Which products should your institution stop selling?

Walk into the head office of a mid-sized commercial bank, microfinance lender, or payment provider in Nairobi and ask the product team for their current tariff guide or product brochure. Most institutions will hand over a schedule listing between twenty and thirty-five separate deposit accounts, loan products, card tiers, and digital wallets. Then ask the chief financial officer a second question. How many of those products generate 85% of the institution’s active balances and net income? In almost every case, the answer sits between four and six.

The remaining twenty products rarely sit idle for free. Each one occupies a slot on the USSD menu or mobile app screen, carries its own general-ledger reconciliation routine, requires separate compliance and anti-money-laundering rules, and adds another set of edge cases whenever the engineering team upgrades the core banking system or connects to a new payment switch. Staff at the branch counter and the call centre are expected to remember the eligibility rules, fee tiers, and exception procedures for products they open twice a month.

Institutions accumulate product clutter because internal incentives reward launches and ignore maintenance. A product manager gets promoted for unveiling a youth account, a diaspora portal, or a niche merchant wallet. Nobody receives a bonus at the end of the year for shutting down three underused product codes and migrating 1,400 customers onto a cleaner core account. Over five or ten years, each chief executive and commercial director adds another layer to the shelf until the institution is spending a third of its technology and operations budget keeping yesterday’s experiments alive.

Counting what a dormant product costs to keep alive.

On a standard monthly management income statement, a dormant or low-volume product often looks harmless. If a specialized savings account or a standalone prepaid card brings in KSh 400,000 a month in fee income and pays out KSh 100,000 in direct interest or commission, management sees a positive line of KSh 300,000 and leaves the product alone.

That view misses the operational bills buried inside shared departmental budgets. To understand the underlying arithmetic, consider an illustrative product review at a mid-tier Kenyan lender offering 24 separate retail and business products. Four core products (the standard current account, the main mobile savings wallet, the salaried personal loan, and the SME working-capital facility) carry 88% of active customers and 91% of net operating income. At the bottom of the ledger sit six legacy products that together serve fewer than 4,500 active customers.

Those six tail products consume cash and staff hours across four departments every month.

First is technology and regression testing. Whenever the bank updates its mobile app, migrates messages to the ISO 20022 standard, or integrates electronic tax invoice validation under the Kenya Revenue Authority’s eTIMS rules, engineers must write and test code for every active product table in the core banking system. Four legacy loan and savings products with custom interest-calculation scripts add three weeks of manual regression testing to every major software release, delaying improvements to the main mobile app used by 600,000 customers.

Second is vendor and scheme minimums. One of the six tail products is a co-branded multi-currency prepaid card launched four years ago for frequent travellers. Only 680 cards remain active. Under the bank’s contract with its external card processor and international card scheme, the product carries a minimum monthly processing and licensing charge of US$4,500 (about KSh 580,000), plus an annual payment-card security audit costing KSh 1.8 million.

Third is manual back-office reconciliation and dispute handling. Another legacy product, a bespoke agricultural input wallet built for a donor-funded pilot that ended in 2023, runs on a separate sub-ledger that does not clear automatically through the main payment switch. A back-office clerk spends two hours every morning reconciling its settlement file by hand and handling failed transfer reversals.

Fourth is front-line confusion and customer error. When a mobile app or USSD tree displays twelve nearly identical account and loan options, customers pick the wrong menu item, pay into the wrong paybill, or apply for a facility they do not qualify for. Front-line staff spend hours explaining the difference between overlapping products instead of resolving real service problems.

Separating avoidable cash costs from allocated overhead.

When a finance team finally decides to prune the product shelf, it often makes the opposite mistake by relying on fully allocated cost accounting. In a conventional cost-allocation model, the finance department takes the bank’s fixed overheads (head-office rent, executive salaries, brand marketing, and the base core-banking licence) and spreads them across every product according to transaction volume or account numbers.

Allocated overhead distorts product decisions in two ways. It makes a high-volume, low-fee product look like a heavy loss-maker simply because it carries millions of transactions, loading it with a huge share of head-office rent and IT salaries that the bank would pay whether that product existed or not. At the same time, it lets a tiny, complex product escape scrutiny because its small transaction count attracts almost zero allocated overhead, hiding the dedicated vendor contract and manual staff hours required to keep it running.

Sound product rationalization ignores sunk corporate overhead and asks what actual cash expenses and staff hours will disappear from the bank’s ledger within 90 days if the institution stops selling the product and migrates its accounts.

Precursor illustrative product-profitability audit for a mid-sized Kenyan financial institution. Avoidable cash costs count only vendor bills, licences, and dedicated labour that cease or redeploy upon product closure.

Look at the fourth row in that table. Under a standard cost-allocation exercise, a finance committee would either add new transaction fees to the school-fees collection account or shut it down to stop a paper loss of KSh 1.98 million a month. Doing either would damage the balance sheet. The school-fees account costs only KSh 380,000 a month in avoidable running expenses while bringing in KSh 1.4 billion in low-cost deposit float during January, May, and September, feeding nearly a third of the bank’s profitable school-bus and infrastructure loan book.

Conversely, the travel prepaid card in row one looks acceptable under allocated accounting because its small customer base attracts only KSh 45,000 of shared overhead. Once the chief financial officer counts the US$4,500 monthly processor floor and the dedicated security audit, the card is burning KSh 420,000 in real cash every month for a feature the bank’s standard debit card already provides.

Tracing cross-subsidies and duplicated functions before cutting.

Before shutting any product down, management has to map how customers move money between accounts inside the institution. Financial services work as an interconnected balance sheet where one product gathers customers or cheap funding and a second product earns the margin.

A zero-margin salary account or a low-fee merchant till earns its keep if the institution converts a clear share of those users into working-capital borrowers, asset-finance clients, or money-market investors. That cross-subsidy argument fails when executives use it as a blanket excuse to defend every loss-making product without checking the ledger.

Testing a cross-subsidy claim requires a simple database query. Take all active holders of the product under review and check what other products they use across the institution, how much deposit float they leave overnight, and how much net revenue those linked accounts generated over the past twelve months. If 80 percent of the customers holding a loss-making youth wallet hold no other balance, generate no float, and graduate to no credit product after three years, the account is a dead-end subsidy.

Duplicated functions inside the same institution deserve the same scrutiny. Many Kenyan banks and fintechs run two or three parallel credit engines built by different teams in different years, such as a branch-based SME overdraft scored on spreadsheets, a mobile merchant advance scored with an external fintech partner, and a supply-chain invoice discounting portal on a separate platform. Maintaining three credit workflows means paying for three software maintenance contracts, running three separate credit-bureau reporting pipelines, and splitting the customer’s transaction history across three databases that do not talk to one another. Consolidating those three products onto a single underwriting engine and retiring the redundant product codes releases engineering and risk staff without taking a single shilling of credit away from viable borrowers.

Comparing algorithmic portfolio tools against a two-day ledger audit.

Software vendors increasingly pitch artificial intelligence portfolio optimization and customer-lifetime-value models to tell banks which products to cross-sell or retire. Those platforms ingest transaction rows, cluster customer behaviour, and output product profitability scores on a dashboard.

For most African banks, SACCOs, and fintechs, buying an AI analytics layer to clean up the product shelf gets the sequence backwards. An algorithm cannot read a paper vendor contract locked in the procurement safe to see that a card processor charges a US$4,500 monthly minimum. Nor can a machine-learning model interview the head of software quality assurance to learn that a legacy savings script from 2019 adds four days of manual testing to every mobile app update.

A two-day audit run jointly by the finance director, the chief operating officer, the head of technology, and the customer-service lead using existing database queries and procurement files will identify 90 percent of the dormant and duplicated products on the shelf at zero software cost.

Migrating affected customers without creating dormant-asset traps.

Stopping a financial product requires more care than pulling a consumer good off a supermarket shelf. Every active account represents a legal contract, a customer’s savings or working capital, and a set of regulatory rules.

In Kenya, the Central Bank of Kenya’s Prudential Guidelines on Consumer Protection require institutions to give customers clear advance written notice, typically at least 30 days, before discontinuing a product or materially altering its terms and tariff structure, while explaining the options available to the account holder. Careless product closures also create a long-term balance-sheet headache under the Unclaimed Financial Assets Act. If a bank freezes or closes a legacy product line without migrating the customer’s balance into an active account they already use, left-behind balances sit untouched until the statutory dormancy clock runs out. That forces the institution to track, report, and surrender the funds to the Unclaimed Financial Assets Authority (UFAA), with stiff compliance penalties if records are incomplete.

Retiring a product cleanly requires treating the migration as a customer-service project rather than an IT switch-off.

First, map every active customer on the retiring product to a surviving core product that offers equal or better terms. If a bank is closing a legacy savings tier or a niche mobile wallet, the receiving account must not hit the customer with higher monthly ledger fees or lower interest rates than the product they are leaving.

Second, move the plumbing automatically where the law and customer mandate allow. Standing orders, direct debits, salary routing, and historical statement access should carry over to the surviving account without forcing the customer to take a day off work to queue at a branch with fresh photocopies of their national ID card.

Third, brief the front line and give them authority to waive friction costs. Thirty days before the product closes, branch officers and call-centre agents need a one-page screen prompt showing which accounts are moving, why the change is happening, and how to waive any accidental transfer or closure charge on the spot.

Deciding which products to retire, merge, or protect for 2027.

Before signing off on 2027 product and technology budgets, a chief executive and chief financial officer can clear the institutional backlog by sorting every product on the tariff guide into three decisions.

• Retire and migrate any product where active usage has fallen below a viable threshold (such as fewer than 1,000 active transacting customers for a retail product), where avoidable vendor, compliance, and manual reconciliation costs exceed the direct cash revenue it brings in, and where the product does not serve as a verified deposit or lending feeder for the rest of the institution. Assigned to the Chief Financial Officer and Head of Operations, retiring these products means serving 30 to 60 days’ written customer notice, migrating active balances fee-free into a core product, terminating the underlying vendor licence, and removing the code from the live regression-testing suite.
• Merge and consolidate any product that duplicates the economic job of another product on the shelf, such as running three separate SME working-capital products or four overlapping retail savings tiers with slightly different minimum balances. Assigned to the Chief Commercial Officer and Chief Technology Officer, merging means keeping the single cleanest product engine, moving existing customers onto that single rulebook, and removing the duplicate options from the USSD tree, mobile app, and branch onboarding forms.
• Protect and rebuild low-margin feeder products, such as school-fee collection accounts, merchant settlement tills, or payroll accounts, that look weak under allocated overhead rules yet supply verifiable, low-cost deposit float or qualified borrowers to the core balance sheet. Assigned to the Head of Product and Chief Risk Officer, protecting these products means stripping out manual back-office steps to lower their direct running cost and wiring them directly into the institution’s primary lending and wealth workflows.

To stop the clutter from growing back next year, the board can adopt one permanent rule for the 2027 budget cycle. No product manager may launch a new product code unless the submission names the legacy product code it will replace, includes the contractual minimum vendor bills in the break-even math, and defines the exact customer and cash thresholds that will trigger automatic closure if the product misses its 12-month target. Pull your institution’s full list of active product codes tomorrow morning and set it beside the last 90 days of support tickets and vendor invoices. You will likely find next year’s engineering capacity sitting inside five products your customers stopped using years ago.

Companion Working Tool: Product Rationalization and Migration Decision Matrix

For Chief Executives, Chief Financial Officers, and Operations Leads to audit the existing product shelf before approving 2027 product and IT budgets.