Autowelt

Generative Engine Optimization (GEO) & AI Referral Intelligence

An analytics platform for Autowelt to track visitor traffic, brand mentions, and car sales inquiries from ChatGPT, Perplexity, Gemini, and Claude.

Industry01

Automotive Retail

Scope02

BI Engineering, Dashboard Build

Timeline03

6-8 Weeks

Tools04

Next.js, TypeScript, Tailwind

The Problem

Car shoppers increasingly ask AI chatbots for car recommendations instead of search engines. Autowelt lacked visibility into these conversational leads.

Before

  • •Analytics Blindspots: Standard tools could not see visits from AI chatbots.
  • •Missing Brand Mentions: No way to know if chatbots recommended Autowelt cars.
  • •Lost Buyer Leads: Chatbots directed car shoppers to competitor listings.

Goals

  • •Track AI Traffic: Track visits and inquiries from ChatGPT, Perplexity, and Claude.
  • •Track Brand Mentions: Measure how often chatbots recommend Autowelt cars.
  • •Fast Dashboard: Deliver fast Next.js views with simple filters.

Requirements

Interviews with leadership, marketing, and development teams shaped core requirements:

Leadership Team

Sales Growth & ROI

Key Questions

"How many car sales inquiries come from AI chatbots, and is the traffic growing?"

Dashboard Requirement

High-level overview showing AI traffic share, sales inquiries, and engine comparisons.

Marketing Team

AI Traffic & Brand Mentions

Key Questions

"Which AI chatbots mention our cars, and which queries do we miss?"

Dashboard Requirement

Breakdown of traffic by AI engine, brand mentions, and search intent.

Operations & Developers

Fast Queries & Data Accuracy

Key Questions

"How do we parse crawler logs and show vehicle data without slowing down the page?"

Dashboard Requirement

Centralized BigQuery storage, fast Next.js endpoints, and instant filters.

Engineering & Build Process

A focused 6-8 week sprint from referral log audits to a working frontend application:

STAGE 01·

AI Referral Audit & Interviews

Analyzed web server logs, identified AI crawler user agents, and gathered team requirements.

Referral Schema & Metric Rules
STAGE 02·

Data Architecture & Structure

Organized reporting into 5 dedicated views covering engines, vehicle models, and intent.

Schema Mapping & Dashboard Architecture
STAGE 03·

Data Modeling & Viz

Engineered responsive citation tables, LLM search share charts, and modular metric cards.

Analytics Component Library
STAGE 04·

Frontend Engineering

Built a fast Next.js application with instant date range filtering and responsive layouts.

Working Next.js App
AI Overview

AI Overview

Tracks total AI search visits, conversion value, and platform growth benchmarks.

Key Decisions

Simple interface choices directly informed by team needs:

Information Structure01

Top KPI Scorecard

Problem

AI search numbers were mixed with general web traffic, hiding key trends.

Solution

Top cards with dedicated AI search share and assisted conversion value.

Data Visualization02

Visits by AI Engine

Problem

A single visitor total did not show which AI chatbots sent traffic.

Solution

Clear breakdown bars comparing visits from ChatGPT, Perplexity, Gemini, and Claude.

Automotive Attribution03

HSN/TSN Model Tracking

Problem

Marketing teams could not connect conversational prompts to specific inventory cars.

Solution

Catalog-level attribution linking generative queries to German vehicle specification codes.

Frontend Engineering04

Reusable Code Components

Problem

Static mockups broke when vehicle models or citation metrics updated.

Solution

Modular React components engineered to display dynamic data cleanly.

Referrers

Referrers

Compares referral visits and lead conversion rates across ChatGPT, Perplexity, Gemini, and Claude.

Content

Content

Tracks top cited landing pages, product URLs, and citation positions in AI responses.

Tech Stack

A clean, modern setup built for fast loading and responsive filtering:

Layer 01

Data Sources

Syncs daily ad spend, impressions, and sales across channels.

Marketing APIs
Layer 02

Data Warehouse

Stores ad records and matches them with Shopify store orders.

BigQuery Warehouse
Layer 03

Query Layer

Calculates daily marketing efficiency and retention curves on demand.

Next.js Server Actions
Layer 04

User Interface

Clean visual interface with responsive charts, dark mode, and video previews.

Next.js & Tailwind

Results & Impact

The new system gave Autowelt clear visibility into conversational search traffic:

Modular Analytics Architecture

Built a modular component architecture with reusable React charts and metric tables for rapid dashboard scaling.

Reusable Primitives

Automated Referral Pipeline

Connected server referral logs into an automated pipeline, eliminating manual data exports.

Continuous Data Sync

Single Source of Truth

Unified AI search and conversion definitions across marketing and leadership.

Standardized Metrics

Faster Marketing Decisions

Gave marketing teams instant visibility into engine citation ranks and car model demand.

Brand Citation Overview
Demographics

Demographics

Analyzes visitor country, device distribution, and conversion rates across regions.

Ongoing Maintenance & Usability

After delivering the dashboard, work shifted to fixing bugs, improving usability, adding new metrics, and adjusting existing calculations.

Continuous Support

Planning Additional Data Sources

Planning the addition of new crawler log feeds and emerging AI engine APIs to expand citation reporting coverage.

Data Expansion

Gallery

All dashboard views developed for the platform. Click any view to inspect full size: