Case study · Head of Growth, 2024–2026 · Updated September 2026
Quadratic: building the growth function for an AI spreadsheet.
- ~300k
- users
- −60%
- blended CAC, quarter over quarter
- 7,000+
- organic signups per month
- 1M+
- monthly search impressions
The product and the job.
Quadratic is the AI spreadsheet. It connects to live data and lets AI write the formulas, Python, and SQL to analyze it, inside a spreadsheet the whole team can read and audit. The buyers split in two: finance and FP&A teams on one side, analysts and technical users who would rather write Python than nest another VLOOKUP on the other. They search differently, evaluate differently, and convert differently.
I was the growth function. Every channel, the budget, the agencies and contractors, the reporting, the launches, and the weekly signup number. The team was small, so I did most of the execution myself.
What I walked into.
- A category squeezed between spreadsheets everyone already has and BI tools everyone already pays for
- Two audiences with almost no overlap in the queries they type or the content they trust
- Paid acquisition that was expensive and hard to attribute
- Ten posts a month with no keyword architecture underneath them
- Five sources of truth and no single answer to "what produced the customer?"
What I owned.
Measurement. I built the reporting system first, because nothing else could be judged without it: product events from Mixpanel and Postgres joined with GA4, Search Console, the CRM, and ad spend into one weekly view of visitors to signups to activated to paid, by channel, with cost. It became the company's growth dashboard and the basis for every budget decision after it. How I build these →
Organic. Keyword architecture for the whole site before a word of content, with separate themes for each audience. A content operation that went from ten to 50–100 posts a month, AI-assisted for volume and human-reviewed for quality. A programmatic pipeline of roughly 100 steps turning product data into use-case and connection pages, with review gates before anything shipped. It produced 7,000+ organic signups a month and more than a million monthly search impressions. SEO + GEO →
Paid. Conversion tracking rebuilt around value, brand separated from non-brand, and budget reallocated weekly by cost per paying customer instead of clicks. Blended CAC fell 60% quarter over quarter. Paid search →
Creators. I built the creator channel from nothing into a repeatable program: sourcing, vetting, negotiation, briefing, and attribution for roughly 50 creators, mostly on YouTube and extending into short-form and newsletters. Every partnership was measured against a branded-search baseline captured before it ran, and the program as a whole delivered paying customers at a fraction of the cost of paid search. Creator marketing →
Launches and the rest. Campaigns for the Agent Sync and MCP releases. A rapid-response campaign when a competitor announced it was shutting down, built to catch their users while they were looking for a home. An automated agent surfacing Reddit and X threads so the team could show up while they were still live. Plus the website, the positioning, and whatever else the number needed that week.
What it added up to.
- User base grown to roughly 300,000
- Blended CAC down 60% quarter over quarter
- 7,000+ organic signups a month, 1M+ monthly search impressions
- Content output scaled five to ten times with human review on every piece
- A creator program of roughly 50 partnerships, acquiring paying customers at a fraction of paid search CAC
- A single weekly reporting view used across the company
Systems that outlived the engagement.
The reporting stack, the content engine, the programmatic landing-page pipeline, and the listening agent all kept running after I left. That's the test I hold growth work to: not whether the chart went up while I was there, but whether the machine keeps working when I'm not.