DropHog
Date
May 2026
Role
AI Product Designer and Engineering (Solo role)
Company
DropHog
Product Engineering · UX design · Information architecture · MVP delivery
Key Outcome
How DropHog closes the Africa-specific analytics gap
DropHog is an AI-powered onboarding funnel diagnostic tool built exclusively for product managers, growth teams, and UX designers at African digital companies. It diagnoses why users drop off during onboarding and prescribes specific, market-aware fixes without requiring existing analytics instrumentation.
Background
The analytics tools were built for the wrong market.
Most analytics tools like Mixpanel, Amplitude, and FullStory were designed for Western markets with fast internet, modern devices, and digitally native users. African product teams using these tools get generic UX advice that ignores the real constraints of their markets.
A fintech PM in Lagos knows their BVN verification step loses 35% of users, according to Mixpanel. What it doesn't tell them is that the NIBSS API times out on 3G connections, that first-time digital users abandon when they don't understand what BVN is; and that the specific fix is progressive disclosure with an offline retry mechanism, not a redesign.
The closest workflow teams use today is Mixpanel + a prompt in ChatGPT: manually exporting data, writing context, and hoping for relevant advice. DropHog replaces that entire workflow in one tool.
The Problem
Generic analytics tools were built for Western markets but African product teams are using them to solve a fundamentally different problem.
Activation rates across African digital products average 30–50%, compared to 70–85% in Western equivalents. The gap is largely explained by market-specific friction that generic tools can't diagnose: infrastructure latency, low-end device constraints, connectivity drop-offs, and first-time digital user behaviour.
Product teams building in Africa face a critical workflow gap: they can see where users drop off, but they can't get contextually relevant explanations or fixes without manually assembling context outside their tools.
The Solution
A four-step diagnostic workflow built for African contexts
DropHog combines market-specific drop-off simulation with AI diagnosis that understands African digital infrastructure realities.
The product is structured around a four-step wizard:
Choose a flow template
Edit onboarding steps
Set market context
Receive an AI-generated diagnosis with ranked, actionable recommendations.
Product Architecture
Secured, serverless AI with persistent diagnostic history
All AI API calls go through Vercel serverless functions; the Groq API key is never exposed to the browser bundle. Clerk handles authentication with email and Google OAuth. Diagnostic history is persisted per user in Supabase, with queries filtered by Clerk user ID.
Diagnostic history view — saved past reports per user with re-access and sharing
Key design decisions
Decision 01 — Simulation as default, not fallback
Most tools require existing instrumentation. DropHog was built for teams without making simulation the primary input method, not a workaround.
Decision 02 — Africa-anchored prompt engineering
Anchoring prompts with infrastructure-specific terms (NIBSS, BVN, MTN OTP, M-Pesa) produced contextually accurate output vs generic fintech framing.
Decision 03 — Serverless AI architecture
All Groq API calls are routed through Vercel serverless functions; the API key is never exposed to the browser. Clerk handles auth. Supabase persists history per user.
Decision 04 — Shareable diagnostic reports Diagnostics are saved per session and exportable so findings can be shared with product leads, engineers, or investors without re-running the tool.
Outcomes & Impact
MVP delivering a complete diagnostic workflow
DropHog v1.0 shipped as a complete, deployed MVP covering the full user journey from landing page through shareable diagnostic results. The core thesis that market-aware simulation plus African-context AI can replace the Mixpanel + ChatGPT manual workflow is validated end-to-end.
Reflections & What Comes Next
What I learned and where this goes
The most important design constraint in DropHog was removing the prerequisite of existing analytics data. Most tools assume instrumentation is already in place, but the teams who most need activation help are often the ones who haven't yet set up Mixpanel or Amplitude. Making simulation the default, not the fallback, was the decision that made the tool genuinely useful.
The AI prompt engineering required significant iteration to produce consistently structured, African-context-aware output. Anchoring the prompt with specific infrastructure terms NIBSS, BVN, MTN OTP, and Safaricom M-Pesa produced noticeably more relevant recommendations than generic fintech framing.
Check Out the Live Product
