Full-funnel growth for B2B and fintech teams
Acquisition, activation, retention and revenue programs across North America, Europe and Southeast Asia, run on GA4, Mixpanel, HubSpot and Salesforce with a fixed experiment cadence.
Diagnose first,
automate last
01 · Diagnose the system
Before I change anything I read the whole funnel as one system: where intent enters, how it routes, where it dies. I rank the leaks by what each one costs, so the first fix is an economic call.
I pull GA4 funnel reads and Mixpanel cohorts to find the drop-offs, then watch Hotjar recordings at each one to see why. I also audit how leads route through HubSpot and Salesforce, and check the tracking itself, server-side included, because a broken pixel looks exactly like a broken funnel. Everything lands in one leak map, ranked by cost.
Cost every leak before proposing any fix02 · Find the bottleneck
Every funnel I have worked on had one constraint capping everything downstream, and most roadmaps ignore it in favor of whatever is easiest to test. I isolate that constraint and size the prize before anyone touches a variant.
I compare stage-by-stage conversion in Looker Studio across regions, because a leak that shows up in NA and SEA alike is structural. Then I put a number on the constraint: what fixing slow handoff or misrouted traffic is worth in pipeline. I write a one-page brief naming the constraint and what it is worth.
Saying no to twelve plausible experiments is the job03 · Run experiments with a process
I run CRO, activation and pricing tests on a fixed cadence, with the success metric registered before launch so nobody re-narrates a loss into a win. I want a pipeline of small certainties that compound.
Tests run through VWO, Optimizely or AB Tasty depending on the stack, each with a pre-registered metric and a fixed decision date. Every result, including the kills, goes into an experiment log finance can audit. That log is how LinkedIn CPL went from $300+ to ~$160 and Google registrations from $90+ to ~$35. The account work did half of it: restructured campaigns, refined audiences, re-scored keywords, rebuilt landing pages. The cadence did the other half, because before this a lead came off a LinkedIn campaign and went straight to a rep with nothing automated in between.
Pre-register the metric or the result is a story04 · Automate & scale
A winning test that still needs me to run it is only a draft. The last step turns winners into infrastructure: lifecycle triggers, routing rules, AI-assisted enrichment.
Winners become HubSpot and Salesforce workflows, wired together with Zapier, Make and Dify, with OpenAI and Claude APIs handling enrichment inside the loop. I shipped 200+ workflows off that discipline and cut lead handoff from ~12 hours to 1–2 minutes. I hand over the running system and a runbook, so the team can change it without me.
If it can't run without me, it isn't doneFull-funnel programs I have owned end to end:
Funnel architecture
Experimentation
Activation & retention
Scale & partnership
What I diagnose and test on:
the analytics and experiment stack
diagnosis
GA4
Looker Studio
Mixpanel
Hotjar
⌁Server-side tracking
VWO
Optimizely
AB Tasty
& CRM
HubSpot
Salesforce
Zapier
Make
Dify
the loop
OpenAI API
Claude API
Deepseek · Qwen
One cross-border payments line, two years, all four steps
Steps 01–03 cut LinkedIn CPL from $300+ to ~$160 and Google registrations from $90+ to ~$35. Step 04 collapsed lead handoff from ~12 hours to 1–2 minutes. Revenue on the line grew from $2M to $76M in year one, past $250M+ in year two.
Read the full case ›His work is thorough and impactful. He has created amazing opportunities for sales with his marketing campaigns and is a huge asset to the whole company.
Tijana SijanicDirector of Human Resources
Need this discipline on your funnel?
The constraint first. Everything else is downstream of finding it.