A festive sale compresses a quarter’s traffic into a few hours. Diwali, a flash-sale midnight, a payday weekend: each brings a concentrated surge that your store has never seen at full scale. Load testing your e-commerce store is the only way to find out what breaks before your customers do.
This guide covers the tools worth using, the benchmarks worth setting, and the bottlenecks that surface first in the platforms we see.
Why Festive Traffic Breaks Websites
Average load in e-commerce store is a comfortable lie. A store that handles a normal Tuesday cleanly can still fail in the first 10 minutes of a sale. Festive traffic arrives as a spike, often triggered by a push notification, an influencer post or a countdown reaching zero.
Three factors make the spike dangerous:
Speed of ramp-up: auto-scaling reacts in minutes, but a surge arrives in seconds.
Concentration on a few endpoints: the same 20 products, the cart and checkout absorb most of the load.
Dependency pressure: payment gateways, SMS providers and logistics APIs face the same surge.
Plan for the peak minute. The average hour does not decide your revenue.
Test Like Real Shoppers
A load test is only as honest as its traffic model. Hitting the home page with 50,000 identical requests proves very little. Real shoppers search, filter, compare, abandon carts, apply coupons and retry failed payments.
Build your scenarios from production analytics:
Journey mix: the ratio of browsers, cart-adders and buyers during your last sale.
Device and network profile: mobile-heavy traffic on variable 4G changes payload and timeout behaviour.
Data variety: thousands of distinct SKUs, users and coupon codes, so caches are not artificially warm.
Think time: realistic pauses between actions, with retry storms modelled explicitly.
Test what shoppers do. Synthetic website/app traffic gives synthetic confidence.
What Numbers to Aim For
Without pass/fail criteria, a load test produces opinions. Agree on thresholds with business and engineering before you start, so results lead to decisions instead of debates.
The following are starting targets. Calibrate them against your own baseline and customer tolerance:
Response time: p95 under 2 seconds for browse and search; p95 under 3 seconds for checkout.
Error rate: under 1% across the full journey, and near zero on payment confirmation.
Throughput: sustained orders per minute at 2x to 3x your highest previous peak.
Resource headroom: CPU and memory under 70% at peak, so a spike beyond plan does not tip over the system.
Recovery time: return to normal latency within minutes of the peak passing.
A benchmark you set after the test is a justification, not a standard.
Where Things Usually Break
Festive failures cluster in predictable layers. Teams often over-invest in web servers and under-invest in the components that actually saturate. Start your investigation where contention is highest.
The usual suspects:
Database connections and locks: inventory updates and order writes contend on the same rows.
Search and catalogue services: uncached queries multiply under filter-heavy browsing.
Payment and OTP integrations: third-party rate limits and timeouts throttle conversion at the worst moment.
Session and cart state: poorly sized caches or sticky sessions create uneven load.
Autoscaling lag: cold starts and slow health checks leave capacity behind demand.
Find the weakest layer first. It sets the ceiling for everything above it.
Problems Teams Often Miss
Some festive failures never show up in a website-only test.Mobile apps, bots and stock logic add pressure that a browser script does not reproduce.
Add these to your test plan:
App API traffic: one app screen often triggers many API calls, so app users load the server harder than web users.
Push notification bursts: one ecommerce festive campaign sends thousands of users to the same screen within seconds.
Bot and scraper traffic: automated buyers and price scrapers consume capacity meant for customers.
Stock accuracy: simultaneous orders for limited items can oversell if inventory updates are not atomic.
Heavy pages and images: uncompressed banners and uncached assets slow every page on mobile networks.
Test the app, the bots and the stock count, not just the website.
What to Do When Traffic Beats the Plan
Every plan has a limit, so decide in advance what happens beyond it. A controlled slowdown is better than a full outage. Agree these fallbacks before the festive sale, not during it.
The fallbacks worth preparing:
Waiting room: queue extra visitors instead of letting them crash checkout.
Feature switches: turn off recommendations, reviews or heavy filters to protect cart and payment.
Rate limits: cap requests per user and per IP to block abuse and retry storms.
Cached fallbacks: serve saved product listing pages when the backend is struggling.
Rollback path: keep the last stable release one command away.
Slow and working beats fast and down.
What One Need to Decide
Peak readiness is a business decision before it is an engineering task. Engineers can show where the system breaks. One must decide how much risk and cost the business accepts.
The decisions to make early:
Cost of downtime: estimate revenue lost per hour of outage and compare it with the cost of extra capacity.
Temporary capacity budget: approve extra cloud spend for the sale window in advance.
Ownership: name one person for the go/no-go call and one owner for each critical system.
Partner readiness: confirm payment, SMS and logistics providers can handle your expected peak.
Go/no-go criteria: agree which test results allow the sale to proceed.
Decide early, because decisions made during an outage cost more.
When to Test
Load testing delivers value as a repeatable practice. A single pre-sale test finds problems too late to fix them properly. Teams that treat it as a release gate arrive at peak season with fewer surprises.
A workable rhythm:
6 to 8 weeks out: baseline test on a production-like environment to find structural issues.
3 to 4 weeks out: fix, retest and validate caching, indexing and scaling rules.
1 to 2 weeks out: full-scale peak and soak tests, including dependency failure scenarios.
Sale week: code freeze, runbooks ready, war-room dashboards live, and rollback paths verified.
Post-sale: review real traffic against your model and update it for next season.
Test early, fix calmly, and let the sale be the least dramatic day of the quarter.
Quick Summary
Design for the peak minute, not the average day.
Build traffic models from real shopper behaviour.
Agree benchmarks before testing begins.
Test the dependencies, not only your own code.
Repeat the exercise every season.
Get Your Platform Ready
If your festive calendar is already set, your testing calendar should be too. Our engineering team works with enterprise retail and commerce teams on website and app development, with performance engineering and scalability. Talk to our experts for a technical assessment of your platform.
In a country that moves more than 800 million UPI payments a day, every second of lag and every minute of outage is a balance-sheet event.
India’s fintech story is a story of scale: UPI processed 24.07 billion transactions worth ₹29.37 lakh crore in September 2026, up 23% year on year. That scale is the achievement, and it is also the exposure. This article looks at what unavailability cost Indian banks, payment apps and fintechs, using NPCI data, RBI actions, recent outages and a few simple models.
1. At UPI scale, a minute of downtime is never small
India’s payment volumes have turned every outage into a national event.
NPCI data shows UPI averaging about 802 million transactions a day in September 2026, worth roughly ₹97,900 crore daily. Spread across the clock, that is a continuous flow, so any interruption cuts off a very large number of people and merchants at once. On March 26, 2025, volumes fell to 550 million against a March daily average of 591 million, a 7% dip from a single disruption.
Per day: about 802 million transactions and ₹97,900 crore (NPCI, September 2026)
Per minute: roughly 557,000 transactions and about ₹68 crore in value (calculated from the daily average)
Per second: around 9,300 transactions
Per month: 24.07 billion transactions in September 2026, up 23% year on year, so the exposure keeps growing
At UPI scale, a minute is never small.
2. Indian outages are a pattern, not a blip
UPI disruptions cluster around predictable peaks, which makes them a planning failure as much as a technical one.
In the span of three weeks in March and April 2025, UPI saw repeated disruptions, drawing a call from the finance minister for NPCI to make the system more modern. Earlier, a brief nationwide disruption in February 2024 was blamed on partner banks, and a ransomware attack on technology provider C-Edge in August 2024 hit more than 300 small banks. The common thread is dependency: when one bank, vendor or switch slows down, the whole network feels it.
March 26, 2025: intermittent technical issues and partial declines at NPCI
March 31 to April 1, 2025: financial year-end closing strained some banks, causing intermittent declines
April 2, 2025: fluctuating bank success rates raised network latency, with an SD-WAN glitch also suspected
April 12, 2025: partial declines hit Google Pay, PhonePe, Paytm and bank apps, with over 2,000 Downdetector complaints logged by noon
Peaks are predictable, so peak failures are preventable.
3. Slow and failing is the new down
A payment rail that is up but slow or leaky does almost as much damage as one that is switched off.
Outage headlines capture only the extreme. The everyday cost sits in technical declines, timeouts and latency. In theory terms, averages hide the tail: a UPI payment touches the remitter bank, the NPCI switch and the beneficiary bank, so one slow link drags the whole transaction, and customers feel the slowest hop, not the average one. Scale makes small percentages large: a 1% technical decline rate on 802 million daily transactions means roughly 8 million failed payments a day.
Technical declines: NPCI data from 2021 showed 1% to 2% of UPI initiations failing for technical reasons
Business declines: a further 6.8% were declined for reasons such as wrong PIN or insufficient balance
Bank variance: in March 2023, technical declines ran at 2.45% for SBI and 7.16% for IDBI Bank, so the weakest link sets the experience
Latency spillover: on April 2, 2025, success-rate swings at some banks raised latency across the whole network
A slow payment is a failed payment.
4. In a zero-MDR market, the bill arrives in retention, not fees
Indian fintechs rarely lose direct revenue to downtime, which is why they underestimate its cost.
UPI person-to-merchant payments up to ₹2,000 carry zero merchant discount rate, and person-to-person transfers are free, so a failed payment seldom costs a fee. The loss shows up elsewhere: in cross-sell, lending and credit revenue that rides on the engaged user, and in the cost of winning that user back. A usable model is:
Incident cost = (lost engagement value + support and engineering labor + compensation + regulatory exposure) x duration + long-tail churn and brand repair.
For a global yardstick, Splunk’s 2026 research puts average downtime at about $15,000 a minute and the Global 2000 bill at $600 billion a year, with nearly 20% of marketers saying brand health takes a full quarter to recover.
Support load: failed-payment complaints flood call centres, chat and social channels within minutes
Compensation: RBI’s time-bound rules on failed transactions carry per-day penalties, for example ₹100 a day for ATM failures not resolved beyond T+5
Engineering time: war rooms, reconciliation and post-incident fixes, often repeated because root cause is not always found
Lost lifetime value: a customer who shifts salary, bills or lending to a rival app takes future revenue with them
The invoice for downtime always arrives later than the outage.
5. With zero switching cost, trust is the only moat
Indian users are loyal to the outcome, not the app, and they punish friction quickly.
UPI apps look and work alike, and a customer can install a second one in minutes, so a failed payment at the till is a reason to try a rival. LocalCircles surveys show how friction-sensitive the market is: 75% of UPI users said they would stop using UPI if a transaction fee were introduced, and 38% make more than half their payments through it. The same sensitivity applies to failures. In one 2025 outage, 72% of Google Pay complaints and 86% of Paytm complaints on Downdetector were about failed payments.
Multi-homing: customers keep several UPI apps, so one failure sends them to another
Everyday dependence: for a large share of users, UPI is the main way they pay
Public fallout: outages trigger social posts urging people to carry cash
Concentrated leaders: PhonePe and Google Pay lead volumes, so a lapse by the leader moves the whole market
In a market with zero switching cost, trust is the only moat.
6. The RBI can switch off your growth
In India, repeated outages invite supervisory action that goes well beyond a fine.
The RBI’s Information Technology Governance, Risk, Controls and Assurance Practices Directions, 2023 took effect on April 1, 2024 and apply to banks, NBFCs and credit information companies. They require board-level IT governance, and enforcement shows the regulator is willing to restrict business lines, not just levy penalties.
HDFC Bank (December 2020): after repeated digital banking outages, the RBI halted new digital launches and the sourcing of new credit card customers
Kotak Mahindra Bank (April 2024): the RBI barred online and mobile onboarding of new customers and new credit cards, citing frequent outages over two years
Bharat Co-operative Bank (September 2025): a ₹3.75 lakh penalty for not reporting unplanned downtime that disrupted customer service
NPCI oversight: performance guidelines and penalties for banks that persistently underperform on UPI success rates
In India, resilience failures can cost you your growth plan.
7. Concentration and third-party risk multiply the blast radius
India’s payment stack is efficient because it is shared, and risky for the same reason.
NPCI is the sole operator of UPI, and an earlier RBI plan for additional umbrella entities was shelved, which industry observers flag as concentration risk. Behind it sit 741 live banks and a web of technology providers. Every payment hits core banking systems at both ends, so a weakness anywhere in the chain becomes everyone’s problem, and a fintech cannot fix what it does not control.
Single rail: one operator and one switch carry the bulk of retail digital payments
Bank readiness: in 2020, nine of the ten large banks with the highest failure rates were state-backed, a legacy-capacity signal that still matters at peak
Vendor dependence: the August 2024 C-Edge ransomware attack disrupted UPI for over 300 small banks
Network faults: a suspected SD-WAN glitch added to the April 2, 2025, disruption
You cannot outsource the customer’s disappointment.
8. Resilience is the next competitive advantage
The Indian fintechs that win will engineer for peaks and failure and treat reliability as a product feature.
The practical shift is from reacting to outages to designing against them. That means service level objectives (SLOs) for the journeys customers care about, tracked on latency percentiles as well as availability, because a p99 target exposes the slow tail that averages hide. It also means rehearsing India’s known stress points and telling customers the truth when something breaks, since users have asked for clearer communication from banks and NPCI during outages.
Measure what customers feel: SLOs for UPI collect, pay and status-check flows, with p95 and p99 latency targets alongside success rate
Plan for the calendar: load-test for financial year-end, festive-season sales, salary days and month-end, not average days
Degrade gracefully: idempotent retries, clear pending-status screens, and fallbacks such as UPI Lite for small payments
Report and rehearse: log incidents to RBI as required, run game days, and publish plain-language status updates
Reliability is the feature customers never notice until it is missing.
The bottom line
In India, speed and uptime are the first things customers judge and the last things they forgive.
With UPI moving over 800 million payments a day, a small failure rate is a very large absolute number; regulators are willing to restrict growth after repeated outages, and users can switch apps in seconds. Counting only lost fees understates the damage, especially in a zero-MDR market.
Count the full cost: engagement, support, compensation, regulatory and brand
Watch the tail: slow and declined payments are where trust erodes quietly
Plan for the peaks: year-end, festivals and salary days are predictable stress tests
Own resilience: make it a board-level metric, not an operations afterthought
Disclaimer: This article is for information purposes only. It is not financial, legal, regulatory, or investment advice. Figures are drawn from public sources and calculations dated in the text and may change; readers should verify them independently and consult qualified professionals before acting.
Fintech in India has grown up. UPI processes billions of transactions a month. Neo-banks and lending apps hold sensitive data for millions of people. Attackers follow the money. Regulators have moved too. “Best practice” has become enforceable, audit-ready requirements.
If you run a fintech app in 2026, three forces shape your security roadmap:
RBI regulations. Rules keep tightening around digital payment security, IT governance, outsourcing and authentication.
PCI DSS v4.0.1. Every formerly “future-dated” requirement is now mandatory.
Zero Trust architecture. It is now the default way to build systems that survive a breach.
This post explains what each one means for your web application. It also shows how to turn compliance from a yearly scramble into a continuous practice.
1. The RBI landscape: what your web app is measured against
RBI’s expectations are spread across several documents. They apply to banks, NBFCs, payment aggregators, PPI issuers and their service providers. These documents affect web and app security the most:
Master Direction on Digital Payment Security Controls. Baseline controls for web, mobile and card-not-present channels. It covers secure development, fraud monitoring and customer protection.
Master Direction on IT Governance, Risk, Controls and Assurance Practices. The board owns IT risk. You need an information security policy and regular independent assurance.
Master Direction on Outsourcing of IT Services. You stay accountable for your cloud, SaaS and API vendors. Expect due diligence, exit plans and audit rights.
Authentication directions for digital payments (effective 1 April 2026). Authentication becomes risk-based and moves beyond SMS OTP. Strong factors are still required.
Payment Aggregator guidelines. They cover data localisation, merchant onboarding checks, security audits and incident reporting.
Two adjacent regimes matter just as much:
CERT-In directions require you to report specified cyber incidents within 6 hours. Logs must be kept for 180 days within India.
The DPDP Act, 2023 and its Rules bring consent, purpose limitation and breach notification. Penalties apply for personal data failures. Obligations are being phased in. Plan now, not at the deadline.
Note: Regulatory circulars change often. Treat this post as orientation. Confirm current text and your entity category with compliance counsel or the latest RBI notifications.
What RBI-aligned security looks like in a web app
Strong, risk-based authentication with device binding and step-up checks for high-risk actions
Encryption in transit (TLS 1.2+, ideally 1.3) and at rest, with managed key rotation
Secure SDLC with code review, SAST/DAST and dependency scanning on every release
Real-time transaction and behaviour monitoring for fraud
Documented incident response with tested playbooks and regulatory reporting paths
Data stored and processed in line with localisation rules for payment data
Domain hygiene: banks are moving to the .bank.in domain and NBFIs to .fin.in, which helps customers tell real sites from phishing look-alikes
2. PCI DSS v4.0.1: the grace period is over
PCI DSS v4.0 was retired at the end of 2024. The future-dated requirements became mandatory on 31 March 2025. They apply if you handle, store or transmit cardholder data. They also apply if you can affect its security. Every assessment now covers them.
The requirements that most often catch web teams out:
Area
What changed
Practical action
Payment page scripts (6.4.3, 11.6.1)
You must inventory, authorise and justify every script on payment pages, and detect unauthorised changes
Use a strict Content Security Policy, Subresource Integrity, and a tamper-detection tool for checkout pages
Authentication (8.x)
MFA required for all access into the cardholder data environment, stronger passwords and phishing-resistant options encouraged
Enforce MFA for admins, developers and vendors, and move toward passkeys/FIDO2
Web application protection (6.4.2)
Automated technical solution for public-facing apps
Deploy and tune a WAF; do not run it in log-only mode forever
Targeted risk analysis (12.3.1)
Flexible frequencies must be justified in writing
Document why your scan, review and rotation cycles are what they are
Logging and monitoring (10.x)
Automated log review and faster detection of failures in critical controls
Centralise logs in a SIEM and alert on control failures
Scope and inventory (12.5.2)
Documented scope confirmed at least every 12 months
Keep an up-to-date data-flow diagram and asset inventory
Pro tip: The best way to shrink PCI scope is to avoid touching card data. Hosted fields, iframes and tokenisation from a PCI-validated gateway help. They can turn a heavy SAQ D into a lighter SAQ A or A-EP. The script-integrity rules still apply to your payment page.
3. Zero Trust: the architecture that makes compliance easier
Zero Trust is not a product. It is a design principle: never trust, always verify. Network location no longer grants access. Every request is authenticated, authorised and evaluated in context.
Use single sign-on with MFA for staff. Use short-lived tokens for services. Offer passkeys to customers wherever possible. Most breaches begin with stolen or phished credentials. Identity is your new perimeter.
Least privilege, everywhere
Use role-based and attribute-based access control. Elevate production access just in time. Review access regularly. Never share admin accounts.
Micro-segmentation
Separate the cardholder data environment, core banking connectors, internal tools and the public web tier. A compromised marketing site should never reach your ledger.
Continuous verification
Check device posture, location, behaviour and risk score on every sensitive action. Use step-up authentication for new payees, large transfers and contact detail changes.
Assume breach
Design for containment and fast recovery. Keep immutable logs. Keep encrypted backups and test restores. Store secrets in a vault. Rehearse your runbooks.
These pillars map almost one-to-one onto RBI’s access control and monitoring expectations. They also map onto PCI DSS requirements 7, 8 and 10. Build once. Evidence many times.
4. The 2026 threat picture for fintech web apps
What your security team should be watching this year:
API abuse and broken object-level authorisation. This is the top cause of data exposure in financial apps. Test every endpoint for privilege escalation.
Web skimming and supply-chain attacks. One compromised third-party script can silently harvest card data from a checkout page.
Account takeover and credential stuffing. Bots test leaked passwords at scale. Use rate limiting, bot management and breached-password checks.
AI-assisted phishing and deepfake social engineering. Voice and video impersonation is now cheap. Add out-of-band verification for sensitive operations.
Mule accounts and authorised push payment fraud. Pair behavioural analytics with consent-aware friction at the right moments.
Third-party and fourth-party risk. Your KYC provider’s vendor is also your risk.
Prompt injection and data leakage in AI features. Treat model inputs and outputs as untrusted. Keep customer data out of prompts where you can.
Longer-term cryptographic risk. Inventory where you use RSA and ECC. Be ready for post-quantum migration.
5. A practical 90-day roadmap
Days 1 to 30: See clearly
Build an asset, data-flow and third-party inventory
Map your obligations: RBI directions that apply to your licence, PCI scope, DPDP and CERT-In
Run a gap assessment and rank findings by risk
Days 31 to 60: Close the biggest gaps
Enforce MFA and remove standing admin access
Lock down payment pages with CSP, SRI and script monitoring
Put a WAF and API gateway with rate limiting in front of public endpoints
Centralise logs and set up alerts for the top 10 attack patterns
Days 61 to 90: Operationalise
Add SAST, DAST, SCA and secrets scanning to CI/CD with release gates
Run an incident-response tabletop covering the 6-hour CERT-In clock
Commission an independent penetration test and fix criticals before launch
Start automated evidence collection so audits stop being a fire drill
6. Compliance is not security, but it can fund it
A passed audit is a snapshot. Attackers do not care about your certificate. The best teams treat compliance as a by-product of good engineering:
Shift left. Run security checks in pull requests. Do not wait for the week before the audit.
Automate evidence. Use policies-as-code, infrastructure-as-code and continuous control monitoring.
Measure what matters. Track mean time to detect, mean time to patch criticals, share of services behind MFA, and third-party risk coverage.
Train people. Teach developers secure coding. Teach support teams about social engineering. Remind leaders of their accountability under RBI’s governance expectations.
Key takeaways
RBI, PCI DSS and DPDP overlap heavily. Design one control framework. Map it to all three.
PCI DSS v4.0.1 is fully in force. Payment-page script control and MFA are non-negotiable.
Zero Trust gives you one architecture for access, segmentation and monitoring.
Vendors are your risk. Contract for audit rights, breach notification and exit.
Rehearse incident response. Six hours to report is not much time.
Ready to make your fintech app audit-ready?
Our team at Fermion Infotech helps fintechs, NBFCs and payment companies run gap assessments, penetration tests and Zero Trust roll-outs. We align them to RBI and PCI DSS.
Disclaimer: This article is for general information only. It is not legal or regulatory advice. Always check the latest RBI, PCI SSC and other official publications. Consult qualified advisors for your situation.
Co-lending is gradually becoming an important part of the lending landscape, particularly as banks and other regulated lenders look to combine their respective strengths in customer reach, origination, funding and risk participation.
At first glance, the concept appears simple.
Two lenders agree to finance a borrower. The loan is shared between them in an agreed proportion. Interest and other economic benefits are shared, and so is the credit risk.
But from a banker’s perspective, co-lending is much more than simply dividing a loan between two lenders.
The moment two lending institutions participate in the same credit exposure, the traditional lending lifecycle changes significantly.
A conventional loan largely follows a linear relationship:
Bank → Borrower → Disbursement → Repayment → Closure
A co-lending relationship is different.
There is still one borrower and one underlying credit requirement, but behind the loan there are two participating lenders, separate financial exposures, agreed responsibilities, shared cash flows and a continuing requirement to keep both lenders aligned throughout the life of the facility.
Understanding this lifecycle is important for bankers who will increasingly deal with such portfolios.
It Starts Before the Borrower Arrives
In conventional lending, the relationship effectively starts when the borrower approaches the bank.
In co-lending, an important relationship already exists before that happens — the relationship between the participating lenders.
The lenders first need to agree on the broad framework within which loans will be originated and shared.
They need to decide what type of borrowers and products will be covered, which lender will originate the loans, how the credit exposure will be shared, who will interact with customers, who will service the loans and how collections, income, defaults and recoveries will be dealt with.
This arrangement becomes the foundation for every loan subsequently originated under the program.
This is the first important distinction:
A co-lending loan operates within two relationships — the lender-to-lender relationship and the lender-to-borrower relationship.
Both must remain aligned throughout the loan lifecycle.
Who Faces the Customer?
Normally, one lender takes the lead in originating the customer relationship.
It may source the borrower, collect the application, undertake KYC and due diligence, evaluate the proposal, arrange documentation and remain the principal point of contact for the borrower.
The partner lender may not have the same level of day-to-day interaction with the customer, but it nevertheless has a real financial exposure to the borrower.
This creates an interesting operating model.
From the customer’s perspective, the experience should remain reasonably seamless.
Behind the scenes, however, both lenders need sufficient information to understand and monitor their respective exposure.
The customer-facing lender therefore does more than originate a loan. It becomes an important information and servicing bridge between the borrower and the participating lender.
Credit Assessment Remains Important to Both Lenders
The originating lender may undertake the detailed credit assessment, but participation in a loan cannot simply be treated as the purchase of a percentage of an unidentified pool.
The partner lender needs to know what it is participating in.
Depending upon the agreed arrangement, the partner may rely upon predefined eligibility and underwriting parameters or undertake additional assessment before taking its share.
This creates two related but distinct decisions:
Should this borrower be given the loan?
and
Should the partner lender participate in this particular exposure?
For bankers, this distinction becomes important because origination and participation are not necessarily the same credit decision.
One Loan — But Two Financial Exposures
Suppose a borrower receives a loan of ₹1 crore.
Under the agreed co-lending arrangement:
Lender A participates for 20% — ₹20 lakh
Lender B participates for 80% — ₹80 lakh
The borrower sees a ₹1 crore loan.
But each lender sees only its respective financial exposure.
This distinction must remain intact throughout the life of the loan.
If the borrower subsequently repays ₹10 lakh of principal, the overall outstanding may become ₹90 lakh, but the corresponding lender-level positions must also change.
The same principle applies to interest, overdue amounts, prepayments, write-offs and recoveries.
Co-lending therefore requires bankers to think about a loan simultaneously at two levels:
What does the borrower owe?
and
What is each lender’s economic interest in that amount?
Disbursement Is No Longer a Single-Lender Event
In conventional lending, the bank sanctions the facility and disburses the amount.
Under co-lending, funding itself becomes a coordinated activity.
Each lender has to contribute its agreed share, and the borrower ultimately needs to receive the intended disbursement.
Consider a ₹1 crore disbursement under an 80:20 arrangement.
One lender must ultimately fund ₹80 lakh and the other ₹20 lakh.
If one lender’s contribution is delayed, short-funded or incorrectly accounted for, the issue is no longer merely an internal reconciliation problem. It can potentially affect the borrower experience and the financial positions of both lenders.
The lenders therefore need visibility over:
What was supposed to be funded? What was actually funded? When was it funded? What was disbursed to the borrower?
Pricing Introduces Another Dimension
Pricing under co-lending can also be different from conventional lending.
The participating lenders may have different expected returns on their respective portions of the exposure. At the same time, the borrower needs clarity regarding the effective interest rate applicable to the loan.
This introduces the concept of a blended interest rate.
If 20% of a loan carries one lender’s economic rate and the remaining 80% carries another, the overall borrower pricing needs to appropriately reflect the agreed structure.
There is another commercial aspect that should not be confused with interest.
The customer-facing lender may perform sourcing, documentation, servicing, collection and other operational activities on behalf of the arrangement. Compensation for such activities may therefore form part of the commercial understanding between the lenders.
From a banker’s perspective, it is useful to distinguish:
return for assuming credit exposure
from
compensation for performing services.
The Real Operational Challenge Begins After Disbursement
Sanction and disbursement receive considerable attention in lending projects.
In my view, however, the more interesting challenge in co-lending begins after the money has been disbursed.
The borrower starts paying instalments.
Suppose an EMI of ₹1 lakh is received.
That ₹1 lakh is not simply an amount to be divided 80:20.
It may contain principal, normal interest, overdue interest, charges or other amounts.
Each component first needs to be appropriately recognised against the borrower’s obligation.
Only thereafter can the economic entitlement of the participating lenders be determined.
This creates an important distinction:
Collection from the borrower is one event. Distribution of that collection between lenders is another.
The two should not be confused.
Collection Distribution Becomes the Heart of the Operating Model
Consider a borrower payment of ₹1 lakh comprising:
Principal — ₹60,000 Interest — ₹38,000 Other charges — ₹2,000
Principal may broadly follow the lenders’ participation ratio.
Interest distribution may reflect their respective economic entitlement.
Some charges may belong to the servicing lender, while others may need to be treated differently under the agreement.
Therefore, blindly applying an 80:20 ratio to every rupee collected may not produce the correct result.
Every collection needs to answer three questions:
What has the borrower paid?
Against which obligation has it been appropriated?
How much belongs to each participating lender?
This is where a well-managed co-lending operation distinguishes itself from a simple loan-sharing arrangement.
And Then Comes Reconciliation
Once collections start moving, another banking discipline becomes critical — reconciliation.
The customer-facing lender may maintain the servicing position.
Each participating lender maintains its own financial exposure.
Collections move through the agreed banking arrangements.
Accounting entries are passed independently by the participating institutions.
All these positions must ultimately agree.
If the borrower has paid ₹5 lakh, one lender cannot recognise a reduction while the other continues to show the old exposure.
Similarly, differences in principal, accrued interest, overdue amounts or collections cannot remain unresolved indefinitely.
Co-lending therefore creates a continuing requirement for reconciliation rather than merely a month-end accounting exercise.
What Happens When the Borrower Does Not Pay?
This is where co-lending becomes even more interesting from a risk-management perspective.
Suppose an instalment becomes overdue.
Which lender identifies the default?
Who communicates it to the other lender?
How quickly?
What Days Past Due should each lender recognise?
What happens when the account moves into SMA or NPA classification?
A shared credit exposure cannot be effectively managed if one participating lender considers the account regular while the other considers the same underlying co-lending exposure stressed.
Timely exchange of delinquency and asset-classification information therefore becomes an essential part of the relationship.
In other words, co-lending requires not only financial synchronisation, but also risk synchronisation.
Recovery Raises a Different Set of Questions
When an account becomes seriously stressed, another practical issue arises.
Who pursues the borrower?
It would make little operational sense for two lenders to independently call the same borrower, issue conflicting instructions or initiate uncoordinated recovery action.
The co-lending arrangement therefore needs clarity regarding who leads collections and recovery.
But leading the recovery does not mean owning the entire decision.
Important decisions may require consultation or consent between lenders — restructuring, settlement, legal proceedings, enforcement of security, waiver, write-off or compromise.
The operating principle should therefore be clear:
One lender may lead the recovery process, but the economic interests and agreed decision rights of both lenders must be protected.
Recoveries Must Again Be Distributed
Suppose a stressed loan with ₹50 lakh outstanding ultimately produces a recovery of ₹30 lakh.
The question is not merely whether ₹30 lakh has been recovered.
Bankers must also determine:
What portion represents principal?
What portion represents interest?
Were recovery or legal expenses incurred?
How should those expenses be borne?
What amount belongs to each lender?
Does the recovery follow the original participation ratio, or are there other contractual considerations?
Once again, the same basic co-lending principle emerges:
A borrower-level transaction ultimately has to be translated into lender-level financial positions.
Collateral Cannot Be Forgotten
For secured loans, co-lending introduces another important dimension.
There may be one property, vehicle, plant or other security supporting a loan in which two lenders have economic interests.
The security therefore needs to be understood in the context of the overall exposure as well as the respective lender positions.
Bankers need continuing visibility over valuation, insurance, creation and perfection of charge, enforceability and eventual release of security.
The position becomes even more important during recovery.
Who initiates enforcement?
Who controls the security documents?
How are enforcement expenses shared?
How are sale proceeds distributed?
Collateral governance in co-lending therefore cannot be separated from exposure governance.
Even Loan Closure Requires Coordination
In conventional lending, when all dues are recovered, the account can normally proceed towards closure and release of security.
Under co-lending, one additional question needs to be answered:
Have both lenders been completely settled?
A borrower may have paid everything due, but there may still be an unsettled amount between the participating lenders.
Therefore, a co-lending relationship should ideally reach closure only when:
the borrower’s obligation is fully settled;
both lender exposures have been extinguished;
all collections have been distributed;
inter-lender settlements are complete;
reconciliation differences have been resolved; and
the security can be appropriately released.
Only then has the complete economic lifecycle ended.
Co-Lending Is Ultimately About Maintaining Multiple Views of the Same Loan
From a banker’s perspective, perhaps the simplest way to understand co-lending is through four questions that need to be answered throughout the life of every loan:
1. Borrower Position — What does the borrower owe?
2. Lender Position — How much of that exposure belongs to each lender?
3. Cash-Flow Position — What has been received and how has it been distributed?
4. Risk Position — How is the underlying exposure performing and classified?
If these four positions remain aligned, co-lending can operate smoothly.
If they begin to diverge, operational, accounting, customer-service and regulatory issues can quickly emerge.
From Loan Sharing to Lifecycle Management
Co-lending should therefore not be viewed simply as:
“Bank A funds 20% and Bank B funds 80%.”
That is only the starting point.
The co-lending model is illustrated hereunder:
As co-lending portfolios grow, the challenge for banks will increasingly shift from entering into co-lending arrangements to managing them efficiently throughout their lifecycle.
The institutions that do this well will need clear allocation of responsibilities, disciplined information sharing, transparent lender-wise exposure tracking, robust collection distribution, synchronised risk monitoring and strong reconciliation.
The borrower may experience one loan.
But behind that loan are two balance sheets, two risk positions and a continuing financial relationship between the participating lenders.
Vehicles as Collateral: What ESG Information Should Banks Capture?
Vehicle finance represents one of the most widespread forms of secured lending—from personal cars and two-wheelers to commercial fleets, buses, trucks and specialised industrial vehicles.
Traditionally, banks assess vehicle collateral through parameters such as make, model, age, registration, ownership, insurance, market value and depreciation.
These remain fundamental. But as sustainability considerations become increasingly relevant to financial institutions, another question emerges:
Should banks also understand the sustainability characteristics of the vehicles they finance and hold as collateral?
In our opinion, increasingly yes—but the assessment needs to remain practical, proportionate and relevant to lending and collateral risk.
Vehicle ESG Assessment Is More Than EV vs ICE
The first sustainability distinction that comes to mind is usually:
Electric Vehicle or Internal Combustion Engine Vehicle?
That is important, but it is only the starting point.
Two diesel commercial vehicles of similar age and value may have very different emission standards, fuel efficiency, utilisation patterns and remaining economic lives.
Similarly, two electric vehicles may differ significantly in battery technology, battery health, remaining warranty, charging compatibility and expected residual value.
For a banker, the sustainability framework should therefore focus on information that could influence the vehicle’s economic life, regulatory acceptability, operating viability, marketability and residual value.
What Information Should Banks Capture?
I believe a common vehicle sustainability dataset should cover the following broad areas:
Vehicle Identity & Classification
Propulsion & Fuel Technology
Emission & Environmental Performance
Energy / Fuel Efficiency
Vehicle Age & Remaining Useful Life
Regulatory & Environmental Compliance
Safety & Social Parameters
Insurance & Climate Protection
Maintenance & Operating Condition
End-of-Life & Circularity
Green / Sustainable Finance Classification
ESG & Sustainability Risk Indicators
The detailed data requirement can then vary according to whether the financed asset is a personal vehicle, commercial vehicle, fleet or specialised industrial vehicle.
Personal Vehicles: Keep It Proportionate
For a personal car or two-wheeler loan, an extensive ESG questionnaire would neither be practical nor proportionate.
The bank can concentrate on a relatively small dataset covering:
Fuel/propulsion type
Emission standard
Fuel or energy efficiency
Vehicle age
EV battery information, where applicable
Insurance
Regulatory compliance
Green vehicle classification
Much of this information may already be available through vehicle documentation and the loan-origination process.
The objective should therefore be data enrichment—not additional customer paperwork.
Commercial Vehicles Require Deeper Assessment
The sustainability dimension becomes considerably more important for trucks, buses, taxis, delivery vehicles and commercial fleets because their utilisation and environmental footprint can be substantially higher.
In addition to the common dataset, banks may consider information relating to:
Annual kilometres travelled
Fuel consumption
Load/capacity utilisation
Emission performance
Maintenance history
Fleet age
Route and operating profile
Alternative-fuel readiness
At the fleet level, the bank can potentially derive indicators such as EV share, low-emission vehicle share, average fleet age, fuel mix and fleet transition progress.
This begins to convert individual collateral data into meaningful portfolio intelligence.
Industrial & Special-Purpose Vehicles Need Another Layer
The same questionnaire cannot reasonably assess a passenger car, an excavator and a large mining dumper.
Industrial and specialised assets such as excavators, loaders, cranes, forklifts, construction equipment, mining vehicles and agricultural equipment require additional consideration of:
Engine and propulsion technology
Fuel consumption
Operating hours
Energy efficiency
Emission standards
Equipment age
Maintenance condition
Remaining economic life
Availability of lower-emission alternatives
This reinforces an important principle:
Vehicle sustainability assessment should be asset-category driven.
EVs Introduce a New Collateral Dimension—the Battery
For electric vehicles, the battery can represent a significant component of both the vehicle’s economic value and sustainability profile.
Banks may therefore progressively need visibility into:
Battery Type | Capacity | Age | Health | Remaining Warranty | Ownership | Replacement Economics | Recycling / Second-Life Potential
Over time, battery condition could become relevant not only to ESG assessment but also to vehicle valuation and residual-value estimation.
Safety Is Also Part of ESG
Vehicle sustainability should not be viewed only through the environmental lens.
The Social dimension can include vehicle and operator safety.
Depending upon the vehicle category, relevant information may include safety rating, mandatory safety features, accident history, fitness certification, inspection status and driver-assistance systems.
For buses, trucks and industrial vehicles, operational safety assumes even greater importance.
Don’t Create Another Manual Data-Collection Exercise
One of the most important design principles should be:
Do not ask the borrower to provide information that the bank can reliably obtain elsewhere.
Vehicle sustainability information can potentially be assembled from:
Loan origination data
Vehicle registration and documentation
Manufacturer specifications
Insurance information
Inspection and valuation data
Commercial-fleet telematics
External databases / APIs
Customer-reported information
System-derived calculations
The collateral management platform should progressively bring this information together, rather than create another standalone questionnaire.
From Individual Vehicles to Portfolio Intelligence
Once structured sustainability information becomes available, banks can potentially analyse their vehicle collateral portfolios across dimensions such as:
EV versus ICE exposure
Fuel-type distribution
Emission-standard distribution
Average age of vehicle collateral
High-emission vehicle exposure
Green vehicle financing
Battery-related collateral exposure
Commercial fleet transition risk
Vehicles approaching technological or economic obsolescence
This is where sustainability data begins to move from data collection to risk intelligence.
The Banker’s Perspective: Proportionality Matters
The purpose should not be to transform every vehicle loan into a complex ESG assessment.
A practical framework could follow:
Personal Vehicle → Lightweight Sustainability Dataset
This is Series 3 of Sustainability in Collateral Management.
I am sharing this as a generic working framework with the broader banking, vehicle-finance, risk, ESG, sustainability and collateral-management community rather than as a prescribed methodology.
Views and practitioner perspectives on what additional sustainability information banks should consider for vehicle collateral would be most welcome.
An overall snapshot summarises the proactive approach to vehicle finance vis a vis sustainable finance.
Banks are indirectly involved in various manufacturing activities through funding; therefore, ascertaining sustainability aspects is critical when orienting their funds towards environment friendly activities. Evaluating sustainability in respect of manufacturing assets presents a significantly more complex challenge.
A manufacturing facility cannot be assessed only by asking whether the building is energy efficient, exposed to flooding, adequately insured or environmentally compliant.
The industrial activity taking place within that property can itself materially influence the sustainability and risk profile of the collateral.
A steel plant, pharmaceutical unit, automobile factory, textile processing facility and food-processing plant may all be classified broadly as manufacturing assets—but their environmental footprint, resource dependencies, regulatory requirements and transition risks can be substantially different.
From a banker’s perspective, this raises an important question:
How do we build a manufacturing sustainability assessment that is comprehensive without creating hundreds of different questionnaires?
A Layered Approach to Manufacturing ESG Assessment
In our view, the solution lies in a layered data architecture.
Instead of attempting to design one universal questionnaire—or a completely separate questionnaire for every industry—the assessment can be progressively constructed through four layers:
1. Common Manufacturing ESG Dataset: A baseline dataset applicable to manufacturing facilities irrespective of the industry.
2. Industry Group-Specific Dataset: Additional parameters based on the broad nature of the manufacturing activity—for example, metals, chemicals, automotive, textiles or food processing.
3. Industry / Activity-Specific Dataset: Specialised parameters relevant to the actual manufacturing process.
4. Site & Location-Specific Risk Dataset: Physical climate and environmental risks associated with the geographical location of the facility.
The resulting assessment can therefore be represented conceptually as:
Manufacturing ESG Assessment = Common Manufacturing Data + Industry-Specific Data + Activity-Specific Data + Site/Climate Risk
What Should the Common Manufacturing Dataset Cover?
Before getting into industry-specific requirements, Indian banks need a consistent baseline. The common manufacturing ESG dataset that we have been working on covers the following major areas:
The accompanying infographic provides an illustrative view of the data points that may be considered within these areas.
Why This Matters to a Banker
For banks, sustainability data should ultimately contribute to better risk understanding rather than becoming another compliance-oriented data collection exercise.
A structured manufacturing sustainability profile can potentially support:
• Identification of environmentally sensitive collateral
• Assessment of physical climate vulnerability
• Identification of energy-, water- and carbon-intensive assets
• Monitoring of environmental clearances and regulatory compliance
• Assessment of transition risk
• Identification of green or transition-finance opportunities
• Monitoring of climate-related insurance protection
• Portfolio concentration analysis across ESG-sensitive industries
• Identification of assets requiring enhanced monitoring
• Future assessment of sustainability implications for collateral value and marketability
Absolute Numbers Alone May Not Tell the Story
Manufacturing assessment also requires greater emphasis on intensity metrics.
A large manufacturing facility will naturally consume more electricity, water and other resources than a smaller facility.
Therefore, recording only absolute consumption may not provide a meaningful basis for comparison.
Where appropriate, the framework should also derive measures such as:
Energy Intensity = Energy Consumption / Production Output
Water Intensity = Water Consumption / Production Output
Carbon Intensity = GHG Emissions / Production Output
This enables comparison across facilities and, more importantly, monitoring of improvement or deterioration in the sustainability performance of the same asset over time.
Data Collection Should Not Mean Manual Data Entry
Another important consideration is the source of sustainability information.
Not every data point should be manually entered.
A mature collateral sustainability framework should progressively distinguish between:
• Customer / User Reported Data
• Document-Sourced Data
• Existing Bank / Collateral Data
• External API / GIS-Sourced Data
• System-Calculated Data
For example, production and energy-consumption information may come from the customer or supporting documents, while flood, cyclone, earthquake, heat stress or water-stress exposure can increasingly be obtained from external geospatial sources.
Carbon intensity, water intensity and composite risk indicators can then be system-calculated.
This approach can substantially improve both the reliability and operational feasibility of ESG data collection.
A manufacturing company may have a corporate ESG rating, but an individual plant offered as collateral can have its own sustainability characteristics.
The plant may be:
• located in a water-stressed region,
• dependent on carbon-intensive energy,
• exposed to flooding,
• operating under environmental clearances approaching expiry,
• generating hazardous waste, or
• operating with significantly better sustainability characteristics than the borrower’s other facilities.
From Collateral Valuation to Collateral Resilience
Traditional collateral management asks:
What is the asset worth today?
A sustainability-enabled collateral framework introduces additional questions:
How environmentally sustainable is the asset?
How exposed is it to physical and transition risks?
Could these risks influence its future operations, insurability, marketability or value?
And perhaps most importantly:
How resilient is the collateral over the remaining life of the bank’s exposure?
This is where we believe sustainability can become an important additional dimension of modern collateral management.
The objective should not be to turn the collateral management system into a full-scale enterprise ESG platform.
Rather, it should ensure that sustainability characteristics capable of materially affecting collateral risk are identified, structured, monitored and made available to the bank’s wider risk-management ecosystem.
Your app runs smoothly on the office Wi-Fi. But what about the customer scrolling through it on a train, with patchy signal, on a three-year-old phone? Are they getting the same smooth experience — or are they giving up and closing the app?
You won’t know the answer unless you’re monitoring the right way. That’s where two powerful tools come in: Real User Monitoring (RUM) and synthetic testing. Both help you track website performance and app performance, but they do it in very different ways. And most businesses need both — not just one.
In this post, we’ll break down what each one does, how they’re different, and how to use them together to build a solid performance monitoring strategy.
What Is Real User Monitoring (RUM)?
Real User Monitoring, often just called RUM, tracks how actual visitors experience your website or app in real time. Every time someone opens your site, RUM quietly collects data in the background — page load speed, click response time, errors, device type, browser, location, and more.
Think of it as a fitness tracker for your website. It doesn’t guess how things are going. It watches what’s really happening, second by second, for every single user.
Why RUM matters:
Shows real performance across different devices, browsers, and locations
Captures issues that only happen for specific users (like slow load times only on iPhones, or only in certain countries)
Helps you understand the actual user experience, not a simulated one
Useful for spotting patterns tied to peak traffic hours
What Is Synthetic Testing?
Synthetic testing, sometimes called synthetic monitoring, works differently. Instead of watching real visitors, it uses automated scripts (bots) to simulate visits to your website at scheduled intervals — every 5 minutes, every hour, whatever you choose.
These scripts test specific actions, like loading the homepage, adding an item to a cart, or logging in. Since the test conditions stay the same every time, it’s easy to compare results and catch problems early — even before a single real user shows up.
Why synthetic testing matters:
Works 24/7, even when your traffic is low or zero (like at 3 AM)
Great for testing before a new feature or update goes live
Helps with proactive uptime monitoring and catching downtime instantly
Useful for testing critical user journeys (checkout, sign-up, login) on a consistent schedule
RUM vs Synthetic Testing: The Key Differences
Factor
Real User Monitoring (RUM)
Synthetic Testing
Data source
Real visitors
Simulated bots/scripts
Timing
Continuous, live
Scheduled intervals
Best for
Understanding actual user experience
Catching issues before users do
Needs traffic
Yes
No
Consistency
Varies by user/device
Controlled and repeatable
Neither one is “better.” They simply answer different questions.
RUM answers: “How is my site actually performing for real people, right now?”
Synthetic testing answers: “Is my site working the way it’s supposed to, all the time?”
Why You Need Both (Not Just One)
Relying only on synthetic testing means you might miss real-world problems — like a slow-loading page that only affects users in a specific region or on a specific browser. Your automated test might pass every time, but real customers could still be struggling.
Relying only on RUM means you’re stuck waiting for real users to hit a problem before you know about it — including at 2 AM when nobody’s around to test manually.
Together, they give you full-picture visibility:
Synthetic testing catches issues early, before launch, and monitors uptime around the clock.
RUM confirms how real users are actually experiencing your site after it goes live.
This combined approach is often called a hybrid monitoring strategy, and it’s quickly becoming the standard for businesses serious about website performance and application reliability.
How to Build a Simple Monitoring Strategy
If you’re just getting started, here’s a simple approach:
Set up synthetic tests for your most important user journeys — homepage load, login, checkout, search.
Add RUM tracking across your site or app to capture real user data continuously.
Set alerts for both — so you know immediately if something breaks, whether it’s a scheduled test failing or a spike in real user errors.
Review the data regularly, not just when something goes wrong. Look for slow trends, not just outages.
Test across devices and regions, since performance can vary a lot depending on where and how people access your site.
Frequently Asked Questions
1. What is the main difference between RUM and synthetic testing?
Real User Monitoring (RUM) tracks how actual visitors experience your website in real time. Synthetic testing uses automated scripts to simulate visits on a fixed schedule. RUM shows what’s really happening; synthetic testing checks that things are working as expected.
2. Can I use only synthetic testing and skip RUM?
You can, but you’ll miss real-world issues — like slow performance for users on a specific device, browser, or location. Synthetic tests only check the scenarios you’ve programmed, not everything a real visitor might run into.
3. Is RUM enough on its own, without synthetic testing?
Not really. RUM depends on having real traffic, so it won’t catch problems during low-traffic hours, and it can’t test something before it goes live. Synthetic testing fills that gap by running around the clock.
4. Does RUM slow down my website?
A well-implemented RUM script is lightweight and runs in the background, so the impact on load time is minimal. Most modern monitoring tools are built to be as unobtrusive as possible.
5. How often should synthetic tests run?
It depends on how critical the page or action is. Homepage and login checks might run every few minutes, while less critical pages can be checked hourly. Most teams start frequent and adjust based on results.
6. Which one is better for SEO and Core Web Vitals?
RUM is more useful here, since Google’s Core Web Vitals are based on real user data. Synthetic testing is still helpful for catching performance regressions early, before they affect real users and your rankings.
7. Do small businesses need both, or is that overkill?
Even small websites benefit from both. Synthetic testing can be set up cheaply for key pages, while basic RUM tools now come built into many analytics and hosting platforms. You don’t need an enterprise budget to start.
Final Thoughts
Website speed and reliability directly affect user experience, conversions, and search rankings. Choosing real user monitoring and synthetic testing isn’t really the right question — the real question is how to use both together, so you catch problems before they cost you customers and understand exactly what your real users are going through.
At the end of the day, good website performance monitoring isn’t a one-time setup. It’s an ongoing habit — one that helps you build faster, more reliable websites and apps that people actually enjoy using.
Need help setting up a performance monitoring strategy for your website or app? Get in touch to learn more.
The Global Fintech Fest (GFF) 2026 — the world’s largest fintech gathering — is underway in Mumbai from September 8 to 11, bringing together policymakers, regulators, investors, and technology leaders from more than 80 countries. This year’s theme, “Potential to Impact: Agentic AI, Tokenisation, Quantum — Trusted, Connected, Global Systems for Inclusive Finance,” is a clear signal of where the industry is headed next.
For leaders and product teams building fintech apps, GFF 2026 is not just a conference to watch from the sidelines — it’s a preview of the technology decisions you’ll be making over.
Here are five themes from this year’s Global Fintech Fest, and what they actually mean for the product you’re building.
A major focus at GFF 2026 is agentic AI — systems that don’t just answer questions but take actions on a user’s behalf: initiating transfers, flagging fraud, negotiating terms, or managing a portfolio autonomously within set guardrails.
If your roadmap still treats AI as a support widget bolted onto the UI, it’s time to rethink. The next generation of fintech apps will need AI woven into core workflows, with careful attention to permissions, auditability, and fail-safes — not just a chat window in the corner.
2. Tokenisation Is Becoming Infrastructure, not a Feature
Tokenisation — representing real-world assets, payments, credentials, or securities as digital tokens — was front and center in GFF 2026 discussions this year. It’s no longer a niche capability for crypto-native platforms; it’s becoming a baseline expectation for secure, composable financial systems.
Payment and asset-handling architecture needs to be built with tokenisation-readiness from day one. Retrofitting it later is expensive and risky. This affects everything from how you store sensitive data to how your APIs are designed to integrate with card networks, banks, and settlement layers.
3. Quantum-Readiness Is No Longer a Far-Off Concern
Quantum computing made it onto the main stage this year, with discussions focused on both its potential to transform financial modeling and the very real cybersecurity risk it poses to today’s encryption standards.
You don’t need a quantum computer to start preparing. Forward-looking teams are already auditing their cryptographic standards and planning for “quantum-safe” migration paths — especially for apps handling long-lived sensitive data like KYC records or transaction histories.
4. Trust and “Connected Systems” Are the New Differentiator
A recurring thread across GFF sessions was the idea of trusted, connected, global systems — the notion that fintech’s next phase of growth depends less on flashy features and more on interoperability, transparency, and regulatory alignment across borders.
Compliance and interoperability can’t be an afterthought bolted on before launch. Building with open standards, clear audit trails, and regulator-friendly architecture from the start will save significant rework — and it’s increasingly what investors and enterprise partners expect to see.
5. Inclusive Finance Is a Design Requirement, Not a CSR Line Item
GFF 2026’s framing — “Potential to Impact” — emphasizes that emerging tech should produce measurable outcomes for everyday citizens and small businesses, not just enterprise clients. Financial inclusion was discussed as a design principle, not an afterthought.
Accessibility, low-bandwidth performance, multilingual support, and simple onboarding flows aren’t nice-to-haves — they’re what separates fintech apps that scale across diverse markets from those that stay niche.
Where This Leaves Product Teams
Every trend coming out of GFF 2026 points to the same conclusion: the technical foundation of a fintech product matters more than ever. AI-driven features, tokenisation-ready payment rails, quantum-safe security, and inclusive design all need to be architectural decisions made early — not patches applied after launch.
That’s exactly where we come in.
At Fermion, we help fintech leaders and financial institutions design and build secure, scalable web and mobile platforms — from AI-powered features to payment infrastructure built for what’s next. If you’re planning a fintech product and want a technology partner who’s thinking about these shifts alongside you, let’s talk →[email protected]
Sources: Global Fintech Fest 2026 official event materials and press coverage, September 2026.
If your site scores well on Lighthouse but still feels laggy, the moment someone clicks a filter, opens a menu, or submits a form — you have an INP problem, not a load-time problem.
Interaction to Next Paint (INP) became a Core Web Vital, replacing First Input Delay (FID). Unlike FID, which only measured the very first click on a page, INP measures how quickly a webpage responds to every user interaction throughout an entire visit. That distinction matters more than most teams realize: a site can post an excellent FID score and still feel broken the moment heavy JavaScript runs mid-session instead of at load.
This guide covers what INP measures, why most existing INP content stops short of being useful, and the specific, framework-level fixes that bring real sites under the 200ms threshold — with code.
What Is Interaction to Next Paint?
Interaction to Next Paint measures how quickly a website responds to user interactions like clicks or key presses — specifically, how much time elapses between a user interaction and the next time the user sees a visual update on the page. It captures clicks, taps, and keyboard input across the entire session, not just the first one.
A good INP score is 200 milliseconds or less at the 75th percentile. Google scores your site using the CrUX (Chrome User Experience Report) field data collected from real visitors, not a synthetic lab test — so INP reflects what actual users on actual devices experience.
INP Value
Rating
0–200ms
Good
200–500ms
Needs Improvement
500ms+
Poor
Despite two years of attention, this metric is still the industry’s biggest blind spot. Roughly 40% of origins on mobile still fail to meet INP thresholds heading into 2026, making it a persistent liability for search rankings and user experience alike.
The Three Phases of an Interaction
Every user interaction that INP measures is made up of three phases, and knowing which one is broken tells you exactly which fix to reach for.
Input delay — the time between the user’s action and the browser starting to process it. Usually caused by the main thread already being busy with another task.
Processing duration — the actual work your event handler, framework, and re-renders do in response to the interaction.
Presentation delay — the time it takes the browser to paint the visual result to the screen once processing is done.
INP reports the longest of these full interactions observed during a session, so fixing your average case isn’t enough — you have to eliminate the outliers.
Why INP Is Hard to Fix (and Why It’s Worth Fixing)
INP is a frontend queueing problem in disguise. Think of the main thread like a single-lane road: when a large task — a 10,000-row filter, a heavy re-render, a synchronous JSON parse — is running, every user interaction has to wait in line behind it before the browser can even start responding.
The business cost is measurable, not theoretical. Improving INP from 500ms to 200ms correlates with a 22% improvement in user engagement metrics including time on page and return visits, and a widely cited INP case study on web.dev found a 7% sales lift after a company implemented responsiveness improvements tied to INP. For a checkout flow, a lead-gen form, or a SaaS dashboard, every sluggish click is a small tax on conversion.
How to Measure INP (Before You Touch Any Code)
Don’t optimize blind. Start with field data, then move to lab tools to reproduce and debug:
Chrome UX Report (CrUX) / PageSpeed Insights — real-user field data at the URL and origin level.
Search Console → Core Web Vitals report — flags which page groups are failing INP in production.
web-vitals JS library — drop it into your app to log real INP scores per session, segmented by device or route.
Chrome DevTools Performance panel — record a session, reproduce the slow interaction, and inspect the flame chart.
Long Animation Frames (LoAF) API — pinpoints exactly which script or function is responsible for a slow frame, which is far more actionable than a generic “long task” warning.
Fix based on the interaction that’s actually failing in the field, not the one that’s easiest to reproduce locally.
Fixing Input Delay
Input delay usually means the main thread is tied up before the click even registers.
Code-split and lazy-load non-critical JavaScript so it isn’t competing with interaction handling.
Defer or async non-essential third-party scripts (chat widgets, analytics, ad tags) — audit these first, they’re a common hidden culprit.
Break up long tasks at their source using scheduler.yield(), which pauses execution to let the browser handle pending high-priority work, then resumes where it left off:
async function yieldToMain() { if (‘scheduler’ in window && ‘yield’ in window.scheduler) { return await window.scheduler.yield(); } return new Promise((resolve) => setTimeout(resolve, 0)); }
async function processLargeDataset(items) { for (let i = 0; i < items.length; i++) { processItem(items[i]); if (i % 50 === 0) await yieldToMain(); // give the main thread breathing room } }
This single fix — breaking long tasks into smaller chunks with scheduler.yield() — is often the highest-leverage change you can make, since it directly reduces the wait time for every pending interaction. Note browser support: Chrome and Edge support it from version 129, Firefox from 142, and Safari does not yet support it — always ship the setTimeout fallback shown above.
Fixing Processing Duration
This is where most of the real engineering work lives.
Debounce, don’t just throttle, expensive input handlers. Throttling still recalculates on an interval while the user types or scrolls; debouncing waits until they stop, which is usually the better trade-off for search and filter inputs. One team found that replacing a throttled scroll handler with a debounced one produced a noticeably smoother result because the calculation only ran once activity settled.
Move heavy computation off the main thread with Web Workers — JSON parsing, sorting large datasets, or client-side filtering are good candidates.
Virtualize long lists and tables so you’re only rendering what’s visible, not re-rendering thousands of DOM nodes on every keystroke.
Avoid layout thrashing — batch DOM reads and writes instead of interleaving them, which forces the browser into repeated synchronous reflows.
For a search filter over a large table, a typical fix chain looks like: debounce the input, paginate or virtualize the results, and render only the visible rows instead of re-rendering the full table on every keystroke.
Framework-specific notes
React: memoization, debouncing, concurrent rendering features, and virtualization are the techniques that most directly move the INP needle. React 18’s concurrent rendering yields back to the main thread roughly every 5ms during rendering to check for higher-priority work like user input, so upgrading rendering strategy alone can meaningfully help — but it isn’t a substitute for fixing genuinely long synchronous handlers.
Vue / Angular / vanilla stacks: the same principles apply — the framework name changes; the fix doesn’t. Identify the long task with the Performance panel or LoAF, then split, defer, or move it off-thread.
Fixing Presentation Delay
Once processing finishes, the browser still has to paint. Reduce this by:
Simplifying complex CSS selectors and reducing DOM depth in the area being updated.
Avoiding non-composited animations (animate transform/opacity instead of properties that trigger layout).
Giving users instant visual feedback (a pressed state, a spinner, a skeleton) — this doesn’t reduce the technical INP number by itself, but it’s good practice alongside the real fix, since perceived responsiveness matters for user trust even while you’re shipping the underlying fix.
The Prioritized INP Checklist
Work through these roughly in order of impact per hour of engineering time:
Audit and defer/async third-party scripts.
Find your top 3 slowest interactions in CrUX/Search Console field data.
Reproduce each in DevTools and identify the responsible script via the LoAF API.
Break up the long task with scheduler.yield() (with fallback).
Debounce expensive input handlers; replace throttling where appropriate.
Virtualize large lists/tables; paginate where virtualization isn’t feasible.
Move CPU-heavy work (parsing, sorting, filtering) to a Web Worker.
Re-measure with field data — lab data alone won’t confirm the fix.
FAQ
Is INP a ranking factor?
Yes — INP is one of Google’s three Core Web Vitals that impact search result rankings, alongside Largest Contentful Paint and Cumulative Layout Shift.
Can I test INP without real users?
Not directly — INP relies on real user interactions and is primarily measured through field data, since it requires an actual interaction to occur, though you can script interactions in a lab environment to approximate it.
What replaced FID, and when?
INP replaced First Input Delay as a Core Web Vital in March 2024.
The Bottom Line
INP rewards continuous engineering discipline, not a one-time fix. You can’t optimize only the initial page load and call it done — every interaction throughout the session counts toward your score. Teams that treat responsiveness as an ongoing budget, not a launch checklist, are the ones that stay under 200ms as their product grows.
If your team doesn’t have the bandwidth to profile long tasks, restructure rendering, and re-test against field data on an ongoing basis, that’s exactly the kind of performance engineering work we take on for clients — get in touch and we’ll audit your site’s real INP data before you spend a single engineering hour guessing.
For decades, banks have assessed property collateral primarily through a familiar set of parameters—ownership, title, valuation, location, marketability, legal enforceability and insurance.
But the risk environment surrounding physical assets is changing.
A property may have a strong market value today, but its long-term value, marketability, insurability and resilience could increasingly be influenced by environmental and climate-related factors.
Consider a few questions:
Is the property exposed to flooding?
Is it located in an area vulnerable to cyclones, earthquakes, wildfires or rising sea levels?
How energy-efficient is the building?
Does it have a recognised green-building certification?
Could environmental litigation or regulatory non-compliance affect its value?
Is adequate insurance available against material climate events?
These questions suggest that collateral management may need to evolve beyond maintaining a static financial and legal description of an asset.
Building a Sustainability Profile of Your Collateral
For properties and physical assets, we believe a structured ESG assessment should consider data across five broad dimensions:
Environmental Attributes
Climate Risk
Social Parameters
Governance Parameters
Financial Sustainability
The accompanying infographic provides an illustrative framework of the data points that banks may consider capturing under these dimensions.
A structured sustainability dataset can potentially help banks develop capabilities for:
Climate vulnerability assessment
ESG risk profiling
Green asset identification
Sustainability-linked financing assessment
Climate insurance monitoring
Portfolio-level sustainability analytics
Identification of collateral exposed to emerging environmental and climate risks
Borrower ESG Risk and Collateral Sustainability Risk Are Not the Same
An important distinction needs to be made between Borrower ESG Risk and Collateral Sustainability Risk.
A borrower may have a strong corporate ESG profile, while a particular property offered as collateral may still be located in a high flood-risk zone, have poor energy efficiency or face other environmental vulnerabilities.
Conversely, an individual asset may have strong sustainability characteristics even when the broader ESG profile of the borrower requires improvement.
Therefore, the sustainability profile of collateral deserves to be assessed as a distinct—but interconnected—dimension of enterprise risk and collateral management.
The Question for Banks Is Changing
As banks progressively strengthen their ESG and climate-risk frameworks, the question may no longer be only:
“What is the current value of my collateral?”
It may increasingly become:
“How sustainable and climate-resilient is my collateral—and what could that mean for its value over the life of the exposure?”
This is the first article in my series on Sustainability in Collateral Management.
In the next article, we will discuss the significantly more complex challenge of ESG assessment of manufacturing and industrial assets, including how a common manufacturing ESG dataset can be supplemented with industry-specific sustainability parameters.