How to build a KPI dashboard your leadership team will use

Why most dashboards are ignored

Dashboards fail in predictable ways. They show too many numbers, so nothing stands out. They show numbers nobody can act on. They rely on manual updates that stop after a month. Or they disagree with the accounting system, so nobody trusts them. The result is a nice-looking screen that the leadership team stops opening.

The fix is to design backward from decisions. What does the owner or leadership team decide each week or month? Whether to hire, where to focus sales effort, which jobs or clients need attention, whether cash is tight. The dashboard should show the numbers that inform those decisions, and little else.

There is also a cost to building it badly. An owner who spends two or three hours every Monday pulling numbers from four systems is doing work that should be automatic, and is making decisions on numbers that are already a few days old by the time they are assembled.

Choose a small set of metrics

Aim for 6 to 10 metrics across the stages of your business: getting work, doing work and getting paid. Include a mix of leading indicators, which predict what is coming, and lagging indicators, which report what happened. The right list depends on your model, but most service businesses draw from a similar pool.

Resist adding a metric just because it is easy to measure. Website visits, social followers and email open rates are tempting because the data is readily available, but they rarely drive a leadership decision in a service business. If a number would not change what you do next week, leave it off the main dashboard.

• New enquiries and lead response time (getting work)
• Quotes or proposals sent, and close rate
• Booked revenue or backlog for the next 30 to 90 days
• Utilization or billable hours for staff or technicians (doing work)
• Jobs or matters completed, and average job value
• Callbacks, rework or complaints
• Revenue and gross margin by service line (getting paid)
• Accounts receivable and days sales outstanding
• Cash position against a minimum threshold

Define each metric precisely

For every metric, write a one-line definition, its source system, how often it updates and who owns it. 'Close rate: quotes marked won divided by quotes sent in the same month, from Jobber, updated daily, owned by the sales manager.' This sounds tedious, but it prevents the most common dashboard argument, which is two people meaning different things by the same word.

Agree targets or ranges where they help. A metric with a target turns green, amber or red at a glance. Without one, people have to remember whether 42 hours is good or bad. Base early targets on your own recent history rather than industry figures, then adjust as you learn.

Connect the data

Most of the numbers already exist in your systems: the CRM (HubSpot, Pipedrive, Salesforce), the field service or practice platform (ServiceTitan, Jobber, Housecall Pro, Clio, Karbon), accounting (QuickBooks Online, Xero) and the phone system. The job is to pull them into one place automatically.

Some platforms include their own dashboards, which may be enough if most of your metrics live in one system. When data spans several systems, a dashboard tool such as Looker Studio, Power BI, Databox or Geckoboard can connect to many sources directly. QuickBooks Online Advanced also now offers custom KPIs and dashboards built on accounting data. Where a connector does not exist, Zapier, Make or a scheduled export into Google Sheets or Excel can bridge the gap.

Avoid manual entry wherever you can. Every number someone has to type in weekly is a number that will eventually be late or wrong.

Decide the time frame for each metric too. Some numbers make sense weekly, such as enquiries, response time and jobs completed. Others are noisy week to week and better viewed monthly or as a rolling 90-day figure, such as close rate, margin and retention. Showing every metric on the same weekly basis can make normal variation look like a crisis.

Check the numbers against reality

Before launch, reconcile each metric against its source for a past month. Revenue on the dashboard should match the profit and loss report. Job counts should match the field service system. Where they differ, find out why. It is often a definition issue, such as booked versus invoiced revenue, or a data issue, such as jobs that were never closed.

Data quality problems surface quickly once numbers are visible. That is useful. If the CRM shows half the leads with no source, the dashboard will expose it, and it may be worth a short CRM cleanup before relying on lead metrics.

Document what you find. A short note explaining known quirks, such as 'revenue excludes pass-through materials' or 'leads before March lack a source', saves repeated questions and keeps trust in the numbers when new people join the meeting.

Make it part of the week

A dashboard nobody reviews is decoration. Put it at the center of a short weekly or fortnightly leadership meeting. Go through each metric briefly, focus discussion on the ones outside their target range, and agree specific actions with owners. Over time the meeting gets faster, because the numbers do the reporting and the discussion focuses on decisions.

Review the metric list itself every quarter. Remove numbers that never prompt action and add ones that answer new questions. InTheory would usually build a first dashboard around one workflow being improved, so that the baseline and progress are visible to everyone, then expand it once the habit is established.

Make the dashboard easy to reach. Pin it in Teams or Slack, set it as a browser homepage for the leadership team, or send a scheduled snapshot by email before the meeting. Every extra click reduces how often people look.

Using AI with your dashboard

AI features in dashboard tools and accounting platforms can now summarize what changed, answer questions in plain language and point to the customers or jobs behind a movement. These can save time, especially for owners who are not comfortable building reports.

Treat AI explanations as hypotheses, not conclusions. Check the underlying numbers before acting, and be careful with financial interpretations that could affect pricing, staffing or lending decisions. The person accountable for the metric should confirm the explanation.