JointCommerce
I hired every team member, owned the $300K–$500K/mo P&L, and built a two-sided AdTech platform - a consumer ordering app (2K to 200K MAU in 11 months) plus advertiser activation, with a Customer Data Platform (CDP) as the connective tissue. Held 85% gross margins (nearly double the industry average) while my 15-person team managed 200+ active accounts at 85% client retention with zero long-term contracts.
Do it yourself. Train the team. Then automate it.
When I joined JointCommerce, there was nothing - no platform, no team, no processes. I hired every team member. Built every workflow. Wrote every line of code. I started by doing every operational role myself - running campaigns, entering data, managing accounts, closing deals - so I understood every process from the inside. Then I hired and trained a 15-person team across engineering, ad ops, sales, and data entry. Then I automated the operational workflows with a 950K-line Django platform so the team could focus on strategy and client relationships. The result: a consumer ordering app at app.jointcommerce.com that grew from 2,000 to 200,000 monthly active users in 11 months (100x), 200+ simultaneous advertiser accounts, 85% client retention with zero long-term contracts, and 85% gross margins where the industry averages 45%.
That's how you get 85% margins with a 15-person team. You can't automate what you don't understand from the inside - and you can't scale what you haven't built yourself.
Two sides. One CDP. One compounding engine.
A two-sided AdTech platform: the consumer ordering app (app.jointcommerce.com) on one side, advertiser activation on the other, with a Customer Data Platform (CDP) as the connective tissue feeding first-party purchase and behavioral data into programmatic. Click each segment to explore.
- Consumer-facing ordering app at app.jointcommerce.com - the source of first-party purchase and behavioral data
- Grew from 2,000 to 200,000 MAU in 11 months (100x)
- Built from scratch on Next.js, Django, PostgreSQL with Stripe checkout, multi-location inventory, loyalty programs, and delivery tracking
- 55K SEO articles + 13K directory pages drove 680% organic traffic growth and fed the top of the funnel
- Customer Data Platform (CDP) - the connective tissue between the consumer app and advertiser activation
- 21M+ unique consumer profiles and 1B+ behavioral data points built from first-party app activity
- Custom JavaScript pixel on the consumer app captures first-party behavioral signals across the full ad-impression → product-view → engagement chain
- 6 DSP integrations — including activation by a major DSP partner for pharmaceutical advertiser campaigns
- 200+ active advertiser accounts driving $300K–$500K/mo P&L at 85% gross margins (industry avg: 45%)
- Programmatic activation across 6 DSPs using purchase-intent audiences sourced from the CDP
- Shoppable ad engine generates interactive units from product feeds in under 2 hours
- On-platform display ad network with CPM billing, geo-targeting, and frequency capping
- Major DSP partner activated the CDP for pharmaceutical advertiser campaigns — first-party data monetized through media partnerships
950K lines. One engineer.
JointCommerce wasn't one product. It was six interconnected systems that gave a 15-person agency the data infrastructure of a company ten times its size.
Consumer Ordering App
app.jointcommerce.com - 2K to 200K MAU in 11 months (100x). 55K articles, 13K directory pages, e-commerce with Stripe, multi-location inventory, loyalty, delivery tracking.
Advertiser Activation
200+ accounts, $300K-$500K/mo P&L at 85% gross margins. Programmatic via 6 DSPs, on-platform display ads with CPM billing, shoppable ad generator, full creative pipeline.
Customer Data Platform (CDP)
21M+ unique consumer profiles, 1B+ behavioral data points, custom pixel on the consumer app, circuit breakers for sync resilience, audience segmentation. Activated by a major DSP partner for pharmaceutical campaigns.
Operations Layer
Reporting automation (60x faster), campaign management, client portal, 589 templates, and 4 Celery worker queues.
What it looked like
I created an ad format
- 9 industry presets, 7 templates, A/B testing
- Generates interactive units from product feeds in under 2 hours (was 3-day manual cycle)
- Adaptive DCO re-sorts products by user dwell time
- Deep-linking sends users directly to product pages
- 3x CTR vs standard display
20K to 156K monthly clicks
AI-assisted engineering at production scale
One engineer. 950K lines. The leverage came from pairing deep platform knowledge with AI-assisted development tools - and shipping production code with an AI stack baked in.
Built With AI Pair Programming
The platform was developed using Claude Code, Codex, Cursor, and Copilot. AI-assisted engineering let one operator-builder ship the equivalent of a 5-engineer team's output - on a 15-person company budget.
Production AI Stack
AI shipped in the product, not just the dev workflow: Vercel AI SDK, Anthropic and OpenAI APIs powering audience clustering, creative generation, campaign Q&A, and operational copilots inside the platform.