Insights
Notes on operations
and execution.
Practical perspectives from our work with fintech companies, payment providers and financial services organizations.
All articles
-
Building operational structure before scale
When a growing fintech business stumbles, the cause is usually the structure under its strategy: the processes, ownership and systems that turn a plan into delivery you can repeat.
Read -
The difference between a process and a workaround
Workarounds stay invisible in reporting and expensive in practice. How to find them, put a cost on each one and decide which to formalize before they become the process.
Read -
Implementation debt: what happens after go-live
Projects get declared complete long before the operation can run on them unaided, and the manual work in between is where the benefits leak away. How to record that debt and pay it down.
Read -
Choosing systems your team will actually use
Whether a platform gets adopted is settled in the operations queue, by the people who work in it every day. What to test with them before you commit to a contract.
Read -
Who owns the data in a partner bank model
Sponsor bank contracts often name a ledger of record without saying which file feeds it or who reconciles it. What to settle on balances, customer data and exit rights.
Read -
Before you put an AI model in the operations queue
Pick one narrow decision and test the model against your own analysts before it touches live work. Then decide who handles the cases that fall below its confidence threshold.
Read -
Controls operations teams can live with
Maker-checker and four-eyes checks accumulate one incident at a time until nobody has added up what they cost. How to design them into the work and measure what each one buys.
Read -
Post-incident reviews that change something
A review that ends in vague actions with no owner tends to produce the same incident six months later. How to write actions that get finished, and how to spot repeats.
Read -
Planning capacity for peak season in payments
White Friday is close enough now that the plan is whatever it is. The peaks behind it, from the Dubai Shopping Festival to salary week and Ramadan, can still be planned properly.
Read -
Vendor due diligence beyond the questionnaire
A completed questionnaire shows what a supplier is willing to write down. Five checks that test those answers against evidence, plus what to keep watching after the contract is signed.
Read -
Why your operating model document is already out of date
Most of these documents describe the organization as it stood on the day they were approved. How we keep the RACI, the service catalog and decision rights close to current practice.
Read -
ISO 20022 cutover: what changes for operations teams
The end of MT and MX coexistence in November is usually run as a technology project. Screening, investigations and the exception queue will feel it more than the payments engine does.
Read -
Running a follow-the-sun operation from the UAE
Gulf Standard Time overlaps European mornings and much of the Asian working day. How to build handovers and a rota around that, including the hour that moves twice a year.
Read -
Three kinds of automation, in the order to try them
Fix the rules and remove steps first, then add workflow and integration, and consider machine learning last. Why operations teams that start at the other end often automate their own noise.
Read -
Exception handling is what customers remember
Customers form their view of a payments company when a transfer comes back or a refund stalls. What operations controls in those moments, from the status message to the follow-up.
Read -
Measuring operational cost per transaction
How to build a rough unit cost for each kind of operational work from time allocation and fully loaded salaries, and what it tells you about automation, outsourcing and pricing.
Read -
Handing a project to the team that will run it
Operations readiness needs its own acceptance criteria, agreed long before go-live. What to test, how to train, what documentation to leave behind and how to end hypercare cleanly.
Read -
The spreadsheet that runs your operation
Somewhere in most payments operations sits a workbook nobody designed and one person understands. When to replace it, and when to give it an owner and proper version control.
Read -
PCI DSS 4.0: the operational work behind the March deadline
On 31 March the future-dated PCI DSS 4.0 requirements become mandatory. Script control, tamper detection, MFA and risk analyses each need an owner and a routine that runs between assessments.
Read -
Moving payment providers without dropping transactions
A PSP migration usually means months with two providers live. Card data transfer, staged routing, reconciliation during the overlap, a rollback you can use, and what to tell merchants.
Read -
Service levels your partners can actually measure
Targets agreed with sponsor banks, PSPs and KYC vendors often cannot be checked from your side. Where the clock should start and stop, whose timestamps count, and why credits go unclaimed.
Read -
Instant payments leave no overnight window for exceptions
Euro instant payments now arrive at any hour. Fraud holds and failed credits need someone who can decide at 2am, with written limits and alerts that reach them.
Read -
The change freeze and the work that still has to happen
Certificates still expire over the holidays, and two EU regulatory dates land in the first weeks of January. Planning a freeze around the work that cannot stop.
Read -
How we scope an operational assessment
What fills the two to three weeks of a GSO assessment: interviews, sitting with the team while they work, a data request, and a short report.
Read -
Chargebacks: the cost that sits between teams
Dispute costs land in finance, operations, support and product, so no team sees the total or owns the causes. Giving the rate one owner and a monthly root-cause review changes that.
Read -
Getting ready for DORA when most of your stack is vendors
DORA applies from January 17, 2025. For a firm built on SaaS and outsourced services, most of the work sits in the vendor register, the exit plans and the contract terms.
Read -
Writing runbooks people open during an incident
At 3am, a tired engineer will open a runbook only if its first screen says what to check and who to call. Most runbooks are written with an auditor in mind.
Read -
KYC backlogs are rarely a staffing problem
Most onboarding queues grow because applications sit waiting and loop back for rework. Measuring waiting time separately from working time shows what to fix before you hire more analysts.
Read -
What the July outage showed about manual fallback
A faulty security update on July 19 crashed Windows machines worldwide. For payments teams it set a practical test: what could you still do by hand, and for how long?
Read -
Reconciliation is an operations problem first
Most unmatched items start upstream, in product decisions and operational shortcuts, long before finance sees them. We set out who should investigate each break and how to keep open items from aging.
Read