AF
Andrew Fonseka Performance Marketing Specialist
Get in touch
Performance Marketing Specialist Vancouver, BC

I don't run campaigns. I build revenue systems.

Across financial services, e-commerce, and B2B — I find the real constraint, build the system that removes it, and measure everything back to revenue.

Flagship outcome
ROAS
$400K+
in new deposits from a single financial-services campaign.
Gold Effie Award · Financial Excellence
Impact at a glance
  • 22% lower CPA Aegon · 2025
  • 18% conversion lift Aegon · 2025
  • 70% faster lead response Aegon · 2025
  • 40% revenue growth Kites Global · 2023
Work

Three systems. Revenue, retention, and operational leverage.

Each study leads with the outcome, then opens into the diagnosis, the critical decisions, how it was measured, and what I would test next.

Financial services Paid media · CRO

Turning a gold-loan campaign into $400K+ in new deposits and 5x ROAS

A financial-services client needed deposit growth in a trust-sensitive, long-consideration category. I rebuilt the audience architecture and the post-click experience around proof — and tied campaign data back to qualified outcomes.

$400K+
new deposits
5x
ROAS
220%
conversion uplift
Gold
Effie Award
E-commerce · Retail Acquisition · Retention

Building an acquisition and retention system that grew revenue 40%

A pet-food retailer had plateaued on acquisition-only spend. I balanced prospecting with remarketing and lifecycle offers so revenue and repeat purchase moved together — not just lead volume.

40%
revenue growth
4x
ROAS
30%
repeat purchases
2.8×
returning customer sessions
Marketing automation n8n · Zapier · CRM

Eliminating 72 hours of weekly manual work through automated CRM synchronization

Lead routing was manual, slow, and error-prone. I engineered trigger-based synchronization between lead-generation channels and the central CRM — with validation, enrichment, and human review on exceptions.

72 hrs/wk
eliminated
70%
faster lead response
~3,744 hrs/year annualized
Faster lead operations · fewer handoff errors · more consistent data

Client names are anonymized, and sensitive figures rounded.

How I work

I don't begin with a channel. I begin with the constraint.

Most marketers show ads → results. I connect the system that produces growth — and know exactly where to intervene when it stalls.

Media stage

Paid Media Acquisition

Full-funnel paid search and social with custom audience segmentation across Google, Meta, LinkedIn, and TikTok Ads.

Proof
5x ROAS · 22% lower CPA
See the full case study →
The process

The same six steps, whether the problem is traffic, conversion, or retention.

01

Find the real constraint

Find the real bottleneck — traffic, creative, conversion, tracking, offer, or operations.

02

Rank by impact, not instinct

Rank hypotheses by impact, confidence, and ease before spending a dollar.

03

Wire measurement first

Wire GA4, GTM, and CRM so optimization targets qualified outcomes.

04

Run controlled experiments

Run controlled experiments; keep what wins, kill what doesn't, and record why.

05

Expand what compounds

Expand the winners without breaking the unit economics that made them work.

06

Remove the repeatable parts

Turn the repeatable parts into systems so the team compounds, not repeats.

The framework in action

Every growth problem lives at a specific stage in the funnel. Here's exactly where I found it and what I did.

Most campaigns fail at one stage and get fixed at another. I find the actual break point first.

Observed

Strong traffic, weak consultation bookings.

Diagnosis

Message mismatch and insufficient proof.

Action

Revised hierarchy, trust signals, conversion path (Clarity-informed).

Outcome

18% conversion improvement.

See the full case study →
Try it yourself

Before I change a dollar of spend, I run a diagnostic sequence. Here's exactly what that looks like.

This is the method, not a guess — the same diagnostic sequence I run on a live account before changing spend.

Diagnostic sequence
  1. 01 Validate tracking is firing correctly
  2. 02 Segment by audience, device, campaign, and landing page
  3. 03 Check ad-to-page message match
  4. 04 Review on-page behavior in Microsoft Clarity
  5. 05 Evaluate traffic quality by source
  6. 06 Isolate the highest-impact hypothesis
  7. 07 Run a controlled test
  8. 08 Check downstream lead quality in the CRM
Most likely cause

Post-click message mismatch, or rising low-intent traffic — rarely the auction.

Recommended first move

Align the landing message to the ad promise and add proof before scaling spend.

AI & automation

I don't automate for the sake of it. I automate what's repeatable so the team compounds.

The differentiation isn’t using AI — it’s using it responsibly, with a human in the loop, to remove repetitive marketing operations. Here’s exactly how one workflow runs end to end.

Campaign data Automated collection Anomaly detection AI diagnostic summary Human review Recommended action Approved implementation
80% faster
campaign anomaly detection
n8n + Zapier
orchestration layer
Human
approval on every exception
30%
marketing operations cost reduction
Andrew Fonseka
MBA · 4.00 GPA University Canada West LSE · Management & Digital Innovation
About

Most useful when growth has stalled, and the real constraint isn't obvious.

I’m Andrew Fonseka, a performance marketing specialist based in Vancouver, BC. I connect paid media, conversion strategy, measurement, and automation into complete growth systems — built to compound, not just convert.

Across financial services, e-commerce, retail, education, and B2B, I’ve delivered measurable revenue, lower acquisition costs, and operational leverage. I hold an MBA with a 4.00 GPA from University Canada West and a BSc in Management & Digital Innovation from the London School of Economics. I build the systems that turn marketing investment into growth — and keep producing it.

Get in touch

Let’s talk about what you’re building.

Currently exploring senior performance marketing roles at agencies building for growth. If that’s you, let’s talk.

Your details are never stored or shared.