Back to All Writing
    Sunday Observer2026-06-07

    The unified ERP decision that determines whether Sri Lanka scales

    Originally published in Sunday Observer on 2026-06-07.

    Read Original Article on Sunday Observer

    Core Argument

    Sri Lanka's recovery is gaining ground and owners who spent two years in survival mode are asking how to grow. The direct answer is that most cannot scale the way they plan to, and the reason is not market size, capital access or talent. It is that the operational architecture underneath the business is fragmented. Finance runs on one platform, inventory on another, procurement on spreadsheets, sales on a CRM nobody updates. Every attempt to grow adds complexity to a foundation never designed to carry it. The businesses that grew through the 2010s did so on improvisation, and those arrangements create the illusion of systems without the reality of integration. Data is entered in several places, which means it is wrong in at least one. Decisions get made on last month's numbers because this month's are still being compiled. Margins compress without explanation because cost data and revenue data are never in the same room at the same time. A unified ERP does not mean a complicated system. It means one platform where finance, inventory, procurement, sales, manufacturing and HR write to and read from the same data. A sales order updates inventory. An approved purchase order hits the accounts in real time. Month close becomes a function of the system rather than a manual exercise. What it eliminates is the gap between what a business is doing and what it knows it is doing, and at scale that gap is the difference between executing quickly and always catching up. The platform is not the variable that determines whether the investment pays off. The implementation partner is. A well-implemented lighter platform outperforms a poorly implemented enterprise one every time. Too many local firms treat implementation as deployment: configure the minimum, run one training session, hand over a system the client cannot extend or troubleshoot. The go-live date becomes the ceiling rather than the starting point.

    What I'd Revise Now

    One thing has happened since June that makes this column more urgent, and one thing in it I would now write differently. The tariff overhaul ratified in July halves CESS in 2026 on 693 intermediate and capital goods codes, with further cuts in 2027 and 2028. Input costs are falling for exactly the manufacturers this piece addresses. But a firm cannot capture relief it cannot see. If procurement sits on spreadsheets and costing is reconstructed monthly, cheaper inputs arrive as an unexplained margin variance rather than as a pricing decision. The businesses that unified will price against their new cost base within a quarter. The ones that did not will find out at year end. What I would drop is the revenue banding. Naming Odoo for one range, Acumatica for another and SAP above that reads as a selection rule, and it is not one. I said in the same column that the implementation partner is the variable that decides the outcome, then supplied a table that invites owners to choose on revenue instead. Those two things do not sit together. The banding is a rough guide to operational complexity, nothing more. A single-entity firm at two billion rupees may need less than a three-entity exporter at four hundred million. Diagnose the data flows, count the entities and the compliance obligations, then choose. The revenue figure is the least informative number in that decision, and I gave it the most prominence.

    Key Takeaways

    • The constraint on scale is fragmented operational architecture, not market size, capital or talent
    • Improvised systems built for a team of ten break at fifty, and the cost of deferring the fix compounds each growth cycle
    • Unified means one data set, so month close becomes a system function rather than a manual exercise
    • The implementation partner determines the return, not the platform
    • A partner unwilling to commit in writing to twelve-month operational outcomes is selling deployment, not scale

    Related Insights