No cost ownership
The bill arrived monthly, but no one was responsible for reviewing what each charge was actually for.
Back to case studies
Cost reduction
Cutting a Growing Company's Cloud Bill Without Business Disruption
A growing Philippine retail and ecommerce company was paying for cloud hosting and software subscriptions that had quietly grown far beyond what the business actually used. I reviewed every recurring charge against real usage, replaced four paid services with systems running on hardware the company already owned, and cut the recurring bill by about 93 percent, roughly 1.8 million pesos a year, with a documented path back at every step.
IndustryRetail and ecommerce
Company sizeGrowing SME
Annual savingAbout ₱1.8M a year
Overview
The bill was the symptom. Nobody reviewing it was the cause.
Once every recurring charge was matched against real business use, most of the monthly spend turned out to be idle capacity, duplicated databases, and paid subscriptions that could run on hardware the company already owned.
The business problem
The monthly bills kept growing and nobody could explain them.
Cloud costs are easy to start and hard to stop. Resources get created for a project, the project ends, and the charges continue quietly every month until someone sits down and traces each line item back to a business reason.
Idle and forgotten resources
Servers, databases, and network components kept billing long after the work that needed them had finished.
Oversized database tier
The company paid for database capacity and duplicate read copies far beyond real production demand.
Paid tools with free equivalents
Analytics, search, and log-monitoring subscriptions were billed every month for capability that could run on hardware the company already owned.
Fear of touching anything
Without a tested rollback plan, turning things off felt risky, so everything stayed on by default.
Cost of delay
Every month of inaction repeated the same avoidable spend on infrastructure the business had outgrown the need for.
Challenges
Saving money is easy. Saving money without breaking the business is the job.
- Live revenue systemsThe ecommerce storefront and internal operations ran on the systems being moved, so an outage would cost real sales.
- No acceptable data lossOrder and transaction records had to arrive complete, with no gap between the old and new systems.
- Unknown true baselineThe real recurring cost had to be measured from finalized invoices and provider billing data, not estimated from memory.
- ReversibilityThe business needed a way back if the new setup failed after the switch.
- Capacity confidenceThe replacement had to be proven able to handle peak traffic before, not after, the switch.
- Ongoing protectionBackups, monitoring, and recovery had to be in place so the savings did not create a new risk.
Solution
Measure first, prove the replacement, switch in a controlled window, then remove cost in stages.
I started by establishing the true cost baseline from the provider's own cost reporting across several months, then inventoried every running resource and matched it against an actual business purpose. Anything that could not be justified was flagged, not immediately deleted.
In parallel I built the replacement environment on owned infrastructure and load tested it against roughly five times the company's normal peak traffic, requiring a clean result before the switch was scheduled at all.
The production switch itself ran in a planned ten-minute window at midnight, chosen so customer impact was minimal. The database was brought forward to an exact matching position so no records were lost in the handover.
Only after the new environment ran cleanly did I remove the old infrastructure, in deliberate stages with the owner approving each round, and only after preserving named recovery snapshots and a written rollback procedure.
Approach
From an unreviewed bill to a measured, reversible reduction.
Each stage had to pass before the next one started. That sequence is what made an aggressive cost reduction safe to perform on live business systems.
Measure real baseline
Inventory every resource
Build replacement
Load test at 5x peak
Planned cutover
Verify in production
Staged teardown
The numbers
The same bill, before and after the decision to review it.
January to August are finalized actuals from provider invoices. September onward reflects the migrated path, with the remaining months modelled at verified rates. Figures in Philippine pesos.
Nothing changes for eight months, because nothing had been reviewed. The gap opens in September, when the migration completed, and stays open every month afterwards.
Business results
About 93 percent lower recurring cost, roughly ₱1.8 million a year.
These figures come from the providers' own billing data and paid invoices, compared against what the same services would have cost if nothing had changed. January to August are finalized actuals; the remaining months are modelled at verified rates.
About ₱150,000 saved monthly
Recurring spend fell from roughly ₱162,000 per month to about ₱12,000 per month across the four services reviewed.
About ₱1.8 million a year
The reduced monthly run rate continues every year without further work, making it a permanent change to the cost base rather than a one-off saving.
About 93 percent lower
The combined bill for hosting, analytics, search, and log monitoring dropped by roughly 93 percent after the migration completed.
Paid for itself in under a month
The temporary cost of running both environments during the migration was recovered by the savings within the first month.
Ten-minute planned window
The production switch ran inside a scheduled midnight window rather than an unplanned outage, with no data loss.
Faster product search
Moving search onto owned infrastructure cut typical search response from hundreds of milliseconds to about one millisecond.
Where the money went
Four recurring bills, reviewed one at a time.
Each service was assessed on its own: what it cost, what it actually did for the business, and whether an equivalent could run on hardware the company already owned.
- Cloud hostingThe largest line item. Application servers, databases, and staging environments moved to owned infrastructure, reducing this bill by roughly 96 percent.
- Product analytics subscriptionReplaced with a self-hosted equivalent, removing the monthly fee entirely.
- Hosted search serviceMoved onto owned infrastructure, removing the subscription and making search dramatically faster at the same time.
- Log monitoring subscriptionReplaced with open-source log collection and dashboards, removing the monthly fee while keeping visibility.
- Honest accountingThe savings figure subtracts an operating allowance for running the owned infrastructure, so it reflects net benefit rather than a headline number.
Three subscriptions were removed outright. The remaining hosting bill dropped by about 96 percent once idle and duplicated resources were retired.
Safeguards
Every step had a way back before it was taken.
- Capacity proven before the switchThe replacement passed a load test at about five times normal peak traffic with no failed requests before any cutover was scheduled.
- Written rollback procedureA step-by-step return path to the previous environment was documented and kept available, with an expected completion time well under an hour.
- Named recovery snapshots retainedPoint-in-time database snapshots from immediately before the switch were preserved as the recovery anchor rather than deleted with everything else.
- Staged teardown with owner approvalOld infrastructure was removed in separate rounds, each one explicitly approved, instead of a single irreversible cleanup.
- Nightly offsite backupsAutomated nightly database backups to separate storage were verified as actually landing, not assumed.
- Monitoring kept in placeLog collection and dashboards were re-established on the new environment so problems stayed visible after the savings were realized.
Technologies
Standard, well-understood infrastructure rather than novel tooling.
AWSCost ExplorerMySQLDockerLinuxCaddyTypesenseGrafana LokiLoad testingAutomated backups
Why this matters
Recurring cost is the easiest money a business ever gets back.
Most growing companies never audit recurring technology spending, because it is nobody's job and everybody is afraid to switch something off. A careful review with a tested rollback plan turns that fear into a permanent reduction in the cost base, money the business keeps every month from then on without further effort.
Not sure what your cloud bill is actually paying for?
Book a free discovery call and we will look at where the recurring spend is going and whether it can be reduced safely.
Book My Free Discovery Call