Your product works
but your numbers don't?
Let's figure out why.

Situation

Enterprise users were failing to get useful data — not because the data wasn't there, but because building Boolean search queries required technical fluency most users didn't have. Low adoption was being misread as a UX problem.

Diagnosis

The barrier wasn't the interface. It was the query language itself. Users couldn't translate business questions into Boolean logic, so they leaned on CX support or gave up. The product was only accessible to power users.

What I built

CA-Pilot — an LLM-powered assistant that converts plain-English questions into structured Boolean queries. Also established a structured AI validation protocol with QA and Data Science, cutting iteration cycles significantly.

Outcome

15% improvement in search success rate. AI delivery speed increased 45% through the validation protocol. Reduced CX dependency for query-building across the enterprise client base.

Situation

A home services marketplace in Saudi Arabia was bleeding on two fronts: high cancellation rates and weak repeat usage. Both were being treated as operations problems — more calls, more manual intervention.

Diagnosis

Cancellations weren't random. Behavioral data showed predictable signals before a cancellation occurred — but no one was acting on them upstream. On retention, there was no structured reason for customers to come back.

What I built

An AI-based cancellation risk scoring model that flagged high-probability cancellations early, giving Ops a prioritized intervention list. A loyalty system mapped user segments, defined activation triggers, and modelled retention incentives.

Outcome

25% reduction in cancellations within one quarter. Loyalty system drove 30% activation in Q1 and 45% retention uplift for the engaged segment. Overall retention improved 15%.

Situation

A B2B marketplace in Egypt had a BNPL vertical with low activation and a demand side growing slowly despite an active growth team. The roadmap was feature-heavy but the real blockers were in the funnel and the credit logic.

Diagnosis

BNPL activation was failing at the qualification stage — too much friction and scoring logic that wasn't surfacing the right leads. On the demand side, growth was driven by promotions without structural retention, so profit leaked through discount dependency.

What I built

BNPL credit scorecards with the data science team, plus a redesigned onboarding and lead validation flow. On the demand side, ran structured growth experiments — loyalty programs, offer segmentation — to replace discount-driven growth with behavior-driven retention.

Outcome

40% increase in BNPL activation. 35% profit growth and 25% retention improvement across the demand vertical. 80% reduction in manual sales reporting via a Sales Management System built in parallel.