Creator tooling · AI content QA
SignalScore
Know what to post next on LinkedIn
Imports your LinkedIn analytics, builds a baseline for your own account, and returns one experiment to run this week.
Business problem
LinkedIn's native analytics show impressions, reactions and comments, but never a baseline and never a next step. Operators and founder-creators post into a vacuum, can't tell what 'good' looks like for their own account, and each post feels like starting from scratch.
AI-enabled workflow
- 01Operator exports per-post analytics from LinkedIn (Post, then View analytics, then Export) and uploads the XLSX files. No LinkedIn login, no API access.
- 02Each post is tagged with a format (carousel, text, video) and an optional topic. Five posts is enough to start; more sharpens the signal.
- 03SignalScore builds a personal baseline across format, topic and cadence, so every new post is scored against your own history, not a global average.
- 04The app surfaces what held a post back and returns one falsifiable experiment for the next week: a specific format, hook or hour to test.
Prototype built
- Browser-only data pipeline. Analytics files are parsed locally and never leave the device.
- Three-screen flow: Import, Analysis, Insights, with zoomable screenshots in the marketing site.
- Local persistence so weekly experiments stack into a real learning loop.
- Privacy-by-design: no LinkedIn login, no API connection, delete-all from Settings.
Tools used
Commercial use case
Sold to solo creators, founder-led brands and B2B marketing teams that own LinkedIn as a channel. Replaces a layer of agency reporting with a self-serve, privacy-safe learning loop the operator runs themselves.
ROI / adoption potential
Roughly one high-confidence content experiment shipped per week, instead of zero.
Illustrative assumptions
- Most operators run zero structured content experiments because building a personal baseline alone takes a half-day each time.
- SignalScore collapses baseline and recommendation into one session.
- Assumes the operator posts 2 to 4 times per week and acts on at least the next recommendation.
Numbers above are illustrative, not measured outcomes. Included to make the commercial logic legible to a hiring conversation.