Workflow Configuration: Review Routing and Approvals

The study workflow is a state machine that the platform owns end to end: the states, their order and the role permitted to move a study between them are fixed and identical in every tenant. What a firm configures is the people standing at each gate. This page separates the two, because confusing them is how firms get surprise bottlenecks.

What is fixed: the states

The lifecycle a study moves through is not configurable. A study is created as a Draft, reaches Study Initialized once its workbook is attached and the run is submitted (a dispatch that fails sends it back here), is run as Processing, lands in In Review when the pipeline completes, is approved into Report Generated, carries the on-demand Document Generated step, is Signed Off, and is Archived — with two terminal side-states alongside the spine: Rejected and Failed. A rejection does not send the study back to Draft: it enters Rejected, the analyst addresses the reasons, and re-submitting returns it to In Review. A failed run re-submits into Processing. The complete state machine, with the role that drives each transition, is documented in study lifecycle.

Two properties of the spine are worth internalising:

• The review gate is a different person’s role. The analyst who built the analysis cannot approve it — the In Review → Report Generated transition belongs to the Manager tier (and up).
• Archiving is the sign-off authority’s act. A study becomes immutable only in the hands of the Partner, Firm Admin or Superadmin tier.

What is configurable: nothing in the machine, everything in the people

There is no workflow configuration surface in Quartyl. No tab of Firm Settings edits the review path — the panel offers Security, Single Sign-On and Branding and nothing else (see Firm Settings) — and the state machine ships as a fixed constant:

• The ten states and their ordering.
• ALLOWED_TRANSITION_ROLES: the exact role set permitted on each transition, identical for every firm.
• The two system-only states (PROCESSING, REPORT_GENERATED), which no human can advance out of — the worker does. The one human action on a Processing study is to cancel it back to Draft.

What a firm shapes is who stands at each gate, through the team console and a study’s own assignment slots:

• Manager review is held by the Manager tier and routes through the reports-to chain: when an analyst submits, the study is assigned to that analyst’s supervisor as its reviewing manager, and the manager’s own supervisor is recorded as the sign-off partner. Set the chain before the first submission; a study with no resolvable reviewer is a study that waits.
• Partner sign-off is held by the Partner tier — and Firm Admin and Superadmin sit in the same role set on that transition, so a firm’s operator is never blocked from signing off.
• Roles bound by the firm’s role scope. Who can hold which tier is provisioned by the platform when the firm is created, so a firm that cannot invite a Manager has no manager-review tier to route to.
• The per-study Assignments card. The auto-assigned reviewer and sign-off are a default, not a sentence. On a study’s detail page the reviewer slot can be re-pointed by a Manager, Firm Admin or Superadmin and the sign-off slot by a Partner, Firm Admin or Superadmin; a senior role may fill a junior slot but a junior role may never fill a senior one, the target must be an active member of the same tenant, and an archived study refuses the change. Each reassignment is written to the audit trail as an assignment change. The card hides entirely on studies created by a Partner, Firm Admin or Superadmin — there is no one above the creator to route to.

So the model is: the platform owns the gates, the firm owns the people at them. There is no way to add a tier, remove one or redefine what a gate does; a slot can be re-pointed at a different person, not at a different kind of gate.

Approval queue behavior

There is one study list with one set of status tabs; what varies by role is which rows a member sees and what they may do to them:

• The task list is one list. Every role sees the same eight status tabs (Draft, Processing, In Review, Report Generated, Document Generated, Signed Off, Rejected, Failed) plus All — the tab bar is not re-cut per role. What rows appear is scoped: a Partner or Firm Admin sees every study in the tenant, while an Analyst or Manager sees their own work, their direct reports’ work, and any study assigned to them as reviewer or sign-off. Within that scope, role decides which action a study offers: In Review shows a Manager the approve/reject controls, Document Generated shows the Partner tier the sign-off control. The queue mechanics are in My Tasks and Approvals.
• Approve or send back. Approval advances the study (with a note kept on record); a send-back puts it in Rejected with the reasons attached, and the analyst addresses them and re-submits straight back into In Review — no re-run. Cancelling a study that is still Processing is the one action that returns it to Draft. Every transition — who, what, when — is written to the study’s audit trail.
• Bottleneck visibility. The Audit Trails view lists how long studies have sat in each state, with active and at-risk counts, so a queue that is quietly growing is visible before it becomes a missed deadline; the audit trail records every wait.

FAQ

Can we skip manager review for some studies? No. There is no workflow tab and no per-firm setting for the path: the In Review → Report Generated transition requires the Manager, Partner, Firm Admin or Superadmin role, and it is the same rule in every tenant. What you can decide is who stands in the reviewer slot. It is filled from the creator’s reports-to link, it can be re-pointed on the study’s Assignments card, and because Partners and Firm Admins share the transition’s role set, a firm can put one of those accounts in the reviewer slot and review in a single step.

A study is waiting and no one has seen it — why? Two usual causes: the study’s submitting analyst has no reports-to set (no resolvable reviewer), or the reviewer’s queue is not being worked. The bottleneck view shows which state is holding the study; the chain is the first thing to check.

Can the states themselves be customized? No — the states, their order and the role rules of each transition are fixed by the platform and shipped as one constant for every firm. Firms configure the people at the gates, not the machine.