CRM. Analytics dashboards. Automated data pipelines. Lead attribution. AI-powered insights. The conventional SaaS stack for all of that runs about $2,000/month, plus implementation costs.

We built it for $47/month. All in. We call it Lighthouse.

This article explains how we did it, what we learned, and why we believe it signals a major inflection point for content operations companies.

‍

The Challenge

At Coastalview Advisory, we needed to centralize eight data streams into a single platform: website analytics, LinkedIn performance across three accounts, ad spend, lead attribution, pipeline status, assessment funnels, qualitative intelligence, and AI-powered marketing analysis.

We also needed a CRM. But enterprise CRM pricing didn't match how an advisory firm actually sells. Advisory firms don't operate on high-volume pipelines. The work is high-value and relationship-driven. Salesforce at $300/month felt like renting a warehouse for a jewelry case. The conventional answer would have been HubSpot + Salesforce + Sprout Social + a dashboard tool + Zapier. We chose a different path.

‍

What We Built

Lighthouse is a marketing analytics platform with an integrated CRM, built on simple, accessible tools:

Google Sheets (16 tabs) serves as the central database for all marketing data. Every row is visible, editable, exportable, and portable.

Make (6 automated scenarios) pulls data daily from GA4, DepthCheck assessment events, and Webflow form submissions, plus a weekly AI insights trigger. Zero manual effort. A companion web app on Vercel provides nine modules for manual entry: lead tracking, LinkedIn analytics uploads, ad spend imports, pipeline management via drag-and-drop kanban, profile statistics, a UTM builder, qualitative notes, and Claude-powered AI insights.

Looker Studio delivers executive dashboards with KPI scorecards, trend charts, and filterable views by person, date range, and campaign.

Claude API powers AI-driven marketing analysis with full access to CVA's operational data, streaming structured insights on demand.

Total monthly cost: $47. ($29 for Make, $18 for Claude API. Everything else is free.)

‍

How It Was Built

The person who built Lighthouse has 20+ years in media, content operations, and consulting leadership. He has never written production code. He doesn't know JavaScript.

Lighthouse was built using Claude Code, an AI coding agent from Anthropic that takes instructions in plain English, and a free plugin called Superpowers that made the process far less intimidating. Instead of requiring technical specs, Superpowers walks you through what you're building conversationally: what matters most, what could go wrong, what the priorities are. Then it creates a plan. Then it builds.

Total build time: approximately 30 hours. Some longer sessions, but mostly an hour here and there between client work. Not a dedicated sprint. Work done in the margins of running a business.

‍

The CRM That Shouldn't Exist

The CRM was built directly into Lighthouse. Same data layer. Same interface. Every lead is tracked from source to close. The pipeline kanban provides drag-and-drop stage management. Full source attribution connects each lead back to the campaign or interaction that generated it.

This is arguably better architecture than the conventional approach. In a typical setup, analytics data lives in one platform, lead data in another, pipeline data in a third. Connecting them requires integrations and middleware. In Lighthouse, there is no integration because there is no separation.

‍

The AI Advantage SaaS Can't Match

Every SaaS platform in the MarTech stack is racing to bolt on AI features. Salesforce has Einstein. HubSpot has AI assistants. And almost universally, the experience is mediocre compared to going directly to Claude or ChatGPT.

Because CVA owns its data layer, we connected Lighthouse directly to Claude. Every piece of marketing data feeds into our project workspace in Claude with full operational context. When we ask for analysis, strategic recommendations, or pattern recognition, Claude has real business data. Not a generic chatbot embedded in a dashboard. Our actual AI working with our actual data.

When you own the data, you choose the AI. When you rent the platform, the vendor chooses it for you.

‍

Iteration Is Free

Traditional development locks you into specs. Every change requires a ticket, a scope conversation, and a timeline adjustment.

Twenty-three modifications were made to Lighthouse after the initial build. In a traditional engagement, that's 23 tickets and a budget overrun. With Claude Code, each change took under a minute. The twentieth version cost the same as the first.

‍

What This Means for Content Operations

If you run a content operation (in-house team, independent agency, media company, production studio), consider how many platforms you're renting for workflows you could own:

The CRM your pipeline doesn't justify, but your business needs the discipline of tracking every opportunity from source to close.

The production tracker bent into shape because you bought Asana or Monday for general project management, then spent months customizing it for content workflows it was never designed for.

The client approval workflow duct-taped across email, Slack, and a shared folder because no single tool does it the way your business actually works.

The freelancer rate cards and capacity tracking managed in someone's head or a spreadsheet nobody else can read, directly impacting margin visibility.

Every one of those can now be built as a custom tool, designed around exactly how your operation works. No engineering team. No vendor evaluation. No six-month implementation.

We've entered a post-SaaS era for small and mid-size content companies. Building custom tools that meet your exact needs, for a fraction of the cost, in a fraction of the time, is about to be the new normal. The only thing standing between your team and a custom-built toolset is knowing what to build and the willingness to try.

Download the Detailed White Paper

The full architecture breakdown, cost comparison, and build methodology in a downloadable PDF.

Download PDF