Not long ago, we broadcasted a session called Flexibility by Design: How Financial Services CRMs Thrive in the Next Decade. We covered a lot of ground in half an hour, so here's the short version for anyone who missed it, or wants the highlights before catching the full recording.
Eighteen-odd years in this industry, and the one thing I keep coming back to is that there's no finish line. New regulation, new products, new expectations: something changes every year, and you'd be mad to expect otherwise. Consumer duty was the big one before AI, and before that it was something else. There's always a backlog of things a business wants to improve.
That means the software you run on is never really complete. It's complete until the next change. If you haven't built on the right foundation, making that next change costs more, takes longer, and carries more risk than it should.
Financial services has actually changed a lot in how it thinks about this. Ten years ago, the instinct when something new came along was to bolt it on: a self-service option here, a dialler integration there. That's worked, and it's still workable for plenty of businesses. But it tends to open a gap between your data and your operations. A data warehouse is brilliant for reporting and analytics. It's not an operational database, and AI in particular struggles to work off the back of one, because it needs to operate in real time, not in a batch report.
The clearest warning sign that a platform is holding you back isn't financial, it's a feeling. Everything starts taking too much time. A workflow tweak that should be a configuration change instead needs operations, testing and change management all pulled in. When a simple update needs that many people involved, the problem usually isn't the people. It's that the underlying platform was never built to let them just make the change themselves.
That's really the whole distinction between customised and configurable. Custom platforms need developers to make a change. Configurable platforms let the people closest to the business, the operational users, make it themselves. On the call, I used a disputes process as the example: new database fields, a form, some automation, all built through a low-code interface. That should be an hours job, not a days job. On a custom system, the same change means release management, testing cycles and a lot more waiting around.
The same thinking applies to the customer side. If you update a process in the agent's view, you want that reflected in the customer portal at the same time, not a version behind it. Keeping those two in step is exactly what a unified platform is for.
How long a move to a platform like ours actually takes varies enormously. Small businesses with simple processes can be live in a day or two. A mature bank with established processes is a different conversation, and I'd say sensibly it's usually three to six months, sometimes up to twelve for the largest, most complex moves. There's no real disruption either way: we do the groundwork, get the new platform ready, and go live either in phases or over a single weekend, depending on what suits the business.
On AI specifically: I think we're looking at meaningful savings, somewhere in the region of fifty to sixty percent on efficiency, over the next decade. It won't replace people entirely, and for regulated businesses I don't think it should. We build approval steps in deliberately, so a subject matter expert reviews and signs off before anything goes out. What AI can do well is take the busywork off an agent's plate and summarise the recommended next step, so the person picks up from a much better starting point.
The bit that's easy to underestimate is what AI actually needs in order to work properly: unified, real-time data. An AI agent handling a complaint needs a different context to one handling a query about payments, and building that context on top of a fragmented set of systems is difficult, if not impossible, to do cost-effectively. That's the real argument for a single platform. Not because it's tidier, but because it's the only way AI ends up paying for itself rather than becoming another expensive thing to maintain.
If there's one thing I'd want a leadership team to take from this: build your CRM as something that needs to serve the business for the next decade, not just get you through the next product launch. Get the foundation right, and you can keep making changes as the business changes around you, rather than having this same conversation again in five years.
Prefer to watch than read? Catch the full session on demand.
