The CRM decision looks like a software purchase. For a funded-trading firm it is closer to a risk decision: the system you choose will hold your evaluation economics, your client book, your flow intelligence and your payout process — the four things you can least afford to have wrong, leaked or locked in. This guide is the framework we would use to evaluate anyone, including ourselves: seven questions, the tests behind them, and an honest note on when the cheapest option is the correct one.
1. Tenancy: who else lives in your database?
Most prop-firm platforms are multi-tenant: your firm is a tenant_id in a shared database alongside your competitors. That is not automatically disqualifying — it is how the low price is possible — but you should know exactly what it implies: isolation enforced by application code rather than architecture, export on the vendor’s terms, and the structural possibility that your patterns feed the platform’s own models and benchmarks. Ask two questions in writing: can we take the full database and leave, at any time, in a documented format? and is our data ever aggregated or mined across tenants? Vague answers are answers.
2. Where does risk surveillance live?
The single sharpest differentiator. In most stacks, risk is a report: a dashboard refreshed on a schedule, reviewed after the P&L moved. In an operating layer, risk is in the workflow: every payout request arrives with behavioural flags attached, every evaluation verdict carries its evidence, and toxic-flow detection runs against live positions rather than last week’s export. The test is concrete: ask the vendor to show you a payout request as the approver sees it. If the risk picture is on another screen, in another tool, or in a PDF, the surveillance is decoration.
3. Is the rulebook code or folklore?
Every evaluation dispute in this industry has the same root: nobody can prove which rules governed an account when. The mature pattern is rules as configuration — drawdown definitions, news windows, consistency requirements encoded once, enforced deterministically, and versioned per account, so “which rule set did this trader sign under?” is a lookup, not an investigation. Ask when rules changed last, and how the platform answers that question for an account funded before the change.
4. Platform and payment coverage — real, not logo-wall
Coverage claims are cheap. The operational questions: does account, trade and balance data sync in real time or on schedule? Does provisioning happen automatically at checkout, or does someone paste credentials at 2 a.m.? Are payment providers orchestrated — routing by region, method and risk — or is each PSP another tab? For funded firms the end-to-end test is checkout to funded account with no manual glue; for the platforms that matter to you (MetaTrader 4/5, cTrader, Match-Trader, DXtrade, TradeLocker and whatever you run next), ask which integrations are first-party and which are “roadmap.”
5. Follow the money in the commercial model
Pricing structure is incentive structure. Revenue share puts your vendor inside your P&L — and gives them opinions about your risk decisions. Per-trader and per-account fees tax your growth precisely when you succeed. Setup and migration fees tell you how the vendor thinks about switching costs — yours, and the next firm’s. None of these is illegal; all of them are information. Prefer fixed, published anchors with itemised scope, and treat any vendor who will not put a number in public as quoting you by your appearance.
6. AI features: leverage or liability?
In 2026 every platform demos an AI assistant. The difference between leverage and liability is governance, and it is checkable. Which external models, under what API terms — is your data excluded from training, and what is retained? What exactly crosses the boundary per request — scoped tool outputs, or the dataset? Is every AI action logged and role-scoped? And is there an off switch for deployments that demand it? A vendor fluent in those answers designed the feature for institutions. A vendor who answers with model names designed it for the demo.
7. Exit: read the divorce clause first
The best predictor of a partnership’s quality is how it ends. Before signing, ask for the exit runbook: what you receive (full schema, documented), when (a standing capability, not a termination ceremony), and what it costs (a number, not “contact us”). Ask equally about entry: is migration from your current system part of the engagement — data mapping, a parallel run, phased cutover — or an invoice surprise? Vendors confident in retention make leaving easy.
The red-flag shortlist
- Cannot state in writing whether your data is mined or aggregated across tenants
- Risk “module” sold separately from the payout workflow it should live in
- No rule versioning — disputes settled by screenshots
- Revenue share presented as “alignment”
- Setup fees plus migration fees plus exit fees on the same rate card
- AI features with no answer on training, retention or an off switch
- Export described as “CSV of major objects”
When turnkey is the right answer
Honesty the industry rarely offers: if you are launching a standard-template firm this quarter, on a tight budget, with no in-house operations muscle yet — a multi-tenant turnkey platform is probably correct. Fast, cheap, adequate. The framework above starts to bind when the operation itself becomes your edge: when your evaluation economics, flow intelligence and client book are exactly the assets you cannot put in a shared system, and when a vendor’s incentives inside your P&L stop being a rounding error. Most durable firms cross that line; the mistake is noticing years after the data did.