past about twelve, the reader scans instead of deciding. Put the rest in a drill-down.
3. Showing outcomes only. Revenue and EBITDA report the past. Without drivers and early signals the page cannot prompt an action.
4. No owner. If nobody can explain a number within a day, it will be argued over rather than acted on.
5. Hiding the data age. A beautifully formatted view of three-week-old data misleads. Show the last refreshed time on the page.
6. Mixing definitions across periods. If a definition changes, restate the history. Otherwise a change in definition looks like a change in performance.
7. Thresholds with no basis. A red status that nobody believes teaches people to ignore red.
8. Colour as the only signal. Use labels or symbols as well, so the page works in print and for colour-blind readers.
9. Treating the dashboard as a report card. If owners fear the page, they manage the number rather than the business. Use it to ask what help the owner needs.
10. Never retiring anything. The page that only adds tiles ends up as the old 40-tile pack in a new format.

AI prompt assistance

AI is useful for drafting specs and testing definitions. It does not know your data, so verify every formula against the ledger and the source system.

1. Draft metric specs.

I will give you a list of twelve KPIs for a mid-market distributor. For each, write a spec with: the decision it supports, an exact formula with numerator and denominator, likely source system, refresh frequency, and two common definition disputes to resolve. Do not suggest target values.

Check: confirm each formula against how your ERP actually records the fields, and have each owner accept the spec.

2. Find definition conflicts.

Below are three reports that each show "revenue" or "gross margin" for the same month, with the numbers. List every reason the figures might differ, such as timing, rebates, freight, intercompany, and returns, and the question I should ask to confirm each.

Check: trace the difference between two reports to the ledger yourself and confirm which reason is real.

3. Suggest a tile layout.

Here is my list of metrics grouped as outcomes, drivers, and early signals. Propose a one-page layout with the number of tiles per row and the order. Explain what a reader should see first and why. Do not choose colours.

Check: test the layout with two leaders who do not work in finance and ask them to state the health of the business in a minute.

4. Draft owner comment prompts.

For an on-time delivery tile that is amber, write five questions an owner should answer in a two-sentence comment so the CEO can decide whether to intervene.

Check: use only the questions that match your process, and drop the rest.

Implementation guide

Governance: the CEO owns the purpose and the review. The CFO owns definitions and reconciliation. Each metric owner owns the explanation and the action. A single person, usually in FP&A, owns the page and the change log, so definitions do not drift without a recorded reason.

Expect the first month to be about definitions rather than visuals. The page will look unfinished for a while, and that is acceptable. A plain table that everyone trusts is worth more than a polished one they do not.

Downloadable assets

The metric spec, page layout, weekly review agenda, and status rule table above are designed to be copied into a spreadsheet. No calculator is attached to this page.

Related resources

• KPI design: choosing 12 metrics that run the company: how to decide which metrics earn a tile.
• Operating cadence: weekly, monthly, quarterly rhythms: where the weekly dashboard review fits.
• Management reporting playbook: turning the dashboard into the monthly pack for the board and lenders.
• Decision rights matrix template: who may act when a tile turns red.
• Financial KPIs for executives: definitions for the financial metrics most dashboards start with.
• CEO toolkit hub: the rest of the operating playbooks for CEOs.

Frequently asked questions

What should be on an executive dashboard?

An executive dashboard should hold about twelve metrics, each tied to a recurring decision: a few outcomes such as revenue, margin, EBITDA, and cash, several drivers such as price realization, on-time delivery, and DSO, and a few early signals such as backlog or the 13-week cash low point. Each needs a named owner and a defined threshold.

How many KPIs should a CEO dashboard have?

Around twelve is a workable limit for a single page, because beyond that readers scan instead of deciding. Keep further detail in drill-downs for the owners. If a metric has not led to a decision in two consecutive quarters and no lender or board requirement keeps it, retire it.

How do you set thresholds on a dashboard?

Set thresholds by method, not by copying a benchmark. Start with contractual floors such as covenants or customer agreements, then your own best trailing six or twelve months, then named public peers if they publish the same metric. Write the basis next to each threshold, and set the act level where inaction costs more than intervention.

How often should an executive dashboard be updated?

Match the refresh to the decision. Cash and operational metrics often need weekly or daily updates, while margin and EBITDA follow the monthly close. Show a last-refreshed stamp on the page and mark any tile older than its agreed window as stale, so no one acts on old data.

Why do executive dashboards fail?

Most fail because definitions and owners were never agreed, not because of the chart design. When two functions use different formulas, the meeting argues about the number instead of the action. Fix definitions, name an owner per metric, attach an action to each threshold, and review the page on a fixed schedule.

How this works in a System of Performance

A dashboard is only as good as the records beneath it. A Performis routine can take on a recurring job like this, with the scope and fit agreed with you: it works from the records your team supplies, prepares the work for your named reviewer, and keeps each approval on record. If you want to see which of your numbers disagree today, bring it to us as a launch partner.