What metrics should I actually track?

The thing most dashboard exercises skip: deciding what the dashboard is for.

The right way to start is one sentence per dashboard. "This dashboard exists to tell me [what question] in [what timeframe] so I can [what action]." If the question is fuzzy, every metric chosen to answer it will be fuzzy too. If the timeframe is wrong (you're looking at a metric monthly that only moves quarterly), the dashboard creates anxiety instead of insight. If there's no action you'd take from the answer, the metric is decorative.

A useful sentence: "This dashboard tells me whether our service delivery is profitable enough this month to invest the surplus in marketing, so I can decide whether to release the Q2 paid-search budget." Specific. Time-bounded. Actionable.

An unuseful sentence: "This dashboard shows me how the business is doing." Too broad. Will produce a sprawling dashboard nobody uses.

Most service businesses need two or three dashboards, not one. One for leadership (strategic health). One for operations (delivery and quality). One for sales and marketing (pipeline and conversion). Each one answers different questions for different audiences. Trying to merge them produces a single dashboard nobody can read.

Two ways to slice any metric, and both matter.

Leading vs lagging. A lagging indicator tells you what happened (revenue this month, win rate last quarter, customer churn for the year). A leading indicator tells you what's coming (qualified opportunities created this week, time-to-respond on inbound enquiries, customer health score trend). Lagging metrics are easier to measure and useless for action. Leading metrics are harder to measure and what you act on. Most dashboards over-weight lagging. The good ones balance the two.

Financial vs operational. Financial metrics speak the language of the income statement: revenue, gross margin, EBITDA, cash. Operational metrics speak the language of the work: utilisation, on-time delivery, defect rate, employee retention. Leadership needs both. Looking only at financial metrics flies blind to operational reality. Looking only at operational metrics flies blind to whether the operations translate into financial results.

The 2x2 (leading vs lagging, financial vs operational) is the simplest way to make sure your dashboard isn't lopsided. Every quadrant deserves at least one metric. Most dashboards we audit have all four in the top-right quadrant (lagging financial) and very little in the others.

Not all of these belong on every dashboard. Read them as a catalog. Pick the 8 to 12 that match your business.

Financial health (lagging, financial)

• Revenue. Self-explanatory. Track total, by service line, by customer segment. The split tells you concentration risk.
• Gross margin. Revenue minus direct cost of delivery. The single most informative profitability metric for service businesses. If you only track one financial number, this is it.
• EBITDA. Earnings before interest, tax, depreciation, amortisation. The number a buyer would talk about. Track monthly with year-over-year comparison.
• Operating cash flow. Cash in minus cash out from operations. Lagging by definition but more honest than profit. Tells you whether the business is funding itself.
• Days sales outstanding (DSO). Average days between invoicing and getting paid. Rising DSO means cash flow is getting harder, often before the income statement notices.
• Cash runway. Months of operating cost the bank balance covers. A safety metric, not a growth one, but the one that wakes you up at 3am if it's wrong.

Sales and marketing (mix of leading and lagging)

• Leads / qualified opportunities created. The top of the funnel. Leading indicator of future revenue. Track weekly.
• Conversion rate (lead to close). What percentage of opportunities become paying clients. Tells you whether the funnel is healthy. Lagging by a sales cycle.
• Win rate (on submitted proposals). What percentage of quotes you actually win. Different from conversion rate because it excludes the leads that never made it to a proposal.
• Average order value (AOV) or average transaction value. What a typical job pays. Useful for sizing the funnel ("we need 8 of these to hit target").
• Customer Acquisition Cost (CAC). Total sales and marketing spend divided by new customers acquired. If CAC is rising faster than AOV, your unit economics are degrading.
• ROAS (return on ad spend). Revenue attributable to paid ads divided by ad spend. Specific to paid acquisition. Helpful for deciding when to scale or pull back.
• Pipeline coverage. Total open pipeline value divided by quarterly target. 3-4x is healthy for most service businesses. Below 2x is a warning signal.
• Time-to-first-response (on inbound). How long between someone reaching out and getting a real reply. Strong leading indicator for win rate.
• Views and engagement (content marketing). Useful as trend lines if content is part of your acquisition strategy. Less useful in isolation than people think.

Service delivery and operations (mix)

• Project margin (versus quote). Did the project make the margin you priced for? Variance is one of the most diagnostic operational signals.
• Utilisation (billable / total hours). For time-billing businesses. For fixed-price ones, the equivalent is throughput per FTE per period.
• On-time delivery rate. Percentage of projects delivered by the committed date. A leading indicator of customer satisfaction.
• Rework rate. Percentage of jobs that required a callback, revision, or fix. Often the most actionable operational signal.
• Cycle time. How long it takes a project to go from start to delivery. Trends matter more than the number.
• Drift between documented process and actual. The Oversight metric we covered in Process Maps That Don't Gather Dust. Increasingly tracked via AI-assisted comparison.

People (leading, operational)

• Voluntary turnover rate. Annual % of employees who left voluntarily. Above industry norms is a warning. Trend matters more than the absolute number.
• Time-to-fill (open roles). Days from req opened to offer accepted. Rising time-to-fill is a leading indicator of capacity issues.
• Employee NPS (eNPS). "How likely are you to recommend working here?" on 0-10. Cheap to measure quarterly. Surprisingly diagnostic.
• Onboarding ramp time. How long until a new hire is producing at expected level. Decreasing ramp time is a sign your Playbook is doing its job.

Customer (mix)

• Customer retention / churn rate. Especially important for recurring-revenue or repeat-engagement businesses.
• Net Promoter Score (NPS). Customer-side version. Trend more useful than absolute.
• Customer concentration. What percentage of revenue comes from your top 3, top 10, top 20% of customers. Diagnostic for risk and for sale-readiness.
• Repeat purchase rate / share of wallet. Are existing customers buying more? The cheapest growth lever.

The list above has 30+ metrics. Most service businesses need 8 to 12 on their primary leadership dashboard. A good filter is the test below.

For each metric you're considering, answer four questions:

1. Do we have a reliable source for the data? If the answer is "we'd have to manually compile it from three tools," the metric won't get tracked. Either fix the data first, or pick a different metric.
2. Does this metric, if it moved, change a decision we'd actually make? If you can't name the decision, drop the metric. It's vanity.
3. Are we already pretty sure what the answer is? Some metrics confirm what you already know. They're less valuable than metrics whose answer you couldn't reliably predict.
4. Will somebody on the leadership team own this number? A metric without an owner is a metric that drifts unsupervised.

Metrics that fail any of the four don't belong on the dashboard. Save them for a secondary view or drop them entirely.

The wrong cadence is the most common dashboard failure mode. A metric reviewed too frequently produces noise and anxiety. A metric reviewed too infrequently produces surprises.

The same metric can sit at different cadences depending on the business. Revenue might be a daily metric for a high-volume transactional business and a monthly metric for a project-based services business. Map your metrics to the cadence that matches how fast they actually move.

Five principles. Honour them and the dashboard gets opened. Skip them and it doesn't.

1. One screen, no scrolling. If the dashboard doesn't fit on one screen, it's two dashboards pretending to be one. Force the prioritisation by limiting the real estate.

2. Trends, not just numbers. A number on its own is meaningless. A number with a sparkline (last 12 weeks/months) is informative. A number with a sparkline and a comparison to plan is actionable. Show trends.

3. Source of truth, not source of more truths. Each metric should pull from one canonical source, not be reconciled across multiple. If your dashboard requires manual reconciliation each week, it will be late or wrong, and over time both.

4. Named owner per metric. Every number on the dashboard has a person whose job it is to know what's going on with it. The owner answers questions about the number when they come up. Without owners, the dashboard is a museum.

5. Built into the meeting cadence. The dashboard doesn't get opened randomly. It's the agenda for the weekly leadership meeting. The monthly business review walks through it. If the dashboard exists outside any meeting cadence, it's a hobby project.

Tooling matters less than these principles. Excel with a few good charts beats a fancy BI tool that nobody opens. A simple Notion or Google Sheets dashboard with the right metrics and the right owners beats Tableau with the wrong metrics and no owners. Pick the tool the team will keep current. Sophistication of platform is a distant second to fidelity of metric.

The dashboard is the visible surface of the Oversight stage in ControlShift. The eighth stage. The one that closes the loop between the operating intelligence you've captured and the leadership decisions you're making.

A business with a documented Playbook and codified operational intelligence but no working dashboard is half of an operating system. The Playbook tells the team how to do the work. The dashboard tells leadership whether the work is producing the outcome the business needs. Without both, you have either flying blind with documentation, or flying blind with data.

The pattern that works: dashboard metrics tie to operational behaviours documented in the Playbook. Rework rate climbs? The relevant procedures get audited for drift. Time-to-response degrades? The intake process gets re-examined. Conversion rate dips? The sales decision rules get reviewed. The dashboard surfaces the symptom; the Playbook is where the diagnosis happens; the corrective change goes back into the Playbook. Operating intelligence and operating data, working in pair.

This is the stage we work with clients on after the codification engagement (see our 16-week roadmap for what comes first). Without the codified Playbook, the dashboard is just data. With it, the dashboard becomes the early-warning system for the operating asset you've built.

You probably don't need a new tool. You probably need fewer metrics, clearer owners, a tighter cadence, and one screen of trends that ties to the meetings you're already having.

If your current dashboard isn't getting opened, the answer is rarely "make it prettier." The answer is usually "ask the harder question of what it's for, then strip everything that doesn't serve that question." A dashboard with 8 metrics, owned, current, and reviewed in the Monday meeting beats a dashboard with 30 metrics, scattered ownership, partial freshness, and no meeting tied to it. Every time.