Your Data Is Talking. Are You Listening?
Most analytics platforms give you dashboards full of numbers. But numbers alone do not drive decisions. The gap between data and action is where growth opportunities die.
Dashboard Overload
Analytics dashboards are full of metrics, but teams spend hours staring at charts without knowing what action to take. Data without direction is just noise.
Web2 and Web3 Data Silos
Website analytics and blockchain data live in completely separate systems. The critical insights that emerge from connecting wallet behavior to browsing activity fall through the cracks.
Manual Analysis Bottleneck
Teams spend hours every week manually cross-referencing metrics, building reports, and trying to spot patterns. That is time not spent building product, acquiring users, or optimizing campaigns.
Blind Spots Kill Growth
Anomalies in traffic patterns, churn signals from declining wallet activity, and campaign underperformance go undetected until the damage is done. By the time you notice, users have already left.
How AnalyticKit AI Works
A fully automated pipeline that runs daily, transforming raw data from two ecosystems into clear, actionable recommendations.
Collect Web2 Data
Page views, unique visitors, sessions, traffic sources, device and browser data, UTM campaign parameters, geographic distribution, and referral paths are aggregated from ClickHouse in real time.
Collect Web3 Data
Active wallets, transaction volumes, transaction success rates, gas cost metrics, token transfer data, and wallet segmentation are pulled from blockchain records stored in PostgreSQL.
Cross-Link Identities
Wallet addresses are normalized and matched to website sessions, creating unified Web2+Web3 user profiles. This reveals which website visitors become on-chain users and how campaigns drive real blockchain activity.
Compute Trends
The system calculates 7-day and 30-day rolling averages from historical snapshots stored over time. It detects week-over-week changes, identifies anomalies, and computes conversion funnels across the Web2-to-Web3 journey.
AI Analysis
The aggregated summary is sent to GPT-4o-mini within a managed 12,000-token budget. The AI evaluates cross-ecosystem patterns and generates 3 to 7 prioritized recommendations, each with a category, severity level, and specific suggested action.
Deliver Insights
Recommendations are delivered via the dashboard and REST API, categorized by type and sorted by severity. Each includes a clear title, detailed analysis, a suggested action, and references to the specific metrics that triggered it.
What You Get Every Morning
Every day at 6 AM UTC, AnalyticKit’s AI agent completes its analysis and delivers fresh recommendations. Here is what real recommendations look like.
Campaign Attribution
Twitter Campaign Conversion Rate Dropped 40% Below Average
Your Twitter campaign ‘Q1_DeFi_Launch’ drove 47 wallet connections this week, but only 3 completed a swap transaction. The current 6.4% conversion rate is 40% below your 30-day rolling average of 10.7%. Cross-linking data shows that 68% of these users visited the swap page but abandoned before confirming the transaction.
Suggested Action
Review the swap confirmation page UX for friction points. Consider A/B testing a simplified flow. Also evaluate whether the Twitter campaign targeting parameters are attracting users with genuine DeFi intent versus casual browsers.
Churn Risk
23 Previously Active Wallets Show Inactivity Pattern
23 wallets that maintained weekly transaction activity for the past 30 days have not completed a transaction in 12 or more days. These wallets represent a combined $14,200 in historical transaction volume. Web2 data shows that 17 of these wallets also stopped visiting the site entirely, while 6 still browse but no longer transact.
Suggested Action
Segment these wallets into two groups: fully disengaged (no site visits) and browse-but-do-not-transact. Launch a targeted re-engagement campaign for the first group. For the second group, investigate potential UX barriers preventing transactions, possibly related to gas costs or slippage settings.
Growth Opportunity
Mobile Traffic From India Surged 340% With High Wallet Connection Rate
Mobile traffic from India increased 340% week-over-week, jumping from 47 to 207 sessions. Notably, 18% of these visitors connected wallets, compared to your global average of 8%. Cross-linked data reveals these users came primarily through organic search and two Telegram referral links. The wallet connection-to-transaction rate for this cohort is 22%, above your platform average of 15%.
Suggested Action
Consider localizing key landing pages and onboarding flows for the Indian market. Invest in the Telegram channels driving this traffic. The high wallet connection and transaction rates suggest strong intent from this audience segment, making it a high-ROI growth opportunity.
Six Categories of Intelligence
Every recommendation is categorized into one of six intelligence domains. Each category addresses a distinct aspect of your Web2 and Web3 operations, ensuring comprehensive coverage of your entire growth stack.
Engagement Intelligence
Surface patterns in user behavior across your entire funnel. Know which pages convert best, which user flows cause drop-offs, and how engagement trends are shifting week over week. The AI identifies behavioral segments and highlights changes in how users interact with your platform before and after connecting their wallets.
Transaction Health
Monitor the health of on-chain activity tied to your platform. Get alerted to declining transaction success rates, unusual gas cost patterns, shifts in token transfer volumes, or changes in the ratio of new versus returning wallets. Catch issues with smart contract interactions before they erode user trust.
Campaign Attribution
Automatically connect marketing spend to on-chain results by tracing the complete user journey. Know which campaigns drive wallet connections AND which ones drive actual transactions. Measure true ROI by attributing blockchain activity back to specific UTM campaigns, referral sources, and landing pages.
Churn Risk Detection
Identify at-risk users before they leave your ecosystem entirely. The AI flags wallets showing declining transaction frequency, reduced site visits, or both. It distinguishes between users who have stopped visiting and those who browse but no longer transact, enabling targeted re-engagement strategies for each segment.
Growth Opportunities
Spot untapped potential buried in your data that manual analysis would miss. New geographic markets showing unusually high engagement, underperforming campaigns that have optimization potential, user segments ready for upselling, and traffic sources with above-average conversion rates are all surfaced automatically.
Anomaly Detection
Catch unusual patterns the moment they emerge. Sudden traffic spikes, unexpected drops in transaction success rates, sharp changes in conversion funnels, and irregular wallet activity patterns are all flagged immediately. Get alerted before anomalies become full-blown problems, giving you time to investigate and respond.
Why This Matters for Web3
Existing analytics tools were built for one world or the other. AnalyticKit bridges both, and the AI recommendations are what make that bridge actionable.
Traditional Web Analytics
Google Analytics and Mixpanel track page views and clicks, but they have no concept of wallets, transactions, or on-chain behavior. They cannot tell you which website visitors became paying blockchain users.
On-Chain Analytics
Dune Analytics and Nansen excel at blockchain data, but they cannot see what happened before the wallet connected. They miss the entire Web2 journey: which campaign brought the user, what pages they viewed, and where they hesitated.
AnalyticKit
The only platform that unifies Web2 and Web3 data into a single view AND applies AI analysis to the combined dataset. The cross-linking step is what makes our AI recommendations uniquely valuable and impossible to replicate with other tools.
The cross-linked identity layer is the foundation. Without it, AI recommendations are limited to surface-level insights from a single data source. With it, AnalyticKit’s AI can tell you not just what is happening, but why it is happening across the complete user journey from first click to on-chain transaction.
Built for Security From Day One
Your data is sensitive. We designed the AI recommendations pipeline with security and cost control as first-class requirements, not afterthoughts.
No Raw Data Sent to AI
The AI never sees individual user records, wallet addresses, or session-level data. Only pre-aggregated statistical summaries are sent for analysis. Individual identities remain inside your AnalyticKit instance at all times.
Team-Scoped Isolation
All data is strictly isolated by team. There is zero cross-team data access. Every API endpoint requires authentication, and recommendations are scoped to the team that generated them. No data leaks between tenants.
Circuit Breaker Protection
Built-in circuit breaker logic prevents excessive AI API usage. On-demand generation is limited to 3 requests per day per team. The entire AI analysis pipeline costs approximately $0.30 per team per month for daily automated analysis.
API-First Design
Every feature in the AI recommendations system is accessible through a clean REST API. Build custom dashboards, integrate with your existing tools, or automate your response workflow.
| Method | Endpoint | Description |
|---|---|---|
| GET | /api/recommendations/ | List all recommendations with filtering and pagination |
| GET | /api/recommendations/latest/ | Get today’s freshest batch of AI recommendations |
| POST | /api/recommendations/{id}/acted/ | Mark a recommendation as actioned to track follow-through |
| GET | /api/recommendations/summary/ | Get a 7-day overview of all recommendations and actions |
| POST | /api/recommendations/generate/ | Trigger on-demand AI generation (up to 3 times per day) |
| GET | /api/aggregation-snapshots/ | Access historical data snapshots used by the AI engine |
AI Recommendations Are Included in All Plans
Every AnalyticKit plan includes the full AI-powered recommendations engine. No add-ons, no premium tiers, no surprises. Your AI agent starts analyzing data from day one.
Frequently Asked Questions
How does the AI analyze my data?
AnalyticKit runs a fully automated pipeline on a daily schedule. First, it aggregates your Web2 data (page views, sessions, traffic sources, geographic data) from ClickHouse and your Web3 data (wallet activity, transactions, gas metrics) from PostgreSQL. It then cross-links these datasets by matching wallet addresses to website sessions. The system computes 7-day and 30-day rolling averages to identify trends. Finally, only the pre-aggregated statistical summary is sent to the AI model, which generates 3 to 7 actionable recommendations. No raw user data, individual wallet addresses, or session-level records ever leave your instance.
How much does the AI feature cost?
The AI recommendations engine is included in every AnalyticKit plan at no additional cost. The underlying AI processing costs approximately $0.30 per team per month for daily automated analysis using GPT-4o-mini. This cost is absorbed by AnalyticKit and is not passed on to customers. There are no premium tiers or add-on fees for AI features.
How often are recommendations generated?
The automated Celery task runs every day at 6:00 AM UTC, analyzing the previous day’s data alongside historical trends and generating a fresh batch of 3 to 7 recommendations. In addition, you can trigger on-demand generation through the API or dashboard up to 3 times per day if you need fresher insights after a campaign launch, a product update, or any significant event.
What AI model is used?
AnalyticKit currently uses OpenAI’s GPT-4o-mini model, chosen for its excellent balance of analytical capability and cost efficiency. The token budget is managed at 12,000 tokens per request, with intelligent data trimming to ensure the most important metrics are always included. The model is configurable and can be upgraded to more powerful models as they become available, without any changes to your workflow.
Is my data safe?
Security is a core design principle of the AI recommendations system. Only aggregated statistical summaries are sent to the AI model; no individual user records, wallet addresses, or session data leaves your AnalyticKit instance. All data is team-scoped with strict isolation, meaning no team can access another team’s data or recommendations. Every API endpoint requires authentication. Additionally, circuit breaker protection prevents excessive API usage, and all communication with the AI provider uses encrypted channels.
Can I act on recommendations directly?
Yes. Every recommendation can be marked as “actioned” through the dashboard or API, creating a record of which recommendations your team followed up on. Over time, this builds an audit trail that helps you understand which types of AI insights drive the best results for your specific use case. You can also view 7-day summaries to track recommendation trends and your team’s response rate across all six intelligence categories.