Case studies / Data foundations

Data foundations · Grocery retail

Cutting $50K a month from a Snowflake bill — without losing performance

Snowflake spend had grown to the point of real budget pressure, with no clear picture of where the cost was going — and any change risked hurting query performance. We found $50K a month, without giving performance back.

$50K/mo
Saved

Recurring monthly savings on Snowflake spend.

0%
Performance lost

Query performance held steady throughout.

0%
UX impact

No degradation to the reporting experience end users felt.

Sector
Grocery retail
Engagement
Snowflake FinOps engagement
Problem area
Data foundations
Confidentiality
Details anonymised

The challenge

Rising cost, unclear cause, real performance risk

Snowflake spend had grown to the point of real budget pressure, with no clear picture of where the cost was going — and every proposed fix carried a risk of hurting query performance or user experience.

Before

  • Snowflake spend rising with no clear driver
  • No workload-level cost visibility
  • Any change risked hurting performance
  • Cost conversations stalled on uncertainty

After

  • $50K/month saved, recurring
  • Performance held steady on critical workloads
  • User experience unaffected
  • Clear, ongoing visibility into spend drivers

How we approached it

Audit first, then tune without disruption

01

Right-size warehouses

We eliminated over-provisioning while protecting the reporting SLAs and user experience that mattered.

02

Tune workload policies

We set auto-suspend and auto-resume policies separately for batch workloads and office-hours warehouses.

03

Reduce storage overhead

We compressed large transaction and log tables to cut storage cost without losing the data teams needed.

04

Match retention to recovery need

We tuned time-travel and fail-safe retention windows against the client's actual recovery requirements.

Under the hood

SnowflakeSnowpipeWarehouse right-sizingAuto-suspend / resumeCompressionTime Travel & Fail-safe tuning

Next step

Watching a cloud data bill climb with no clear reason why?

We find where the cost is really going, then bring it down without touching the performance people depend on.

Start a conversation →