Dashboard Design with AI: 6 Metrics Every SMB Founder Tracks

When a founder of a 70-person SaaS asked me to "audit her dashboards" last quarter, she opened a Looker workspace with 41 charts across three tabs. She could not, on the spot, tell me which three numbers had moved in the past week. That is not a dashboard problem. That is a thinking problem.

Why does every SMB dashboard get bloated?

Because new charts are cheap and removing charts is political. Someone on the team built each one for a reason. Removing it feels like dismissing the reason. So they accumulate.

Definition: Dashboard bloat — the steady accretion of charts that each made sense individually but produce no decision when viewed together.

The fix is not "build a better dashboard." The fix is to define the founder's weekly question first, then keep only the charts that answer it. AI is genuinely good at the second part. It is useless without the first.

What's the right number of founder-level metrics?

Six. Not five, not ten. Six fits on one screen, fits in a 15-minute weekly review, and forces real prioritisation. Five tempts founders to skip operational health; ten dilutes attention back to scanning mode.

The six split into three pairs:

• Two business-outcome metrics — revenue, retention, gross margin. The "did this week matter" pair.
• Two pipeline-leading metrics — qualified leads, activation, time-to-value. The "what's coming" pair.
• Two operational-health metrics — cash runway, churn risk, team capacity. The "what could break" pair.

How does stage change the answer?

The biggest mistake I see is founders carrying a PMF-stage dashboard into the growth stage. PMF metrics are about whether the product is loved by a few; growth metrics are about whether the engine compounds; scale metrics are about whether the system survives.

PMF stage (under ~$1M ARR, under 30 people)

• Weekly active users / customers
• Activation rate (first meaningful action)
• Qualitative NPS or "would-be-disappointed" share
• Cash runway in months
• Top-1 customer concentration
• Time-to-first-value (signup → first job done)

Growth stage (~$1-10M ARR, 30-150 people)

• Net revenue retention
• New MRR by source (qualified pipeline → win rate)
• Gross margin trend
• Sales cycle length
• Customer acquisition cost payback in months
• Engineering throughput vs roadmap commit

Scale stage (~$10M+ ARR, 150-500 people)

• Rule-of-40 (growth % + margin %)
• Logo retention vs net dollar retention spread
• Pipeline coverage ratio (next quarter)
• Hiring plan vs actual (lead time)
• Free-cash-flow conversion
• Incident rate / SLA breach rate

Definition: Stage-aware metric — a number that becomes diagnostic at one company stage and noisy at another, e.g. NPS-survey scores at PMF are signal; the same scores at scale are vanity unless segmented.

Where does AI actually help?

In three specific places, none of them being "decide what to measure."

1. Translating intent into query. "Show me weekly active customers split by signup cohort, last 12 weeks" — a founder shouldn't write the SQL. AI writes a first-pass query, you review the JOIN logic, you ship.
2. Naming the chart honestly. AI is good at suggesting the question the chart answers, which forces clarity. If a chart can't be summarised as a question, it shouldn't ship.
3. Catching the "this looks weird" moment. Feeding a chart's last 12 data points to an LLM with "anything notable here?" surfaces the inflection a human eye glides past in a Monday scan.

Copy/paste prompt for query generation

You are helping a non-technical founder build a metric query.

Business context: [E.g. "B2B SaaS, ~$3M ARR, ~50 staff"]
Data warehouse: [E.g. "BigQuery", "Postgres", "Snowflake"]
Tables I have: [list with 1-line description each]

The metric I want: [name + 1-sentence business definition]
The cut I want: [time grain + segments, e.g. "weekly, by signup cohort"]
The decision this metric drives: [1 sentence — what will change if it moves]

Please:
1. Propose the SQL query (or BI calc) with explicit assumptions called out.
2. Flag the 2 most likely places this query is wrong (timezone, deduping, etc.).
3. Suggest the single chart title that names the question this answers.

The "decision this drives" line is the one most founders skip. Without it, AI produces a technically-correct query for a metric that no one will act on.

Tool tip (AIAdvisoryBoard.me): The six-metric discipline only sticks if the weekly review surfaces a Plan → Fact → Gap line for each metric — what was committed, what happened, what the gap means. Our daily-management OS automates that exact view across the company, so the founder review starts from the gap, not from the chart. The 7-day diagnostic shows you which of your current dashboards survive the cut and which quietly go away. See it at https://aiadvisoryboard.me/?lang=en.

Manager scan (2-minute digest example)

• Six metrics on one page — anything more goes to a separate ops view
• Each metric paired with a Plan (committed at start of week) and Fact (what happened)
• Gap line written in words, not numbers — "missed activation by 14 because onboarding email broke Tuesday"
• Stage labelled on the dashboard header — so the team knows why these six
• Owner per metric — engineering doesn't own revenue, sales doesn't own latency
• Threshold per metric — a number that triggers a conversation, not just a colour change
• Last-12-week sparkline so direction beats single-point noise
• Token / volume / "active user" charts deliberately excluded from the founder view
• Quarterly review of whether the six are still the right six for this stage
• One copy/paste template per metric so a new analyst can rebuild the dashboard in a day

Micro-case (what changes after 7-14 days)

A 90-person professional services firm had a 32-chart dashboard that the CEO admitted she opened "maybe once a week, then closes." We replaced it with six metrics tied to her growth-stage reality: net revenue retention, qualified pipeline by source, gross margin trend, sales cycle length, CAC payback, and engineering throughput vs commit. The first week's review took 22 minutes and surfaced two issues nobody had named — sales cycle had crept up 11 days versus the prior quarter, and engineering had under-shipped against commit four sprints running. Neither was visible in the 32-chart version because both lived as line-items in larger charts. By week three the CEO was running the review on her own and asking sharper questions of each function lead. The old dashboard wasn't deleted — it was demoted to a "deep dive" link nobody clicked.

Note on this case: This example is illustrative — based on typical patterns we observe with companies of 30-500 employees, not a single named client. Specific numbers are rounded approximations of common ranges, not guarantees.

Tool tip (AIAdvisoryBoard.me): The reason most founder dashboards drift back to bloat is that there's no automated Plan → Fact → Gap loop pushing the six metrics back to the founder every Monday. Our system pulls Plan from your weekly commits, Fact from your operational systems, and writes the Gap narrative in plain language — so the dashboard doesn't depend on a founder remembering to open it. Start the 7-day diagnostic at https://aiadvisoryboard.me/?lang=en.

FAQ

Can AI choose my six metrics for me? No — and any tool that claims it can is selling a template, not a system. Metric selection requires knowing your stage, your bottleneck, and your strategic bet. AI helps after that decision, not before.

What about leading vs lagging indicators? The pipeline-leading pair covers the leading side. The business-outcome pair is mostly lagging — and that's the point. You need both, weighted toward leading for action and lagging for accountability.

How often should the six change? Once a quarter is a healthy cadence. Less often and you miss stage transitions; more often and you lose comparability. Schedule a quarterly "metric audit" the same way you'd schedule a board prep.

My CFO wants 30 metrics. What do I do? The CFO needs 30 — for financial close, audit, and modelling. The founder needs six. They are different jobs. The six is the founder's weekly review; the 30 lives in the finance back-office, not in the founder's Monday.

Is six different for a B2C SMB? The pairs structure holds, the metric names change. B2C swaps "qualified pipeline" for "weekly active buyers" and "sales cycle" for "repeat purchase interval." Same discipline, different vocabulary.

Conclusion

A founder dashboard is not a data product. It is a decision artefact. Six metrics, one screen, one weekly question — anything more is procrastination dressed as rigour.

Pick the stage you're actually in. Choose your six. Let AI build the queries. Delete everything else from the founder view.

If you want a system that surfaces the Plan → Fact → Gap automatically — every day, across the company — see how the 7-day diagnostic works at https://aiadvisoryboard.me/?lang=en.