Profit Driven Platform
Agency intelligence: know if you're delivering, and if it's working
What It Is
This is the app behind Profit Driven, the digital marketing agency where I lead product and delivery. Agencies run on scattered signals: the team talks in Slack, tasks live in Asana, results live in Google Analytics and Google Ads, and the client's actual expectations live in a proposal nobody reopens. The platform pulls those into one place, per client, and keeps score.
It answers three questions, in order of how much they matter:
- Are we delivering? Each client is enrolled in defined services with real cadences. If a content retainer promises four posts a month and one shipped, the platform says so, in those words.
- Is it working? As history accrues, delivery gets connected to the numbers: what we did, and what moved.
- What needs attention? The daily surface: which clients are quiet, which promises are slipping, what the team said this week that should have landed somewhere.
How It Thinks
The interesting design decision: analysis is anchored to the services each client is enrolled in, not to free-floating metric thresholds. The enrolled service is the expectation model. That means the platform is useful on day one, when there is no analytics history to mine, because checking activity against a promised cadence needs no history at all. The deeper insight, connecting work to outcomes, compounds over time on the same substrate.
Under the hood it stays cheap and honest: deterministic detectors find the candidates and produce the numbers; the AI narrates and judges plausibility. The model never invents a metric.
Where It Stands
The platform runs Profit Driven's real client accounts today. The team files knowledge from Slack conversations, delivery is tracked against enrolled services, and the analysis layer is being built out in the open, on live work.
The longer arc: Profit Driven's clients are media organizations and other marketing firms, which means our customers look exactly like the platform's target user. First we run our own agency on it. Then we offer it to agencies like ours.
Tech Stack
- Next.js + Supabase on Vercel
- Slack, Asana, GA4, and Google Ads as signal sources
- Claude for routing, narration, and judgment over deterministic detectors
Same thread as everything else I build: making the invisible visible. This time it's the health of client relationships.