Login
Register
Member Insights
IBM Consulting's Rahul Kumar, Joern Kropfgans and Rory Meffen explain how operators can escape continuous transformation cycles by building their BSS to establish control and change how they fund and govern modernization.
Rahul Kumar, IBM
Joern Kropfgans
Rory Meffen, IBM
Modernizing BSS for control and flexibility; not for the next generation of legacy
Communications service providers (CSPs) are investing in AI to grow, yet research from the IBM Institute for Business Value found that while 88% of telecoms executives expect AI to contribute to revenue by 2030, only 28% can identify where that revenue will come from. Even fewer are ready to capitalize on it without transforming their business support systems (BSS), and by the time BSS transformation programs deliver, the needs have usually changed.
In addition, every CSP's situation is different. Market dynamics, competition, sovereignty and regulation shape what each operator can sell, so no CSP can follow a peer's path and each has to adapt to its own market.
That leaves CIOs with a hard question: how do you build a BSS platform for a revenue model nobody can yet describe? CSPs have diversified into selling products beyond connectivity such as device insurance, energy and financial services. However, the next product has always been hard to predict, and in BSS estates built for today's products a new idea often means another expensive transformation program.
The transformation paradox
These programs take years to realize, and the target state is defined at the start. At completion, customer expectations, regulation, technology capabilities and business models have moved on. Teams then fall back on customizations and workarounds, new products and offers take months to launch, and technology gets added on top of the already rigid estate rather than built into it, raising licensing and operating costs and adding technical debt. Another transformation program then follows to fix the last one. It is little surprise that 42% of telecoms CEOs say their organizations have disconnected technology because of the pace of recent investments. In effect, many CSPs risk paying today for the legacy of tomorrow.
Designing for continuous change would break this cycle, but it is easier said than done.
Control decides how fast a CSP can change
A CSP that cannot change its BSS on its own schedule moves at its suppliers' pace. Standard functions such as rating, invoice generation and order management run well on proven platforms and are usually best consumed as they come. The data, the insights drawn from it and the workflows that act on them are what differentiates a CSP, however, most BSS platforms bury all three inside large applications, which makes change slow and expensive. Separating them, in line with TM Forum's Open Digital Architecture (ODA) and Open APIs (application programming interfaces), lets data be shared across domains, with insights driving decisions and workflows predicting and responding to customers stated and implied needs.
Replacing your BSS platform is not the decision that matters. What matters is who (CSP or supplier) owns the layer where agents decide and act: customer data, insights, decisioning and orchestration. CSPs that hand this control to their BSS vendor's built-in AI will have modernized into the next lock-in.
In the case of customer relationship management (CRM), one application usually holds the data, runs the analytics and manages the process. With the pace at which AI technology is advancing, the data and insights should move to CSP-owned data stores, Agentic AI architectures and insights-enriched workflows. When a supplier holds the data and intelligence along with the platform, the CSP's pace is dictated by the supplier's roadmap.
CSPs need a BSS that gives them the flexibility to stay ahead of what customers will demand tomorrow. Increasingly, a digital agent may decide whether a prospect becomes a customer, and the BSS has to be able to influence that decision
Intelligence belongs inside the capability
While customer service is a natural starting point for agentic AI, in our experience, some of the largest revenue opportunities sit further back, in the parts of BSS that are harder to change, such as billing and product management. Manual checks rarely cover more than a fraction of invoices. In billing, a digital agent can compare pre-invoice records with previous patterns, flag anomalies, make adjustments, and raise tickets for anything that needs human review, cutting revenue leakage and bill shocks. In product management, a digital agent can define and price new offers or create a brand-new product and publish it to the catalog and other systems through Open APIs - enabling the CSP to launch them in days, or hours, maybe even minutes.
What matters is whether the digital agent can take actions or not. A model that only recommends is an insights tool. One that executes within boundaries the business sets, and learns from the outcome, is a business capability.
Take a future-state example: creating a price point based on scan of competitive offers (public forums, customer interaction transcripts, other channels), for customers within a specific region. The agent scans for competitors' offers in that region, proposes a new price point for the customers affected and models the revenue impact. Once a product manager approves it, the agent publishes the price to the catalog. The customer, billing and payment core stays as it is, and if the price does not perform, it can be withdrawn just as quickly.
Customers already ask AI agents to compare plans and prices before they contact a CSP. The next step will be for them to act on their behalf. After a decade of work on omnichannel service, CSPs will have a new participant in the conversation, one that can talk, text, digitally inspect and navigate on the customer's behalf, compare competing products and offers in seconds, work around the clock and expect answers at machine speed. When offers are compared that quickly, the speed at which a CSP can reprice and relaunch becomes a competitive factor, and only capabilities that can act, and can be consumed by AI agents through open interfaces, will keep up.
Funding, staffing and governance decide how fast BSS can change
Architecture alone will not deliver this. Fixed multi-year programs, project teams that disband at go-live and success measured against a predefined target state all made sense when the destination was relatively static. Today the destination is dynamic, leaving little room to adjust, with transformation boards reviewing progress against a plan written years earlier, usually declaring the results as ‘Not Met’.
A better approach is to break modernization into smaller initiatives, each tied to a business outcome such as launching offers faster, reducing billing errors or automating partner settlement, with clarity on how the outcomes will impact CSPs P&L. Agentic AI allows us to undertake this now. Several initiatives can run at the same time. Each has its own business owner and clear measures and keeps its funding for as long as it delivers.
This also changes the finance conversation, and CFOs will notice. The initiatives move from being capital to operating expense during their lifecycle, putting visible pressure on EBITDA. One of the largest hidden costs in BSS modernization is running old and new systems in parallel, and in a multi-year program that cost can carry on for years. With smaller investments, the plan for retiring the old system / capability is agreed before the new one is built, so each release takes part of the old estate out of service. Each investment should also have its own unit economics. Cost per change shows whether the CSP is getting faster and cheaper to change, and inference cost per desired business outcome shows whether AI spend is paying off in the revenue it protects or creates, rather than being buried in a platform budget.
Questions worth asking
Choosing a BSS platform remains an important decision. Over time, though, CSPs will gain more from having flexibility in their platform. Three questions test any BSS program today:
Technology will keep accelerating and no CSP can say for certain where its next revenue will come from. It may be AI applications or something nobody has thought of yet. The CSPs that grow will be the ones who are able to launch these in days or weeks without needing another transformation.
