In Move, events are stored as part of the transaction output and emitted via the Event module. To design a scalable event logging system, I:

1. Define a custom event structure that captures essential analytics data such as user_id, action_type, timestamp, and metadata.

   struct UserActionEvent has copy, drop, store {
       user: address,
       action: vector<u8>,
       timestamp: u64,
   }
   

2. Emit events inside transactions when users interact with the dApp:

   public fun log_action(user: &signer, action: vector<u8>) {
       let event = UserActionEvent {
           user: signer::address_of(user),
           action,
           timestamp: std::time::now_seconds(),
       };
       event::emit(event);
   }
   

3. Use off-chain indexers (like Sui Indexer or GraphQL APIs) to collect, filter, and process these events for analytics dashboards.

Unlike traditional logs, Move events are immutable and verifiable, ensuring real-time analytics remain tamper-proof. To scale, I would shard event storage per dApp module and use batched off-chain ingestion for efficiency.