Frequently Asked Questions

The questions firms actually ask — before they commit, while they adopt, and when a plan decision comes up. Short answers first; the linked page carries the full rule.

Before you commit

Is Quartyl a comparable companies database?

No. Quartyl is an end-to-end benchmarking platform: you upload the tested party’s financials, the pipeline screens the comparable population, the statistical engine computes the arm’s length range, and the workflow governs the study through sign-off. The benchmarking runs against the platform’s maintained comparable population — Quartyl is not a data vendor and not a general-purpose database you can query. What the platform is, in full, is on the What is Quartyl page.

Who can see our data?

Only your firm. Every record in the platform — studies, files, comparables, evidence — carries your firm’s tenant, and every query is scoped to it. There is no cross-firm read path: members of another firm cannot see your studies, your uploads, or your evidence, and evidence search is firm-scoped. That isolation is implemented in code, and it is what this page can answer for you.

Where the data physically sits is a different question with a different answer type: hosting region, encryption at rest, backup cadence, retention beyond the dump window, and any third-party certification are deployment and contract terms, agreed per engagement — the product manual does not assert them. Ask for them in writing before you sign; the engineering controls the code does implement are listed on the Security and Data page.

What leaves the platform during a run?

One thing by default, and it is narrow: the AI screening steps send inference requests to the model gateway configured for the deployment, and those requests carry company names and trade descriptions from your dump. That is the egress surface of a default run; the optional Web Research step, where it is enabled, adds outbound search queries and page fetches.

What does not happen:

• No website contact by default. A run reads no company website unless the optional Web Research step runs — it needs the advanced_ai_screening plan feature, the same one that gates the deep pass. Where it does run, it contributes labelled findings with the URL behind each one, never verified facts.
• No external data pull you did not configure. Screening reasons trace to the text in the workbook you uploaded — plus, where Web Research ran, the labelled web findings the deep pass was given. No legal-entity registry record is joined onto a comparable at any point, and a field left empty means nothing was found.
• No silent third-party send. Download links, storage access, and exports all ride your own configured storage provider.

Whether the gateway you point the platform at retains or reuses what it receives is a term of that provider, not a property of this codebase. Put it in the processing agreement.

What happens to our studies if a seat is removed?

Removal deactivates the account — the user can no longer sign in, and their outstanding sessions are revoked at once — but the studies stay in the firm’s workspace. The work belongs to the tenant, not to the individual account, so nothing is orphaned or deleted. A re-invited user joins the same tenant and the firm’s record is intact.

Seats themselves are a plan entitlement, not a hard product rule: the seat allowance and the per-seat price are fields on the plan, and the seeded defaults are configured figures rather than published tariffs. What you pay, and for how many seats, is agreed per tenancy — see your agreement, not a price list.

What roles exist, and who can do what?

Four firm roles — Analyst (creates and runs studies), Manager (reviews and sends back or approves), Partner (signs off and archives), Firm Admin (everything, plus user management and firm settings) — plus the platform-level Superadmin. The complete transition map is on the roles and permissions page.

Working in the product

Are AI screening decisions final?

No. The AI step is one pass of the screening pipeline; it evaluates qualitative fit and records its reasoning per company. It records four dispositions — accept, almost accept, reject, or flag — and a Manager, Firm admin or Superadmin can override any of them with a documented reason (an analyst cannot), and every override is logged in the audit trail with the person, the reason, and the AI reasoning it overrode. Automated scoring is a starting point, not a verdict. The full rule is on the AI screening page.

How does the study workflow work?

A state machine, not a four-button form. The forward path is Draft → Study Initialised → Processing → In Review → Report Generated → Document Generated → Signed Off → Archived, with two terminal side-states, Rejected and Failed (backend/app/constants/workflow.py).

In practice:

• The analyst works in Draft and initialises the study; the run takes it through Processing.
• When the numbers settle the study sits In Review. The reviewer either accepts it forward or rejects it with a reason.
• A Rejected study goes back to an analyst, who re-submits it for review — no re-run needed if the data did not change.
• A study whose run died is Failed; re-submitting takes it back through Processing.
• Reports and the final document move it through Report Generated and Document Generated; partner sign-off then archives it as a permanent read-only record.

Every transition is recorded in the audit trail with actor and timestamp. The full lifecycle, including which role owns each transition, is on the study lifecycle page.

How are reports delivered, and how long do the links last?

The Excel benchmarking workbook is the primary deliverable, and it is produced inside the run. The seven-chapter Word master report is not automatic: a Partner, Firm Admin or Superadmin requests it from the study, fills in the engagement details (prepared-by, report date, database, search regions, narrative overrides), and an async job builds it from the persisted study record. The branded client-facing PDF is generated from the same context in the same job, and it is best-effort by design — the Word document is already stored at that point, so a PDF failure leaves you with the Word report and no PDF. Treat a missing PDF as “not available”, not as an error to chase.

Evidence packets bundle the supporting documentation for audit readiness. Download links are presigned and short-lived — five minutes by default — so request a fresh one when a link expires, and know that every download is logged. The report format is on the Excel report page.

Can I edit an archived study?

No. Archived studies are immutable to preserve audit integrity. If the study needs to change, it is a new study — or a documented exception through your firm’s governance process, not an edit of the archived record.

Why do my on-screen results differ from the exported report?

The report reflects the latest approved benchmark. If you overrode comparables after exporting, the export is now behind the screen. Re-run the export after the study’s processing has settled and the numbers agree.

Plans and support

Can we white-label the platform for our firm?

Yes, where your plan holds the white_label plan feature: the firm’s brand name and accent colour run through the application shell — the sidebar, the sign-in overlay, and the claim-account page. Configuration is in Firm Settings under the Branding tab. See branding and white-label.

Do we need a plan feature for a given capability?

Some capabilities are plan-gated, and the catalog is closed: eleven keys — result charts, user management, appearance theme, firm settings, extended dump retention, web research + advanced AI screening, predictive risk, the evidence repository, API access, white-label branding, and priority support (PLAN_FEATURES in backend/app/constants/billing.py). A plan-gated capability is hidden, not locked: tenants that do not hold the feature never see an entry point for it, and the backend denies the endpoint regardless. The tier suggestions in the catalog are guidance only — the platform team controls what a plan actually includes, so what your plan holds is the authority, not the list above. It is visible from the Firm Admin surface, and the full breakdown is on the plan features page.

What does priority support add?

In product terms, priority_support changes two things in the application, and both are visible:

• The Support tab renders the priority card instead of the standard card.
• Feedback submissions carry a priority flag in the payload posted to the feedback webhook, so the intake side can route them.

Everything else on that card — the contact addresses, the phone/WhatsApp lines, and the four badges (sub-2-hour response, 24×7 coverage, named account manager, senior-engineer escalation) — is frontend display configuration, not a mechanism the product enforces. The plan catalog describes the feature as “priority support with a named account manager”, but the codebase does not implement a ticket queue, a routing rule, or a measured response time. So the service level you are owed is the one in your engagement agreement; treat what the card says as a rendering default the platform team can change without a backend deploy. One quirk worth knowing: accounts with no tenant and no plan signal fail open and see the card too, which is why the card alone is not proof of an entitlement. Firms without the feature see the standard card and no lock panel — gated UI is hidden, not locked. The details are on the contact support page.