analytics-dashboard.

Most LinkedIn Analytics dashboards reduce a thirty-day export to two top-line metrics and a vague trend arrow. analytics-dashboard does the opposite. It reads every sheet in the export, builds five chart panels that each answer a different strategic question, and writes a paired analysis that names five specific content moves to make next month based on what the charts actually show.

The challenge with LinkedIn analytics is that the platform's native dashboard is shaped for sales conversations, not for content writers. Top-line numbers, vague trend lines, no quadrant analysis on individual posts. analytics-dashboard rebuilds the same data into a writer-shaped artifact: five panels that each answer a question a content writer actually has, plus a written analysis that names what to do about it.

§01What it does

The skill expects a LinkedIn Analytics export as an .xlsx file. It reads every sheet (DISCOVERY, ENGAGEMENT, TOP POSTS, FOLLOWERS, DEMOGRAPHICS), cleans the headers, and merges the two top-posts tables (top-by-engagements and top-by-impressions) into one unified dataset per post.

Then it builds a single React artifact with a fixed dark theme and Recharts visualisations. The headline metrics row carries total impressions, total reach, total new followers, daily averages, average engagement rate, and total posts tracked. Below that sit five panels in order: an engagement trend chart with daily impressions and engagements over the full date range plus markers on the top three spike days; a follower growth area chart with a seven-day moving average; a post-performance scatter that colour-codes posts into four quadrants (stars, viral but shallow, niche gold, underperformers); a day-of-week heatmap showing average impressions and engagements by day; and an audience breakdown with bar charts for job titles, industries, seniority, company size, and top locations.

Below the dashboard, the skill writes a strategic analysis. Performance summary names whether the trajectory is growing, plateauing, or declining. Five data-backed recommendations follow, each tied to a specific chart panel rather than to generic advice.

§02The four-quadrant scatter is the load-bearing chart

Every other panel describes the writer's overall account performance. The post-performance scatter describes individual post types, which is where the actionable content decisions actually live. Stars (high reach, high engagement) tell the writer what to do more of. Viral but shallow (high reach, low engagement) tell the writer which hooks are pulling impressions but losing the read. Niche gold (low reach, high engagement) tell the writer which posts are landing with the right audience even though they did not break out. Underperformers in the bottom-left tell the writer which patterns to drop.

Most LinkedIn dashboards collapse this distinction by ranking posts on a single combined score. The split into four quadrants is what surfaces the asymmetric moves, because more stars and fewer underperformers are different actions and they need different recommendations.

§03Setup

# Get the export:
#   LinkedIn Analytics → date range (30/60/90 days) → Export
#   Result is a single .xlsx

# Trigger phrases:
#   "analyse my linkedin"
#   "linkedin analytics"
#   "build my dashboard"
#   "review my performance"

The artifact is a self-contained React file with Recharts. It runs in any React environment that can render the artifact directly, including Claude's artifact viewer.

§04Caveats

A thirty-day export is the minimum useful window. Shorter ranges produce charts where the spikes overwhelm the trend and the recommendations turn into reactions to a single post.

The five recommendations are only as good as the data behind them. An account with twenty posts in the export window will get coarser advice than an account with two hundred. Run again at a longer window or after publishing more.

Number formatting is consistent across the artifact (67K, not 67000; 1.2M, not 1200000). That is a real readability gain and a deliberate choice.