An on-demand alcohol delivery app with web, iOS, and Android surfaces. This is how iExcel built the growth analytics and customer acquisition operating system underneath it.
Thirstie is an on-demand alcohol delivery platform — web, iOS, and Android — that connects 21+ consumers with local partner stores and routes orders through the last-mile fulfillment loop. The product side works. Users can browse, cart, check out, and get a bottle at the door.
What wasn't yet running was the layer underneath the product: a customer acquisition and growth operating system that could actually tell the business what was working. Where were downloads coming from. Which cohort activated. Which channel produced the repeat order. What the funnel looked like from first tap to second purchase. That layer was the assignment.
When iExcel picked up the engagement, the ask wasn't performance media. It was foundation — the analytics, the KPI framework, the Mixpanel taxonomy, the funnel reporting, the experimentation roadmap. Track it first. Then scale it.
Before events, before dashboards, before a single Mixpanel property, we wrote the KPI definition doc. What are the fields on a user record. What are the fields on an order record. What does a weekly reporting cadence actually need to show leadership on a Monday morning. The doc came first — because the taxonomy that follows only holds up if the definitions are locked.
The centerpiece of the engagement: a full Events & Property List workbook — the Mixpanel implementation spec. Every user action, session, order state, and funnel transition mapped into a structured event with typed properties, example values, and People Analytics fields. Handed to engineering as a build sheet. Web, iOS, and Android reading from the same taxonomy so cross-surface cohorts don't lie.
The Growth Hacker roadmap sat on top of the taxonomy — an experimentation ladder written so a small team could actually run it. Each rung tied to events already in the Mixpanel spec, so the moment a test shipped, the reporting was there waiting for it.
Download attribution wired per surface. Campaign tracking on paid and organic. Bartender and influencer referral tracking. SEO for high-intent product and recipe queries. Landing page A/B tests wired to signup events.
Download-to-signup and signup-to-first-order funnels defined. Onboarding sequence mapped. Email signup capture. Push opt-in prompts scoped. Recipe and how-to suggestions as an activation nudge for users who bounce pre-cart.
Cart abandonment triggers per surface. Exit surveys on abandonment. Alerts on drop-off spikes. Landing page A/B testing framework and roadmap. Conversion-blocking friction points logged against event data — not gut feel.
Repeat-order tracking on the user record. Segmentation by order count, category, and geography. Push notification lifecycle sequences. Drip campaigns tied to lifecycle stage. Attrition monitoring — churn flagged before it fires.
Referral events wired end-to-end. Bartender and influencer partner tracking with named UTM sources. Social proof surfaces on recipe and product pages. Lead gen + nurture roadmap for pre-signup traffic that isn't ready to convert.
With the taxonomy live and the KPI framework locked, the funnel reporting layer wrote itself. Each stage from ad impression to repeat order became a Mixpanel funnel with typed drop-off events underneath it. Cohort by surface. Cohort by acquisition source. Cohort by geography and store.
Attribution ran alongside — download source, session UTM, promo code, referral partner — landing on the user record so a signup could be traced back to the campaign that produced it. No more "we think Instagram is working." Now the funnel showed which Instagram campaign, which creative, which geo, which day-part, and which cohort came back for a second bottle.
The result is a reporting spine leadership can actually read in under five minutes on a Monday — and a growth team that can point at the exact drop-off step they're shipping against this week.
If you're running an on-demand or app-first B2C business — alcohol delivery or otherwise — and you're staring at three platforms firing events into three different tools, a growth team guessing at attribution, and a leadership dashboard nobody trusts, the Thirstie engagement is the shape of what to do about it.
KPI framework first. Event taxonomy second. Funnel and attribution after. Experimentation roadmap only when the layer underneath it can actually catch what the tests emit.
Most agencies want to run the paid. iExcel built the instrumentation so the paid — and the CRO, and the lifecycle, and the referral loop — could all be measured against the same source of truth. Track it first. Then scale it.
A KPI doc leadership can read on a Monday. An event taxonomy engineering can build against. A funnel that shows the drop-off, not a hunch. Book a 15-minute sanity check on your analytics stack — we'll tell you what's under-built and what to do about it.
Book a 15-min sanity check →