RankStacky

Inspiration Search behavior is changing fast. People are no longer only searching on Google; they are asking ChatGPT, Gemini, Perplexity, Claude, and other AI assistants which products, services, and brands to choose. Traditional SEO tools show how a website performs in search engines, but they do not show how AI systems perceive, rank, cite, or recommend a brand.

RankStacky was built to solve that gap: helping businesses understand and improve their visibility in AI-powered search.

What it does RankStacky tracks how a brand appears across AI search platforms and regions. Users can create workspaces, projects, tracking groups, prompts, competitors, and scheduled runs. It shows: Brand ranking by provider Competitor rankings Regional GEO testing Citations and sources Semantic topics Sentiment Brand perception Brand favorability Brand tone Share of Voice Recommendation rate Week-over-week visibility changes SEO audit history and technical SEO findings

How we built it RankStacky was built as a full-stack SaaS platform with a Next.js frontend, .NET backend, PostgreSQL databases, and a Python AI service.

The backend uses EF Core and separates user management, project data, SEO audit data, and AI usage tracking. Firebase Google login handles authentication, while backend APIs are protected with bearer-token sessions. Provider API keys are encrypted before storage and decrypted only when needed for execution.

The AI service runs prompts across providers such as Gemini, OpenAI, Perplexity, Claude, DeepSeek, Qwen, Mistral, and Kimi when users have added the relevant keys. Hangfire handles scheduled prompt runs.

Challenges we ran into The hardest challenge was making AI results consistent enough to compare. AI responses vary by provider, model, region, and time, so RankStacky needed normalization for rankings, competitors, citations, semantic topics, sentiment, and brand perception.

Another challenge was product structure. The flow had to be clear: Workspace → Project → Tracking Group → Prompt → Provider Run → History and Analytics We also had to ensure users only see their own data, while admins can manage plans, limits, models, and user-level administration without mixing private project data.

Accomplishments that we're proud of We built a working multi-tenant AI visibility platform with: Google/Firebase login Protected backend APIs Encrypted provider keys Project-based data ownership Multi-provider AI runs Regional prompt testing Competitor tracking Prompt scheduling with Hangfire SEO audits with crawl history Admin model and plan limit management Historical analytics and week-over-week tracking The strongest part is that RankStacky does not just run prompts once. It stores the results over time so users can see whether their AI visibility is improving or declining.

What we learned We learned that AI visibility is not just about being mentioned. A brand also needs to understand how it is described, whether the AI recommends it, which competitors appear nearby, what sources are cited, and how results change across regions.

We also learned that GEO and AI search analytics need a different product model than traditional SEO. Prompts, providers, regions, citations, semantic meaning, and historical comparison all matter.

What's next for RankStacky Next, RankStacky will expand deeper into AI search intelligence: More provider support Better charts for provider and region comparison Stronger recommendation and sentiment scoring Automated competitor discovery More advanced SEO audit reporting Team and organisation-level permissions Billing-based plan limits AI-powered suggestions for improving visibility More reliable long-term trend analysis across countries, languages, and models