AWS exit: Cloud costs reduced from €50,000 to €10,000

everysize is the only search engine on the German web that finds sneakers in your preferred size at the best price. Founded as a bootstrapped start-up, today with over 2 million happy customers. Instead of burning growth on AWS list prices, the seven-person team partnered with We Manage to execute a 1.5-year cloud exit to Hetzner — and today pays a fifth of the original cloud costs.

The numbers at a glance

80%

Cloud cost reduction through AWS exit to the Hetzner cloud.

5×

cheaper: annual server costs reduced from €50,000 to €10,000.

1.5 years

controlled migration with zero downtime, in parallel with live operations.

"We Manage is an optimal partner for us, providing reliable support for our web applications 24/7. Beyond competent and reliable consulting, we are especially grateful for the fast and direct communication."

Eugen Falkenstein

CEO, Everysize GmbH

From bootstrapping success to AWS trap

everysize appeared on Vox's "Höhle der Löwen" (the German Shark Tank) in 2017 — and despite a panning by the celebrity investors, walked away with an unofficial traffic record and a record profit. The founders around Eugen Falkenstein then consistently invested most of their earnings into IT infrastructure rather than marketing.

Like many start-ups, they began on AWS. But over the years, fixed costs doubled — at nearly identical resource consumption. The bootstrapping model couldn't absorb that cost curve. A cloud exit was needed.

3 reasons why everysize left AWS

For bootstrapped businesses with predictable workloads, the benefits of large public clouds quickly become smaller than their price tag.

Escalating fixed costs

Five-figure monthly AWS bills at nearly unchanged resource demand — increasingly complex pricing models led to a cost doubling over two years, with no added value.

Vendor lock-in

AWS-specific managed services made switching expensive and blocked any negotiating leverage. Multi-cloud capability was never part of the original design.

Service-provider trap

The previous agency grew, processes became more formal. Partner-style consulting turned into ticket processing — well done, but not what a bootstrapped startup needs in an equal-footing partnership.

Our approach: divorce in installments

Instead of a big-bang migration weekend, we executed the cloud exit workload-by-workload over 1.5 years. This kept live operations stable throughout.

Migration to Hetzner

Over 1.5 years, we moved everysize's servers from AWS to the Hetzner cloud — structured, without downtime, with continuous A/B performance comparison.

Self-managed MySQL cluster

What AWS RDS provided for everysize, we rebuilt on Hetzner as a highly available MySQL cluster — performant and roughly 10× cheaper to operate monthly.

URL shortener for social media

A self-hosted URL shortener on Hetzner replaces expensive SaaS services, without tracking restrictions or external dependencies.

24/7 operations

From the trial engagement on, monitoring, IT security, patching, and scaling run around the clock — issues are usually solved before everysize notices them.

How does We Manage help?

The tech stack at a glance

Hosting: Hetzner Cloud & Google Cloud
OS: Ubuntu LTS
CDN: Cloudflare & BunnyCDN

We use Ansible consistently to automate server configuration.

everysize's Elasticsearch search engine runs on our Managed Cloud Servers. Parts of the backend are built with Symfony — and we offer the Symfony stack as a managed service to other customers as well.

Technologies

nginx
ElasticSearch
Symfony
PHP
Redis
MySQL
MongoDB
Docker
ClickHouse

Applications

Airbyte
MySQL Cluster
URL Shortener
VPN
Metabase

Full control over the digital value chain

Thanks to the high degree of automation, the sneaker search engine now runs more reliably than before the migration. In emergencies, Eugen Falkenstein no longer has to mobilize service providers — issues are reported proactively and usually resolved in the same breath, even at night or on weekends.

New ideas now ship in weeks rather than months. everysize has regained the agility that small digital companies can wield against large corporations.

The results after 1.5 years of migration

The cloud exit paid off twice for everysize: significantly lower costs and better performance at the same time.

5× cheaper

Annual server costs reduced from €50,000 to €10,000 — at consistent or better performance of the sneaker search.

Weeks instead of months

New features and experiments ship in weeks rather than months — the agility of a small digital company is back.

Proactive operations

We Manage doesn't report problems — we solve them. Including nights and weekends, without anyone at everysize having to step in.

The full everysize story as PDF

Cloud exit from AWS to Hetzner over 1.5 years — methodology, numbers, and lessons learned, first hand.

How we migrated the AWS stack workload by workload
Which AWS services Hetzner lacks — and how we rebuilt them
Why async collaboration beats standing meetings

Conclusion

With 24/7 managed operations, Hetzner cloud infrastructure, and a sparring partnership on equal footing, everysize is ready for the next growth phases — from marketplace expansion to CMS modernization to potential international expansion.

Also paying five-figure AWS bills and wondering whether it has to stay that way? Let's talk.

Get in touch

Frequently asked questions

What we are often asked about AWS exits, cloud migration, and multi-cloud strategies — answered from the everysize migration and comparable projects.

How much in cloud costs can you save with an AWS exit?

For bootstrapped businesses with predictable workloads, 60 to 90% reduction is realistic. everysize saved 80% — from €50,000 to €10,000 per year. The prerequisite: AWS-specific managed services either need to be rebuilt in-house or replaced with alternative provider services.

How long does a cloud migration from AWS to Hetzner take?

For everysize we planned and used 1.5 years — workload by workload, without a big bang and without downtime. Key factors are the complexity of the existing architecture, the parallel-running live business, the number of AWS-specific services, and data volume. Migrating significantly faster risks performance or availability dips.

Which AWS services are not directly available on Hetzner?

Hetzner has no direct equivalents to AWS RDS, S3, Lambda, ElastiCache, or Managed Kubernetes. These services must be self-operated — at everysize, for example, we set up a MySQL cluster on Hetzner servers ourselves. Advantage: roughly 10× cheaper than RDS at comparable performance. Disadvantage: the operations effort doesn't disappear, it shifts to the cloud partner.

When does a self-managed MySQL cluster beat AWS RDS?

For workloads with largely constant load and a platform architecture that should remain multi-cloud-capable. RDS bills per compute hour plus storage plus IOPS — at stable demand, these line items quickly add up to a factor of 10 compared to a self-managed cluster on Hetzner. With unpredictable auto-scaling demand, RDS stays simpler.

How do you organize collaboration between internal team, dev agency, and cloud partner?

At everysize, communication between Eugen Falkenstein (internal), the development agency Intradesys, and We Manage (operations) runs primarily asynchronously via Slack and email — without classic standing meetings. This gives everyone involved focus and time to think before answering, producing better decisions overall than spontaneous calls. Prerequisite: trust and clearly defined roles.