Spec a new dashboard with audience and layout
intermediateClaude SonnetData & AnalyticsBusiness Intelligencedashboardbidesign-speclookerbusiness-intelligence
Use case
Use this prompt when scoping a new dashboard request from a stakeholder. Forces the audience-and-decisions framing that prevents the standard 30-tile dashboard nobody opens.
The prompt
You are a senior BI developer scoping a new dashboard. The stakeholder is asking for "a dashboard for X" — your job is to translate that into a specific, decision-driven spec before opening the BI tool. Inputs: - Stakeholder request (verbatim):{{stakeholder_request}}- Stakeholder role and context:{{stakeholder_role}}- Decisions or actions the dashboard should support:{{decisions}}- Cadence of those decisions:{{cadence}}(daily, weekly, monthly, ad-hoc) - Source of truth tables / models available:{{source_tables}}- BI tool:{{bi_tool}}- Existing dashboards that overlap:{{existing_dashboards}}- Constraints:{{constraints}}(refresh latency, access permissions, cost) Produce a dashboard spec with these sections: ## Audience and Decisions (the framing) - **Primary audience:** who opens this every{{cadence}}- **Decisions:** the 2–4 specific decisions or actions this dashboard should support. Be concrete: "Decide which 3 channels to reallocate spend to next week," not "Understand marketing performance." - **What this dashboard is not for:** out-of-scope uses that should redirect to other dashboards or ad-hoc analysis. ## Top-Level KPIs (the headline tiles) 3–5 metrics that go at the top of the dashboard, with for each: - The metric, with a clean definition - Comparison frame (vs. last period, vs. target, vs. baseline) - Source model / measure name in{{bi_tool}}- Why this is in the headline (which decision does it support) ## Detail Sections Group the remaining content into 2–4 sections, each with a clear question it answers. For each section: - The question (e.g., "Where is conversion dropping?") - The 3–6 charts that answer it - For each chart: chart type, dimensions, metric, default time range - Any filter controls ## Layout Describe the layout in plain text — top row, second row, etc. Keep the dashboard to one screen for the headline; details below the fold. ## What to Leave Out Three to five things that are tempting but should not be on this dashboard: - Tiles that don't tie to a decision - Tiles that overlap with another canonical dashboard - Tiles that require manual interpretation the audience won't do - Vanity metrics ## Filters and Drilldowns - Global filters (date range, segment, geo) - In-tile drilldown behavior - Saved-view requirements (e.g., per-region default views) ## Performance and Maintenance - Expected query cost / refresh latency - Owner team for maintenance - Review cadence (when this dashboard gets revisited) ## Open Questions Two to four things you need from the stakeholder before building. Tone: opinionated, decision-driven, willing to push back on requests that don't tie to action. The best dashboard spec eliminates 60% of the original request.
Variables
{{{{stakeholder_request}}}}Replace with your {{stakeholder request}}{{{{stakeholder_role}}}}Replace with your {{stakeholder role}}{{{{decisions}}}}Replace with your {{decisions}}{{{{cadence}}}}Replace with your {{cadence}}{{{{source_tables}}}}Replace with your {{source tables}}{{{{bi_tool}}}}Replace with your {{bi tool}}{{{{existing_dashboards}}}}Replace with your {{existing dashboards}}{{{{constraints}}}}Replace with your {{constraints}}
Example
Input
stakeholder_request: "Hey can you build us a marketing dashboard? We need to see all our channels and how they're performing — spend, CAC, conversion, the funnel. Should be everything." stakeholder_role: VP of Marketing at a B2B SaaS company; she runs a weekly Friday review with her team leads (paid, content, lifecycle). decisions: | 1. Decide which paid channels to reallocate spend across next week 2. Decide which content / SEO topics are working enough to invest more in 3. Identify when CAC has broken from its trend so we can investigate cadence: Weekly Friday review; some tiles glanced at daily by the paid team source_tables: | fct_marketing_spend, fct_marketing_clicks, fct_signups, fct_trials, fct_subscriptions dim_channels, dim_campaigns, dim_content, dim_dates bi_tool: Looker existing_dashboards: | "Acquisition Overview" (high-level CAC and signups, exec-facing) — overlaps but at less granular cuts "Paid Channels Deep Dive" (paid team's daily ops dashboard) — overlaps for paid-only views constraints: | Looker dashboards above ~30 tiles slow to load badly Weekly cadence on most data; daily on paid spend No PII; no individual-customer detail (use existing CRM tooling for that)
Output
## Audience and Decisions - **Primary audience:** VP Marketing and her three team leads (paid, content, lifecycle), in the Friday weekly review. Loaded once a week, scanned in 15 minutes. - **Decisions this dashboard supports:** 1. Reallocate paid spend across channels for next week. 2. Identify content and SEO topics worth doubling down on. 3. Spot CAC anomalies that warrant investigation. - **Not for:** daily paid ops (use the existing Paid Channels Deep Dive); exec-facing CAC narrative (use Acquisition Overview); individual lead pursuit (use CRM); attribution debates (use a separate attribution review dashboard, not yet built). ## Top-Level KPIs (headline tiles, top of dashboard) 1. **Blended CAC, this week vs. trailing 4-week average vs. target.** From `fct_marketing_spend` ÷ `fct_subscriptions.new_paid` over matched windows. Triggers decision 3. 2. **Total marketing-attributed paid customers, this week vs. last week vs. plan.** From `fct_subscriptions` filtered to marketing-source. Volume read. 3. **Total marketing spend, this week vs. last week vs. plan.** From `fct_marketing_spend`. Spend pacing. 4. **Marketing-sourced ARR added this week.** Quality companion to the volume metric. ## Detail Sections ### Section 1: Where Is Spend Going and What Is It Producing? (decision 1) - **Spend by channel, last 4 weeks** — stacked bar by channel × week. From `fct_marketing_spend`. - **CAC by channel, last 4 weeks** — line chart, one line per top-5 channel. Channels with too few signups are suppressed (min 20 signups/week to display). - **Cost per qualified opportunity by channel** — table sortable by CAC and trend. - **Channel mix shift** — heatmap of (channel × week) with spend share, last 8 weeks. Helps see whether spend has been quietly shifting. Filter: date range (default last 4 weeks); channel multi-select. ### Section 2: Content and SEO Performance (decision 2) - **Top 20 content pieces by signups, last 90 days.** Table with: content title, channel (organic/paid), signups, CAC. - **SEO trended sessions and signups, last 12 weeks.** Line chart, two axes (sessions left, signups right). - **New-content signup trajectory** — content published in the last 90 days, with weeks-since-publish on x-axis and cumulative signups on y. Identifies which new content is compounding. Filter: content topic; channel. ### Section 3: CAC Trend and Anomaly Watch (decision 3) - **Blended CAC trended weekly, last 26 weeks** — line chart with shaded ±1σ band based on 26-week rolling. CAC outside the band is a flag. - **CAC by acquisition channel, last 13 weeks** — small multiples, one panel per channel. - **Spend × volume × CAC table** — current week vs. 4-week avg by channel, with deltas highlighted. ## Layout - **Row 1 (above the fold):** the four headline KPI tiles, each with a sparkline below. - **Row 2 (Section 1):** 4 tiles in a 2×2 grid. - **Row 3 (Section 2):** 3 tiles, full width for the table. - **Row 4 (Section 3):** 3 tiles. The CAC trend with band is full width, the rest in a 2-column. Total: 14 tiles. Well below the 30-tile slowness threshold. Decision-anchored. ## What to Leave Out - **Email open rates and per-send metrics.** Lifecycle's existing dashboard handles this; including it here invites debate over an attribution model that doesn't tie to weekly decisions. - **Per-customer revenue or retention.** Out of scope — different dashboard, different team. - **Multi-touch attribution charts.** Attribution is a separate conversation; including it here turns every weekly review into an attribution debate. - **Funnel-by-funnel drilldown for every channel.** Tempting but each becomes a separate sub-dashboard. Build the parent dashboard first; spawn channel-specific drilldowns only if the team uses the parent for 4+ weeks. - **Brand search vs. non-brand split.** Important for SEO but should live in the SEO team's dashboard, not the leadership review. ## Filters and Drilldowns - **Global filters:** date range (default last 4 weeks; presets for 7/30/90 days); channel multi-select. - **Drilldowns:** every channel-aggregated tile drills to per-campaign breakdown; every content tile drills to a single-content view. - **Saved views:** "Last week" (default for Friday review), "Last 4 weeks," "Last quarter." ## Performance and Maintenance - Expected refresh: nightly for most; paid spend tiles refresh every 4h. - Owner: BI team. - Review cadence: re-evaluate the dashboard's tiles every 6 months. Tiles not viewed in 60 days are candidates for removal. ## Open Questions 1. **Is "marketing-attributed" first-touch, last-touch, or a model?** Need a single answer before building CAC tiles, since the existing Acquisition Overview uses last-touch. 2. **Are paid social and paid search bundled or separated as channels?** Affects channel mix tiles. 3. **Is there a target CAC per channel, or only a blended target?** Affects the comparison frame. 4. **Confirm spend data is loaded daily by 9am UTC.** If later, the Friday review may see stale data.
Tips for best results
- 1Always start with the decisions. A dashboard that doesn't enable a specific decision is a dashboard nobody opens twice.
- 2The headline tiles must answer 'are we okay?' in 30 seconds. If the audience has to scroll to find a target comparison, the dashboard isn't doing its job.
- 3Aggressively cut tiles. Every tile has a maintenance cost; tiles that don't drive a decision still cost the BI team time when something changes upstream.
- 4Spec what the dashboard is NOT for. Out-of-scope uses redirect to other dashboards and prevent your dashboard from accumulating every passing request.
- 5Always include a review cadence. Dashboards rot — without a scheduled review, you'll be debugging tiles built for someone who left 18 months ago.
Related prompts
Outline a data-storytelling deck for leadership
intermediateProduce a slide-by-slide outline for a data-storytelling deck — built around a single argument, with charts that earn their place and a clear ask.
Data & Analyticsdata-storytellingpresentationbi
Comms plan for deprecating a legacy dashboard
intermediateProduce a deprecation comms plan for a legacy dashboard — audience map, replacement, timeline, and the messages that prevent angry Slack threads.
Data & Analyticsdashboarddeprecationcommunications
Generate exec narrative on top of a dashboard
advancedProduce a one-page exec narrative that sits on top of a dashboard data dump — the story leadership reads instead of squinting at the dashboard.
Data & Analyticsexecutive-reportingkpinarrative
Need help implementing this prompt in your workflow?
Book a call