Mobile.de
Germany's largest vehicle marketplace
mobile.de ↗1.6M listings, 140M monthly visits. The filter system is exceptional if you know exactly what you want. The majority of buyers don't.
The Opportunity
Mobile.de's search is filter-driven and excellent for buyers who arrive with a specific model in mind. But studies of automotive consumer behavior suggest the majority of buyers (like me) who are still exploring, comparing segments, or simply asking “what should I get?” — have no structured way to express their intent. They hit 30+ filter options and either settle on a make/model they don't fully know, or leave.
US competitors CarGurus and Cars.com launched AI-powered conversational search in 2025. Google's AI Overviews are already intercepting automotive queries before users reach mobile.de. The window to own this experience in Germany is now.
Wireframe concept: what conversational search could look like for buyers who don't know the exact model they want. The existing filter bar remains while this sits alongside it.
Prioritization Summary
| Initiative | Impact | Complexity | Timeline | Priority |
|---|---|---|---|---|
| Smart Vehicle Comparison | High | Medium | 6–8 weeks | Start now |
| Conversational Search | Very High | Medium-High | 3–4 months | Start now |
| Personalized Search | High | High | 8–12 weeks | Plan now |
All three proposals share the same strategic logic: mobile.de's filter-driven search was built for users who know what they want. The next generation of search must also serve users who don't.
Conversational Search Layer
A prominent natural-language search input that sits alongside (not replacing) the existing filter system. The user types “reliable family SUV under €25K with low fuel costs” and the AI parses intent, maps it to vehicle attributes, and returns a curated set with an explanation of each match.
The key UX decision is additive, not replacement: power users keep their filters, undecided buyers get a new entry point. The AI result set can also generate the filter criteria so users can refine further if they want.
Target: 10–15% conversational search adoption within 6 months of launch.
Success Metrics
Key Risks
Query quality variance: Low-quality queries produce irrelevant results and erode trust faster than no feature. Requires robust fallback to standard filters when confidence is low.
Journey Comparison — Current vs. Proposed
Smart Vehicle Comparison
Today, users compare vehicles mentally or across browser tabs. There's no side-by-side comparison within the mobile.de search flow with no structured specs view, no running cost comparison, no AI-generated trade-off summary.
The proposal: users select 2–3 vehicles from search results and open a comparison panel. Specs, price evaluation, running cost estimates (insurance, fuel, tax), and an AI-generated summary: “The Golf is €3K cheaper but the Octavia has more boot space and lower insurance costs.”
Medium complexity, 6–8 weeks to MVP — frontend-dominant, reuses existing listing data.
Success Metrics
Key Risks
Running cost accuracy: Insurance and fuel estimates depend on external data. Inaccurate estimates damage trust. MVP should clearly label estimates as approximate.
Before / After — Comparison Experience
Personalized Search Experience
140M+ monthly visits generate enormous behavioral data that currently goes unused. Whether a user has been searching compact EVs for three weeks or is visiting for the first time, the search experience is identical.
Returning users should see results influenced by their browsing patterns. A “Recommended for you” section driven by collaborative filtering like “Users who viewed the VW ID.3 also looked at the Cupra Born and MG4” which creates cross-brand discovery that benefits both buyers and dealers.
GDPR is non-negotiable. All personalization on first-party data with explicit consent.
Success Metrics
Key Risks
Infrastructure complexity: Real-time personalization requires ML pipeline investment. Phase 1 can use simpler rule-based re-ranking on saved searches and recent views before full ML deployment.
Personalization Signal Flow
Hypotheses to Test
Hypothesis
Undecided buyer segment size
Test Method
Analyze funnel drop-off: % of visits that never apply a filter beyond make/model
Expected Outcome
Quantify the addressable opportunity for conversational search
Hypothesis
Conversational search conversion lift
Test Method
A/B test with 5% of traffic: natural language queries vs. filter-based searches
Expected Outcome
Do Natural Language queries generate higher-intent leads?
Hypothesis
Comparison as conversion trigger
Test Method
Track sessions with/without comparison feature: time-to-lead differential
Expected Outcome
Does structured comparison reduce time-to-lead?
Hypothesis
Cross-brand discovery value
Test Method
Attributed lead tracking on comparison-suggested vehicles
Expected Outcome
Do dealers see incremental leads from cross-brand suggestions?
North Star Metric
Primary Metric
Search-to-Lead Conversion Rate
The percentage of search sessions that result in a dealer contact (call, message, or financing inquiry). Captures search relevance, result quality, and buyer confidence in a single number AND maps directly to the dealer revenue that funds the platform.
Methodology — Analysis based on firsthand usage documentation (one week, multiple buying scenarios), public data (Adevinta annual reports, press releases, AIM Group reporting, app store reviews), competitive research (AutoScout24, CarGurus, Cars.com), and industry trend analysis. No internal data used. All opinions are my own.