How Postscript Reduced Thousands of Redundant Retries per second

‍Fondue was acquired by Postscript - a leading SMS marketing platform for Shopify, making SMS the #1 revenue channel for over 10,000 Shopify brands and growing at pace. 

‍

“The company was making thousands of retry calls per second. Lunar.dev managed to flip the script for us out-of-the-box”. - VP R&D @Fondue, Omri Mosseri

‍

Developing a cashback system is no easy feat. The RnD team at Fondue, acquired by Postscript, understands that every call is a revenue line or refund request, tending to each call is business-critical.

Their mission is to enable brands to grow profitably by replacing the coupon code system with CashBack processes. They help ecommerce brands design and operate SMS programs that keep users engaged, drive revenue, power list growth and increase margins with the smart coupon alternative.

‍

Use Case:

‍

Imagine the following scenario: Thousands of retries every second (we’re talking over 25K), creating a backlog of retries and failed requests. 

‍

This was the case Fondue acquired by Postscript, and the goal was clear: implement an orchestration layer to gain visibility and optimization for API consumption over the massive volumes of third party API traffic generated per second. 

‍

In case of a product like this, speed and accuracy are of the essence and every API request impacts performance and revenue—especially when handling transactions like refunds.

The key challenges were:

• How quickly could the team process refunds while managing API rate limits?
• What happens when multiple API requests get queued by the cloud provider?

In a typical setup, excessive retries occur when requests are throttled or delayed. For example, if seven requests were queued on the cloud provider’s side, they would sit in the backlog, waiting for processing. This caused delays and inefficiencies.

How Lunar.dev Reduced API Retries and Errors

How do you process refunds fast—without running into API rate limits or getting stuck behind a cloud provider's queued requests? In typical setups, throttled requests trigger excessive retries and errors, compounding delays. For example, seven queued requests might sit idle, wasting precious time and compute.

Lunar.dev’s API Consumption Gateway slashed errors completely and retries to just 1 in every 100,000 requests. No custom middleware. No chaos. Just intelligent traffic handling.

With built-in priority queuing, Lunar.dev ensured time-sensitive calls were processed efficiently, reducing backlog and maximizing throughput at scale.

The Game-Changer: API Aggregation Without Data Retention

The biggest impact came from Lunar.dev’s ability to aggregate thousands of requests efficiently while maintaining optimal performance and reliability:

- Prioritization—Each call was processed based on business logic, ensuring mission-critical requests weren’t delayed.

- Seamless integration—Instead of building and maintaining an in-house middleware solution, the team installed Lunar.dev and gained instant visibility into their API consumption patterns.

- No data retention—Lunar.dev never stores or sees the payload.

In high-scale environments, traditional retry mechanisms and rate limit handling fall short—causing delays, wasted compute, and degraded user experiences. Lunar.dev shifts teams from reactive workarounds to proactive API traffic management. With intelligent queuing, built-in retry controls, and full visibility into consumption patterns, Lunar.dev empowers engineering teams to move faster, spend less, and operate with confidence—no custom tooling required.

API consumption doesn’t have to be a bottleneck. With Lunar.dev, it becomes a competitive edge.