Knowledge Hub / Scaling playbook
From 200K to 2 Million Accounts
The CoreFi scaling model sets out the milestones and levers that take an institution from 200K end users toward 2 million. This page is the operating playbook behind that model: what actually breaks at each scale stage, what must be instrumented before the next stage, and who in the organization has to own it. It is written for CEOs, growth leaders and operations leaders, and it is useful whether or not you ever run CoreFi. As on the model page: the 500K, 1M and 2M milestones are a strategic model, not a forecast. They depend on each institution's products, markets and licensing; the failure modes and the instrumentation do not.
The verified baseline
The playbook starts from production figures, not assumptions.
Every number below is a verified production figure, published and maintained on CoreFi in Production together with the counting rules behind it. Where customer deployments produce measured results, they are published as anonymized metric ranges on Client Outcomes; nothing in this playbook is a projection dressed as an outcome.
How to read the three stages
Each stage breaks a different part of the operating model.
The pattern that repeats across scale stages is consistent: the constraint that gets you to one milestone is not the constraint that blocks the next. Conversion problems give way to distribution problems; distribution problems give way to jurisdiction problems. The playbook below describes each stage in three parts: what breaks first, what to instrument before the stage begins, and what organizational readiness the stage demands. The milestones remain a model, not a commitment; the failure modes are what we observe in regulated banking and lending operations generally.
Stage one / 200K → 500K
The conversion stage: growth is throttled by your own funnel and back office.
At this stage most institutions do not have a demand problem; they have a conversion and capacity problem. The demand they already attract leaks out of the onboarding funnel, and the accounts that do land are serviced by an operations team whose cost scales linearly with volume. CoreFi's Customer Onboarding and AI Workflow Control Plane modules carry this stage in the model, but the playbook applies regardless of stack.
What breaks first
Onboarding drop-off: each manual document check and identity review adds latency, and applicants abandon. Exception handling stays manual: AML hits, mismatched documents and edge-case applications pile into queues. Servicing headcount scales linearly: every cohort of new accounts demands another unit of operations cost, which caps growth at the budget line rather than at demand.
What to instrument
Funnel conversion measured step by step, from application started to account funded, so drop-off has an owner and a location. Exception rates per workflow, including how many cases leave the straight-through path and why. Cost per funded account, fully loaded, tracked as a trend rather than a one-off study. No target values belong in the playbook; the discipline is that these metrics exist and are reviewed before the stage begins.
Organizational readiness
Someone must own automation policy: which cases may flow straight through, which require human review, and how that boundary moves as evidence accumulates. In practice this sits with the COO or an operations-policy owner reporting to one; the framing CoreFi uses with operations leaders is on For COOs. Without a named owner, automation decisions are made ad hoc by whoever is closest to the queue.
Stage two / 500K → 1M
The distribution stage: growth shifts from your funnel to other people's channels.
Past the point where converting your own demand carries growth, the next stage runs through distribution: partner brands, embedded-finance channels and white-label journeys. The failure modes change accordingly. In the model this stage is carried by Headless APIs, White-Label Channels and partner ecosystem integrations; what follows is what the stage demands operationally.
What breaks first
Channel and partner onboarding becomes the bottleneck: each new partner takes months of bespoke integration work instead of weeks of configuration. Brand-by-brand configuration drifts: each channel accumulates its own product variants, policies and pricing until no two behave alike. Reconciliation across channels fragments: when partner traffic and direct traffic settle in different records, the books stop agreeing with themselves.
What to instrument
Partner activation time, from signed agreement to first live transaction, measured per partner so the integration path improves with each one. Per-channel unit economics: revenue and operating cost attributed by channel, so a growing aggregate cannot hide a channel that loses money on every account. Both are metrics to watch, defined before the first partner goes live rather than reconstructed afterwards.
Organizational readiness
A partner operations function, distinct from sales: someone has to own partner onboarding, partner servicing and the escalation path when a partner's customer has a problem. API governance with equal standing: versioning, deprecation and change-communication discipline, because every breaking change is now a partner incident. The executive sponsorship question this raises is the CEO's; the framing is on For CEOs.
Stage three / 1M → 2M
The replication stage: growth becomes a regulatory and jurisdictional problem.
The final stage in the model is multi-country replication: launching the proven deployment shape in additional markets. The hard problems here are not technical. CoreFi's deployment patterns are already proven across the six geographies documented on In Production, but in every one of them the customer remains the regulated entity, holding its own license and its own supervisory relationship. That is the defining constraint of this stage.
What breaks first
Per-country compliance variance: consumer-credit rules, disclosure requirements and reporting formats differ enough that a single hard-coded compliance build fails in market two. Data residency: where customer data may be stored and processed becomes a per-jurisdiction architecture question, not a footnote. Supervisory relationships: each market brings its own regulator, with its own expectations of the locally regulated entity.
What to instrument
Time-to-launch per country, measured from market decision to first live customer, so replication actually gets cheaper with repetition instead of restarting from zero each time. Per-jurisdiction audit readiness: whether each market's deployment can produce the records its local supervisor expects, verified on a cycle rather than discovered during an inspection. Again, metrics to watch and review, not targets to publish.
Organizational readiness
A local regulated entity or license per market, because the platform provider cannot stand in for one: customers remain the regulated entities in every market they enter. That means local accountable management, local compliance ownership and a supervisory relationship the institution itself maintains. Lending programs replicated across markets also need per-jurisdiction credit policy; the lending stack behind that in the model is Lending Automation.
The playbook on one page
Stage, failure mode, instrumentation, and the modules that carry it.
The module column matches the stage-to-module mapping on the scaling model. The milestones describe the model, not a commitment; the failure modes and instrumentation stand on their own.
Where a stage produces verified customer results, they are published as anonymized metric ranges on Client Outcomes; we do not publish projections as outcomes, and no figure in this table is a target value.
Walk the playbook against your own numbers.
A growth strategy session takes your current baseline, identifies which stage you are actually in, and works through the instrumentation and readiness questions above before any module discussion. The model behind this playbook is on /growth; the production evidence underneath it is on /in-production.
Continue in the Knowledge Hub
Related reading for the executive team.
Knowledge Hub
All buyer-education guides in one place: scaling, modernization and AI governance for regulated institutions.
Browse the Knowledge Hub →The board guide to agentic AI core banking
What a board should ask before approving AI in the banking operating model, and what evidence to expect back.
Read the board guide →Modernize the core without a big bang
How institutions replace core capabilities one journey at a time, with reversibility at each step.
Read the modernization guide →