How to Count Website Visits Efficiently with Redis: Hash, Bitmap, and HyperLogLog

Counting website user visits is a common analytics need, but traditional relational databases struggle with real‑time high‑concurrency counting. Redis offers several data structures and commands that can solve this efficiently.

1. Hash

Hash is a fundamental Redis data structure. Redis maintains a hash table that maps keys to hash slots; on collisions it chains entries. Using a Hash, we can design a simple visit‑counting scheme.

When a user visits the site, a logged‑in user uses their user ID as the identifier; an anonymous user is assigned a randomly generated string key on the front end.

During a visit, the page URL serves as the key , the user ID or random string serves as the field , and the value is set to 1. The HSET command adds data to the Hash, and the HLEN command returns the total count.

This solution is simple, easy to implement, and provides accurate results. However, as the number of keys grows, memory consumption and performance degrade, making it suitable only for pages with modest traffic.

2. Bitmap

A 32‑bit integer can represent a single user ID, but converting it to a binary bitmap allows each bit to represent a distinct user. Thus, 32 bits can track 32 users, saving up to 32× memory compared with storing full integers.

For logged‑in users, their IDs map to bitmap positions; for anonymous users, a hash of the random identifier maps to a bit. The BITCOUNT command directly yields the total number of set bits, i.e., the visit count.

The bitmap approach uses less memory and offers fast queries, but multiple users may map to the same bit, causing some counting bias. To reduce bias for anonymous users, a separate mapping table can be maintained. In extremely sparse scenarios (e.g., a user ID of 100 million), allocating that many bits can waste memory.

3. HyperLogLog

When a site experiences massive traffic and exact counts are not required, a probabilistic algorithm can be used. Redis implements the HyperLogLog algorithm, which estimates cardinality without storing individual values.

Visit data is added to a HyperLogLog key, and the PFCOUNT command returns an estimated count. This method consumes extremely little memory and scales to billions of users, but it cannot query individual users and introduces a small estimation error.

These three Redis‑based techniques—Hash, Bitmap, and HyperLogLog—provide flexible options for website visit statistics. Choose the one that best matches your traffic volume, memory constraints, and accuracy requirements.