Rohan Gawankar
Agency PM | Systems Builder | Fast Growth Specialist
Ready for work
Rohan is ready for their next project!
The Challenge D2C skincare brand was stuck at £15K/month ad spend with inconsistent returns. Campaigns lacked structure, creative testing was ad-hoc, and scaling attempts crashed ROAS. No clear funnel segmentation between prospecting and retargeting.
What I Did → Restructured accounts with TOF/MOF/BOF campaign architecture → Built creative testing framework (3 concepts × 4 formats weekly) → Launched TikTok Spark Ads driving 35% lower CPAs than Meta → Implemented bid strategies aligned to target CPA by funnel stage → Created weekly reporting cadence with spend/ROAS pacing alerts
Results 5x spend scaling (£15K → £80K/month) in 4 months 4.2x blended ROAS maintained at scale 35% lower CPA on TikTok vs Meta prospecting 62% increase in new customer acquisition £1.2M attributed revenue in 6 months
23
146
Challenge An e-commerce brand running paid ads across Google, Meta, TikTok, and Microsoft had fragmented tracking. Conversions were double-counted, attribution windows misaligned, and iOS 14+ changes left ~30% of purchases untracked.
Solution Rebuilt their measurement stack: configured GA4 with enhanced e-commerce and custom events, deployed server-side GTM for first-party data resilience, implemented Conversions API for Meta and TikTok to bypass ad blockers, set up conversion imports from GA4 into Google and Microsoft Ads, and built a unified attribution dashboard in Looker Studio.
Impact 40% improvement in tracked conversion visibility 95% match rate on server-side event deduplication Unified attribution across all 5 platforms 30% CPA reduction through accurate optimization signals
24
181
Built graph-based identity resolution engine processing millions of records with 95% match accuracy
The Challenge
Breach data investigations required manually connecting scattered identity fragments across millions of records. Analysts spent hours tracing relationships between emails, phones, usernames, and IDs with high false positive rates from common values like "123456" passwords.
What I Did
→ Designed multi-signal matching with weighted confidence
→ Built recursive graph traversal with configurable depth and breadth limits
→ Implemented dynamic supernode detection to filter high-frequency noise
→ Integrated Elasticsearch + Neo4j for search and visualization
Results
95% confidence accuracy on 4-level recursive search, Sub-30 second query response on 1M+ records
12
99
The Challenge A 4-location US restaurant chain with 42K customers was bleeding money on acquisition. 80% of budget went to ads, yet only 17% of diners returned. Over half their base had gone dormant.
Goal: Build loyalty without increasing ad spend.
What I Did
→ Segmented 42K customers by lifecycle and behavior
→ Designed behavior-triggered email/SMS campaigns → Built automated workflows integrated with CRM
→ Created "nudge" sequences to reactivate dormant users → Set up attribution tracking for revenue impact
Results 82% increase in repeat purchases (17% → 31%) £33K additional revenue in 2 months 314 bookings attributed to campaigns 14-point lift across all 4 locations ROI positive in under 8 weeks
23
162