Build vs. buy: collections software for growing lenders.

The short version: most lenders underestimate collections builds by an order of magnitude, because the visible part, sending reminders, is maybe a tenth of the system. The invisible parts are state management, decisioning, channel infrastructure, compliance enforcement and the audit trail.

What "build" actually means

A credible internal build needs, at minimum:

• Account state management. One live state per account, balance, DPD, promises, disputes, eligibility, reconciled continuously against your core systems. This is a data-engineering product in its own right.
• Decisioning. Something has to choose channel, timing, message and offer per account, and recalculate when the account changes. Rules get you started; learning from outcomes is where the uplift lives, and where builds usually stall.
• Channel infrastructure. Messaging routes with deliverability management, voice infrastructure (and in 2026, credible AI voice in Arabic and English), email reputation, a borrower self-service surface. Each has its own vendor stack, compliance regime and failure modes.
• Compliance enforcement. Contact windows, frequency caps, disclosures, dispute freezes, encoded so violations are impossible, and evidenced per interaction.
• The audit trail. Every decision and interaction, reconstructable per account, forever. Bolting this on later is close to impossible.

A serious version of that stack is a multi-year effort by a dedicated team, while your delinquency book compounds monthly.

When building makes sense

Honest cases exist: collections is your core product (you're an agency or servicer); your scale amortizes the build (millions of active delinquent accounts); or your product is so unusual that no platform models it. If you're a lender whose product is lending, none of these usually hold.

The middle path most lenders miss

Build-vs-buy is usually framed as software-vs-software. But the real decision has a third axis: who operates it. Buying software still means staffing an operation. A managed service means outcomes without the operation. The strongest position for a growing lender is often a platform that offers both, start managed for immediate lift, take segments in-house as your team matures, on the same account states. That optionality is worth more than any single feature. (This is exactly the Command / Allocate / Hybrid architecture.)

Questions that settle it

• What is the fully-loaded cost of the build team for three years, versus the recovery uplift a proven platform delivers in quarter one?
• Who carries the compliance risk of a home-built dialler making a mistake at scale?
• Can your build produce, today, a complete evidence trail for any account a regulator picks?
• If your head of collections leaves, does the strategy leave too?

Collections deserves the same discipline as any infrastructure decision: buy the commodity, build only what differentiates you. For a lender, differentiation lives in origination and product, not in owning a dialler.