Social Media Scorecard

Context

A global pharmaceutical aesthetics brand manages social media accounts across dozens of regional markets — US, Mexico, Germany, and many others. Each account operates independently, but the brand needed a way to understand how each region was performing relative to the others.

The ask was a scoring and ranking system: pull metrics from the Sprout Social API, score each profile against its peers, and visualize the results in a way brand managers could actually read.

How the scoring works

Eight metrics are pulled per profile for a given date range: engagement rate, likes, comments, video views, and others. Each metric is scored on a 0–100 scale using a logarithmic, peer-relative formula — meaning a profile's score on any given metric is a function of how it compares to the other profiles in the dataset, not against any external benchmark.

The overall score (called rating) is a simple average of the eight per-metric scores. Profiles are then ranked by rating within the dataset.

Detail views include donut and bar charts breaking down per-metric scores, color-coded by rating band:

• 0–40: red
• 41–70: yellow
• 71–100: green

The main table shows the rating alongside raw metric values — the per-metric scores feed the charts rather than the table, so managers see actual numbers (likes, comments, views) alongside the overall score.

What the scores actually mean

This is worth being precise about. The scoring system is a relative benchmarking tool, not a performance grade. A score of 80 means this profile outperforms most of its peers in the current dataset — it does not mean the profile is objectively performing well at social media.

This has real consequences:

• A profile can score 85 while performing genuinely poorly, if every other profile in the dataset is worse
• The same profile can score 65 in one date range and 45 in another with no change in actual performance, simply because a peer had a strong month that shifted the distribution
• The logarithmic scaling compresses outliers (a profile with 10x the followers doesn't automatically dominate), but scores are still only meaningful relative to each other

For a global brand comparing regional accounts against each other, this framing is actually useful — it answers "which regions are lagging behind peers?" rather than "is any given region good at social media?" Those are different questions, and the tool answers the first one well.

What it doesn't do

The system has no guidance layer. Germany scoring 42 tells a brand manager that Germany is underperforming relative to peers — it doesn't tell Germany whether to post more video, improve response time, focus on engagement rate, or something else entirely.

The per-metric scores exist and are computed, they're just surfaced in charts rather than surfaced as actionable information. A follow-on feature that identified which specific metrics were pulling a profile's score down — and offered some context on what levers move those metrics — would make the tool significantly more useful as a management instrument rather than just a report.

Weights were explored (engagement rate weighted higher than likes, reflecting that engagement rate normalizes for audience size in a way raw likes don't) but ultimately removed in favor of a simple average. The case for weighting was real; the organizational simplicity of an unweighted average won out.

What I'd do differently

Surface the per-metric breakdown as actionable guidance, not just chart data. If a profile scores low because video views are dragging the average, that should be visible and labeled in a way that suggests a direction — not buried in a donut chart that requires interpretation. The data is already there; it's a presentation and product thinking problem more than a technical one.