How Shoptimised cut infra costs by 42%, and deployment time by 30x
Shoptimised sits between ecommerce stores and their ad platforms, rewriting product feeds so Google Ads picks up the items that would otherwise get ignored. More than 8,000 stores use Shoptimised today, from small clothing brands up to 570 larger clients who go a step further and have Shoptimised's team handle their entire ad campaign.
The whole platform is a fleet of Laravel microservices that pushed 18.5 million background jobs and three million requests in its first week on Laravel Private Cloud.
Shoptimised moved from Laravel Vapor to Private Cloud, experiencing only a few minutes of downtime as part of the migration. Since migrating, the team has seen multiple benefits:
A 42% reduction in infrastructure costs compared to running on AWS and Vapor. The team's return on investment with Cloud is roughly 72%: for every dollar they spend on Cloud, they're saving about $0.72 compared to their previous AWS/Vapor spend.
Deployment time dropped approximately 30x, from over an hour on AWS and Vapor to one to two minutes on Laravel Private Cloud.
Full fix cycles (investigation, fix, redeploy) went from around two hours to minutes.
Cloud saved them around 40 hours per week of DevOps work, an estimated $100K to $160K per year in DevOps hiring, and the equivalent of one less engineer managing infrastructure.
Preview environments that used to take a week to stand up on AWS and Vapor now spin up per pull request and scale to zero when idle.
If you are weighing a move off AWS and Vapor, see the Vapor to Cloud migration guide to see what the process looks like for your stack.
About Shoptimised
Shoptimised is a B2B SaaS platform that optimizes product feeds for e-commerce brands, with a managed service on top for Google Ads performance. The company targets non-clicked, non-converting products with rule-based transformations, which they call "incremental sales."
The team is just shy of 50 people, with nine engineers including CTO David Zborowski.
Why AWS and Laravel Vapor stopped working
Shoptimised had been running on AWS with Laravel Vapor on top, plus a custom autoscaling group of EC2 instances underneath. Each instance ran a bespoke Docker container with 25 queue workers, all wired to a database. Deployments took over an hour, and any bug fix took roughly two hours from detection to redeployed code.
Our deployments took over an hour to deploy any changes, so there was growing frustration among our developers," said David Zborowski, CTO of Shoptimised. "We wanted to improve the developers’ experience and to eliminate the need for a DevOps function.
The custom Docker setup made queue management especially painful. Because the container carried its own database dependency, jobs had to be pinned to specific EC2 instances. A bespoke queue manager had been built on top of Laravel's defaults to compensate. "It proved really difficult to track the jobs, to dispatch them correctly, to retry the jobs, to back off the jobs, or really to easily scale the number of queue workers," David said.
Two other requirements pushed the decision. Shoptimised wanted the ability to expand quickly into new regions to meet data residency rules in future markets, and the team wanted to standardize on Kubernetes so the platform was not locked into any one provider.
Moving to a Cloud platform built for Laravel
Removing the engineering work around queues
Shoptimised worked closely with the Laravel Cloud team on the design of managed queues. "We had quite a lot of influence on the managed queues," David said. "We've talked with Kieran [Brown, Staff Software Engineer] quite a lot on how we're going to develop it, and what we needed. Now we don't worry about it. We just set how much memory we want, how many max instances, and it's being taken care of."
The caching microservice alone runs 16 managed queues, each configured to scale from 0 to around 200 workers. That capacity is what makes an enterprise upsell possible: Shoptimised can now offer big-name customers a dedicated priority processing lane. "It's a new methodology to our approach and what we can sell to customers," David explained.
Preview environments that scale to zero
On AWS and Vapor, standing up a preview environment for a single feature took roughly a week of infrastructure work. Shoptimised had production and staging, and that was it.
On Laravel Cloud, every engineer gets preview environments per pull request that scale to zero when idle. "Every single developer almost gets unlimited environments per PR," David said. "They can share the link, our product owner can review it. It's such an improvement to the workflow."
Serverless Postgres and pooled databases
The caching microservice runs on 9 to 14 Serverless Postgres databases, pooled to spread load and connections across the fleet. MySQL runs the main business databases and staging environments.
Looking at the numbers on his dashboard, David said that the caching microservice alone served 153 million queries in about a day and a half.
Migrating five microservices
Shoptimised did not migrate the monolith as-is. The AWS and Vapor setup ran a single application with a very specific Docker setup that Cloud did not support, so the team spent some time decoupling it into three microservices: an app service for the customer-facing product, a data service for Google Merchant and Ads API aggregation, and a caching service for the feed ETL work.
"There were a lot of unnecessary things in one codebase, which created a huge context for the developer," David said. "If they introduced a bug, the bug has a bigger blast radius because it takes out the entire application."
When cutover night came, the preparation paid off. Live traffic switched with roughly 15 to 20 minutes of downtime.
What changed for the team: Costs, DevOps, and AI
Beyond the 30x drop in deployment time, the most significant change for Shoptimised has been cost, with a 42% reduction in infrastructure spend. On a dollar-for-dollar basis, that's a 72% return: for each dollar going to Cloud, Shoptimised is saving roughly $0.72 versus what AWS and Vapor used to cost.
Part of that saving comes from time, not just infrastructure spend. The engineer responsible for Shoptimised's original setup spent the majority of their working hours managing infrastructure. When that person moved on, Laravel Cloud meant Shoptimised didn't need to backfill the role.
"If we hired a DevOps engineer, it would be eight hours a day. That's 40 hours a week (a full-time job) that we don't have to spend," David said. "Each developer can easily handle the management of the Laravel applications on Laravel Cloud."
The team is also leaning into Laravel's AI tooling. Every project has Laravel Boost installed. Internal agents built on the Laravel AI SDK already classify incoming Jira tickets as bugs or feature requests, and low-confidence classifications are surfaced through the human approval flow announced at Laracon US. Deeper use of the Laravel MCP is next on the roadmap.
Try Laravel Private Cloud
Shoptimised cut deployment time by roughly 30x, saved 42% in infrastructure costs (a 72% return on investment on their Cloud spend), and opened a new enterprise revenue line, all while processing 18.5 million jobs a week on Laravel Private Cloud.
David's advice for teams considering the move is short:
Stick to the Laravel way of doing things, and it will work on Cloud no matter what. Don't be scared and take the risk.
If your team is running Laravel on Vapor, read how Vapor and Cloud compare and try Laravel Cloud. If your team has strict security requirements and microservices that require private connections, learn more about Private Cloud.
Shoptimised
Problem
A bespoke setup on AWS with Laravel Vapor on top made deployments take over an hour, required a custom queue manager on top of EC2, and would have needed a dedicated DevOps hire to scale. Standing up a preview environment took roughly a week.
Solution
Shoptimised decoupled its monolith into three microservices and migrated to Laravel Private Cloud on Kubernetes. Deployments now take one to two minutes, infra costs are down 42% (a ~72% ROI on Cloud spend), 16 managed queues on the caching microservice scale from 0-200 workers each, and every engineer gets preview environments per pull request that scale to zero when idle.
Talk to a Cloud expert
Let the Laravel team design the right infrastructure configuration for you.
Contact sales