NEXUS Sustain | Nuvei: 10x Lower Log Costs, Issues Caught Early

Nuvei's payments platform logged all 3 million events a day, so the Azure bill outran the business and real problems hid in the noise. EB Pearls took over operating it with no migration, cut log cost tenfold, and now finds problems first.

The question they were asking: How do we keep a live platform healthy as it grows, without the cost of watching it growing faster than the business?

Book a Discovery Session

NEXUS Sustain | Nuvei: 10x Lower Log Costs, Issues Caught Early

Meet Nuvei

Nuvei moves money between merchants, the partners who resell its platform, and the cardholders who pay through it. Its own engineers built the platform and keep shipping on it every sprint. In regulated payments, every event must be recorded and every outage reaches a merchant within minutes.

The platform worked. Watching it didn't.

Each of the 3 million daily events sent traces, health checks and debug lines to Azure, and Nuvei kept all of it. The log bill grew twice as fast as transactions. The security team, already stretched thin, hunted for real alerts inside the noise, and merchants sometimes reported problems first.

More servers would have paid for more noise. A one-off clean-up would have drifted back within a quarter, because every new service ships with new telemetry.

What another year of watching everything would cost Nuvei

Log spend

Est. A$14k / month

Rising est. 4% a month while transactions grew est. 2%.

Found by customers

Est. 6 issues / quarter

Merchants reporting problems before Nuvei's team saw them.

Security time

Est. 15 hrs / week

Analysts sifting routine warnings for the few that mattered.

A$14k a month × 12 months ≈ A$168k a year

in log spend alone, growing faster than the business, before counting the merchants who found problems first.

We ran it. Nuvei kept building it.

Fluid engagement. Fixed methodology.

Fluid · shaped to Nuvei

Named EB Pearls engineers and a DevOps lead worked inside Nuvei's own Azure tenancy. No migration, no re-platforming. They joined Nuvei's incident channel and change-approval board, while Nuvei's product team kept shipping.

Fixed · never flexed

A baseline before any change. Security, audit and transaction records never filtered. Every production change approved by an engineer and logged. A written improvement report every quarter.

How we took it over, then kept improving it

All NEXUS stages applied, and Operate and Evolve never stopped. Delivered using the method now productised as NEXUS Sustain.

In payments, the log you throw away is the one an auditor asks for, and the alert you miss is the one a merchant finds.

Stage 01

Discover

Reviewed the live platform in two weeks: incidents, cloud spend, alerts, telemetry.

NEXUS: Environment Review

Stage 02

Design

Set three numbers to beat each quarter. Locked audit records as keep in full.

NEXUS: Evolution Board

Stage 03

Deliver

Installed an AI-native watch in Nuvei's Azure. Shadow-ran 14 days, then switched.

NEXUS: No-Migration Onboarding

Stage 04

Operate

The workflow flags anomalies. Engineers triage and approve every change.

NEXUS: Operational Watch

The shadow run caught it. Three months later, it paid off.

In week one of the shadow run, a sampling rule thinned out payment-gateway timeout warnings. One timeout is routine. A burst is a provider outage, and the sampled stream hid the burst. The side-by-side comparison caught the gap. Engineers moved gateway timeouts to keep in full and added a burst alert. Three months later, that alert fired on a real gateway slowdown, est. 20 minutes before the first merchant ticket.

"One timeout is noise. Forty in a minute is an outage."

— Delivery Lead, EB Pearls

The machine watched. People decided.

The workflow scores millions of events a day. Three calls it never makes: what it may touch, how its rules change, and what reaches production.

What stays in

Every record an auditor could ask for

Engineers and Nuvei's security team locked security, audit and transaction records out of the filter. The model can't reclassify them.

What was killed

No rule changes itself

A self-tuning filter was on the table. We killed it: the model proposes a rule, an engineer approves it.

Who signed off

Every change goes through Nuvei's board

Switch-over was Nuvei's call. Since then, every production change goes through Nuvei's change board, with the full approval history logged.

"Couldn't a one-off clean-up do the same?"

For a quarter.

Nuvei ships new services every month, and each adds telemetry. Without someone owning the rules and the numbers, the bill and the noise drift back. Every production change is reviewed and approved by a senior EB Pearls engineer. Nothing reaches production without human approval.

The results, against Nuvei's baseline

Nuvei's result, on Nuvei's volume, measured every quarter. Yours starts from your own baseline.

Metric	Before	After · EB Pearls
Azure log cost	Est. A$14k / month	10x lowerAzure Cost Management, 30 days either side of switch-over
Time to detect	Est. 42 min	Est. 6 minIncident log, alert time vs first symptom
Found by customers	Est. 6 / quarter	Est. 1 / quarterMerchant tickets matched to incidents
Business translation	Est. A$168k a year	A$150k a year(A$14k − A$1.4k) × 12 months

Same Monday. Different job.

Before

The security team opens thousands of routine warnings and hunts for the few that matter. A merchant emails about failed payments. The Azure bill grows faster than transactions.

After

Routine events never land. A gateway burst arrives as one flagged alert, an engineer approves the fix, and merchants never notice. The quarterly report shows the cost per million events still falling.

And Nuvei kept going.Each quarterly report ends with what to fund next. The first recommendation Nuvei funded: putting two newly launched payment services on the same watch before their first transaction.

It's live. It matters. Can you afford to hear about problems from your customers first?

NEXUS Sustain runs your live platform under named engineers, wherever it was built. You hear about problems from us, not your customers, and every quarter you see what got healthier, faster and cheaper.

Book Your Free Discovery Call

1 Your Information

2 Book Meeting

3 Confirmation

Tell us about your app. We'll take it from there.

Share a few details about your project and what you need help with. Once you submit the form, you’ll be able to book your free discovery call.

What to expect on your call

What to expect

1 Share a few details
Complete the form with your contact details and what you need help with.
2 Book your free discovery call
Once you submit the form, choose a time that suits you for your discovery call.
3 Privacy comes first
Sign an optional NDA to ensure the highest privacy level and protection of your idea.
4 Discovery call
We’ll discuss your goals, the support you need and answer your questions. If we’re a good fit, we’ll outline the next steps.

What to expect

1 Share a few details
Complete the form with your contact details and what you need help with.
2 Book your free discovery call
Once you submit the form, choose a time that suits you for your discovery call.
3 Privacy comes first
Sign an optional NDA to ensure the highest privacy level and protection of your idea.
4 Discovery call
We’ll discuss your goals, the support you need and answer your questions. If we’re a good fit, we’ll outline the next steps.