The Evaluation Scorecard
Twelve questions to ask every platform. Ours included.
Vendor decks all say the same words. This is the checklist instead: what to ask, how to verify each answer in an Interpose sandbox in minutes, and — where our own boundary is — a plain note saying so. A vendor confident in the answers publishes the questions.
Open a sandbox and start checkingThe book and the math
Does the engine actually compute what the deck claims?
- 1
Does one strategy rebalance across custodians — or only inside each?
Multi-custodian firms live in the gap between those two answers. A per-custodian rebalancer means your model is really three models that drift apart.
Run it yourself: Your seeded book holds 6 aggregated-custody accounts at outside custodians sitting in the same portfolios as the native ones. Open Rebalancing, run the drift view, and watch one strategy span all of them.
Honest boundary: in overlay mode, outside-custodian data moves at custodian feed speed, not exchange speed. Anyone who tells you otherwise is describing custody they don't have.
- 2
Is the tax engine lot-level — holding periods, wash sales, both terms?
A tax pitch that can't show you a lot ledger is a report, not an engine. Wash-sale handling is where the difference shows.
Run it yourself: The seeded book carries 3 years of history through the real fill path: long- and short-term losses, realized gains in two tax years, and one genuine wash sale. A harvest run is waiting at ORDERS_GENERATED when you arrive — pick it apart lot by lot.
- 3
Is performance labeled net or gross — and does the platform say which?
An unlabeled TWR is a compliance finding waiting to happen. The label matters more than the number.
Run it yourself: Open any account's Performance tab and look for the net/gross labeling and the disclosure text under it. Then check the same thing on whatever you use today.
Your firm on it
Can you evaluate with YOUR book, YOUR brand, YOUR artifacts?
- 4
Can the platform say something about the book you actually manage?
Every demo shows you the vendor's flattering data. The evaluation that matters runs on yours.
Run it yourself: Export positions from your current custodian and paste the CSV into Analyze Your Book (in your sandbox nav). Harvestable losses, concentration and drift against a model — on your own book, nothing imported or stored.
Honest boundary: a CSV has no acquisition dates, so this is aggregate-level analysis. It says so on the page. Lot-level answers come from a real import at onboarding.
- 5
Can you see your own brand on it before you commit?
White-label claims are cheap to make and expensive to retrofit. Self-serve is the proof.
Run it yourself: Open White-Label in your sandbox nav, set your name, color and logo, and save. The portal you're looking at rethemes on the spot; your seeded clients' portal follows.
- 6
Can you hold the artifact your clients will actually receive?
Client statements are the one deliverable every client sees every quarter. Evaluate the PDF, not the screenshot of it.
Run it yourself: Reports → Client Statements: pick a seeded account and a quarter, generate, download. It renders with the branding you just set.
The trial itself
Is the evaluation environment honest about what it is?
- 7
Is the trial a real tenant — or a shared demo login?
A shared demo can't tell you how the platform scopes, permissions, or isolates. A real tenant can — and everything you configure should survive a conversion.
Run it yourself: Signup provisions a real firm with its own rep codes and a 19-client seeded book. Converting later is a flag flip: your firm, models, fee schedules and API keys stay exactly where they are.
- 8
Can your team evaluate together, each with their own login?
An RIA evaluates as a committee — ops checks reconciliation while the founding advisor checks the client experience. Password sharing on a trial teaches bad habits.
Run it yourself: Invite a colleague from your dashboard. They get their own advisor seat on your firm — same book, separate login, three seats per trial.
- 9
Can you see the client side without a second sales call?
Half the product is what your clients see. If evaluating it requires a demo appointment, that tells you something.
Run it yourself: 4 seeded people can sign in to the client portal — advised and self-directed shells both, derived from account shape the same way a real firm's are. Their credentials are on your dashboard.
The plumbing
Will it still look good when your engineers open it?
- 10
Is every screen backed by an API you could call yourself?
A platform that is API-first survives your weirdest workflow. One that isn't makes every gap a feature request.
Run it yourself: Mint a machine credential from your own trial login (Profile → API), then replay anything the portal does from curl. The developer docs' quickstart runs against your seeded book.
- 11
Do events stream, or do you poll?
Fills, settlement changes and rebalance completions should push to you. Polling architectures show up later as reconciliation lag.
Run it yourself: Subscribe a webhook or open a WebSocket stream from the developer docs against your sandbox — then trade a seeded account and watch the events arrive.
- 12
Does the vendor tell you what is simulated — unprompted?
Every trial simulates something. The vendors worth trusting document the boundary; the others let you find it.
Run it yourself: Read "What's real vs. simulated" in your sandbox guide — it names exactly where the simulation boundary sits, down to which paths use the built-in fill simulator.
The checklist is yours either way.
Take these twelve to every vendor on your shortlist — we wrote them because we like our answers. The sandbox is free, provisions in about thirty seconds, and nothing in it asks for your calendar. Pricing is scoped per firm; talk to us when the checklist says it's worth a conversation.