How to create a management dashboard that supports decisions

A management dashboard makes issues needing attention visible and lets you reach records explaining a result. Every metric therefore needs a definition, period and source. This guide uses a fictional sales funnel to show how to prepare metrics and avoid incorrect conclusions.

By Lyntev. Updated on .
About 5 minutes to read.

From data to the next decisionChoose the question, define the calculation and act on the result.

Translation published on . Original Portuguese guide; source revision .

By the end, you can:

  • Define metrics unambiguously
  • Calculate conversions within the same cohort
  • Prepare the dashboard and check results

Choose decisions before choosing charts

Start with three questions someone needs to resolve: which requests await a reply, where opportunities accumulate and what work needs a decision this week? Identify who will act and how often they consult the dashboard. Daily operations may need a list; monthly review may need a trend.

Microsoft’s dashboard guidance recommends considering the audience and clearly presenting essential information. Turn that into a design rule: every metric should explain a situation, open relevant records or guide an action. A number without a definition or consequence may occupy space without helping management.

Write a metrics dictionary

Define what counts, the unit, source fields, period, time zone and exclusions. “Contacts” may mean unique people, requests or opportunities. Choose a unit and retain it throughout the calculation. Record definition changes to avoid confusing historical comparisons.

Separate distinct events. A won proposal is not automatically an issued invoice or received payment. When tracking amounts, state the source and treatment of taxes, cancellations and currencies. Without those rules, a value may appear comparable while measuring something else.

Write a metrics dictionary
Example metricDefinitionRelevant date
Eligible contactsUnique requests admitted to the funnelRequest entry
With a proposalCohort requests with a proposal sentFirst proposal sent
WonCohort requests with recorded acceptanceAcceptance
Awaiting a replyOpen requests awaiting team actionLast action and next action

Determine what each row represents

Before adding values, confirm granularity: does a row represent a request, proposal, proposal line or activity? Joining proposals with activities may repeat the same amount several times. Counting distinct identifiers resolves some counts, but does not automatically correct duplicated sums.

Microsoft’s star-schema modelling separates events and measures from dimensions used for filtering, such as service or owner. This discipline helps even in a small dashboard: keep stable identifiers, explicit relationships and event dates. Preserve history needed to know when an opportunity changed status.

Calculate conversion within the same intake group

Consider a fictional cohort: 100 eligible requests arriving from 1 to 31 October. By the observation on 17 November, 40 had received proposals and 12 were won. Every win belongs to the group with proposals. Observed conversions are 40 ÷ 100 = 40%, 12 ÷ 40 = 30% and 12 ÷ 100 = 12%.

If the 40 proposals comprise 12 won, 18 lost and 10 open, the rate among closed proposals is 12 ÷ 30 = 40%. It measures something different. The 30% includes proposals still open and does not represent the final probability of winning. Compare cohorts with the same follow-up time.

This guide’s calculator counts unique eligible requests, requests with proposals and won requests from the same cohort. Do not divide this month’s sales by this month’s contacts when sales may originate in earlier months. A snapshot of current statuses does not, by itself, show movement between stages either.

The same cohort at three stages
  1. 100 requests

    Defined intake: eligible October requests.

  2. 40 proposals

    40% of requests reached proposal stage by the cut-off.

  3. 12 wins

    30% of proposals and 12% of the cohort, observed up to the cut-off.

Fictional example. Observation date: 17 November. Groups are cumulative and contained within the preceding group.

Organise reading from summary to action

At the top, show priority metrics with their period and update. Then show relevant trends or comparisons. End with records needing action, indicating status, owner and next step. Selecting a metric should open a list respecting the same filters.

Use consistent formats, understandable scales and readable labels. Colour may reinforce a status, but must be accompanied by text. Check mobile reading and keyboard use, especially filters and tables. Show the last update so a synchronisation failure does not appear to mean no activity.

Organise reading from summary to action
QuestionUseful presentationNext action
What is awaiting a reply?Count and list ordered by waiting timeOpen and assign the request
Is intake changing?Time series with comparable periodsConsult source and quality
Where do opportunities stall?Stages and time spent in eachReview cases and reasons

Check the result before interpreting the trend

Choose some records and follow each from source to metric. Check identifiers, statuses, dates and filters. Compare totals with the authoritative source. Test an empty period, a proposal without an associated contact, a time-zone change and a repeated record.

Distinguish zero from having no basis for calculation. With zero contacts, conversion should show “no data” or “not applicable”. DAX’s DIVIDE function, for example, returns blank by default when the denominator is zero. Define the same behaviour in your chosen technology.

Check the result before interpreting the trend
Error signalWhat to check
Conversion above 100%Cohort, unit, duplicates and differing filters.
Tripled valueRelationships and totals repeated on detail lines.
Sudden drop on one dayTime zone, incomplete import and update.
Percentage without recordsDivision by zero and distinction between blank and zero.

Make rules visible to dashboard users

Attach a worksheet to each metric and name who maintains its definition. During review, ask whether the team could decide and act. A trend suggests where to investigate; explaining its cause requires cases and context. Add metrics when they address a specific need.

Worksheet for your project

Management metric worksheet
Question and decision it should support:
Responsible person and consultation frequency:
Metric name, unit and definition:
What each source row represents:
System, table and fields used:
Numerator and denominator, where applicable:
Cohort, period, time zone and observation date:
Filters, exclusions and duplicate handling:
Behaviour when no data exists:
Update frequency and last update time:
List of records explaining the result:
Expected action and person responsible:
Source reconciliation and test cases:
Definition version and approval date:

Read your funnel with clear denominators

Enter figures for the same group of requests, with the same intake period and cut-off date. Initial values are a fictional example.

Data for the same group of requests

Unique requests arriving in the chosen period.

How many of these requests received at least one proposal.

How many of these requests with proposals resulted in a deal.

12 of 100 requests were won in this group.

Received a proposal
40%Requests with a proposal ÷ requests
Wins among proposals
30%Requests won ÷ requests with a proposal
Wins in the initial group
12%Requests won ÷ requests

60 requests still without a proposal and 28 with a proposal but no deal won yet. They may be open or have another outcome.

How the calculation works

Each rate divides the requests reaching the next stage by the number at the starting stage, multiplied by 100.

Count each unique request once per stage. Three proposal versions for the same request still represent one request with a proposal. The same person may make distinct requests.

A zero denominator appears as “No data”, because there is no basis for calculating the rate. Open cases are not automatically lost. A rate for this group is neither a forecast, target nor market benchmark.

Calculations run in this browser. Values are neither sent nor saved.

Sources and criteria

Content prepared with AI assistance and research using technical sources, with checks of the relevant examples and calculations. The references support the guide’s decisions; examples identified as fictitious let you try out the method.

Wellinton Oliveira

Written by

CEO of Lyntev

Technology with context, original analysis and sources you can consult.