Assemble Performance
Isabella Viracachá
issaviracacha@gmail.com
Total Hours
Delivery (Adj.)
Projects Worked
Clients
Projects Worked (3)
Hours by Task
Monthly Trend
Forge production
Australian clients (polaris + mmg) · 2026-06-01 → 2026-06-30 (Australia/Sydney)
Velocity
Five clocks, kept separate on purpose. Production is the operator clock the studio is accountable for; queue time is how long a brief waited before being claimed.
| Clock | Median | P75 | Mean | Max | Sample |
|---|---|---|---|---|---|
| Production in_production → internal QA | 1.6h | 2.12h | 1.78h | 3.4h | 4 |
| Internal QA internal QA → ready for review | 0.44h | 0.49h | 0.33h | 0.53h | 3 |
| First delivery in_production → first proof | 2.23h | 3.04h | 2.2h | 3.85h | 3 |
| Revision turnaround revision requested → next proof | 1.79h | 2.14h | 1.79h | 2.48h | 2 |
| Queue (not operator) created → claimed | 25.35h | 25.35h | 25.35h | 25.35h | 1 |
Production time by product
Production time by complexity
By build type
Deliverable mix
Categorised by what was claimed in the period. Product is the ad family (Tabloid, Tabloid Special, A4/A5 Mag, Flyer…); product type is the concrete size.
Product
| Tabloid | 1 |
Product type / size
| T22 (92 x 129 mm) | 1 |
Build type
| new build | 1 |
Artistic discretion
| ad3 | 1 |
Turnaround
| overnight | 1 |
Client
| McPherson Media Group | 1 |
Sources: 1 ingested (MMG feed) · 0 created in the portal · 1 dealer sites · regions AUSTRALIA 1
Internal QA
Revisions & confirmed omissions (assemble-fix)
A round an Assemble ADMIN ruled our fault ("assemble-fix"): either classified internally (internal QA path) or flagged by the client and confirmed afterwards. Pending flags, rejected flags and no_agreement outcomes are excluded.
No confirmed omissions in this period.
SLA
Two phases, never blended: first delivery runs against the TAT deadline; each revision opens its own fresh overnight clock.
First delivery
2 on time of 3 measured · median margin 0.27h
Revisions
2 on time of 2 measured · median 1.79h
Late first deliveries
| Deliverable | TAT | Deadline | Delivered | Late by | Queue | Production |
|---|---|---|---|---|---|---|
| Tabloid — 198304McPherson Media Group | 2h | 2026-06-30 03:30 | 2026-06-30 03:46 | 0.27h | 22.76h | 1.7h |
Queue vs production tells you where the miss came from: a large queue with fast production means the brief was claimed too late, not produced too slowly.
Efficiency
Throughput
| Deliverables per active day | 0 |
| Output units per active day | 0 |
| Weighted units per active day | 0 |
| Cohort median (weighted) | 2.04 |
| Throughput index | — |
| Active days | 1 |
Score components
Components with too small a sample are omitted and the remaining weights renormalised. speed_index and throughput_index are cohort-relative (1.00 = cohort par); 50 + 50 × (index − 1) maps them onto 0-100.
Cohort comparison (4 operators with activity)
| Operator | Completed | Weighted / day | Production median | Speed index | First-pass QA | Omissions / 100 | Score |
|---|---|---|---|---|---|---|---|
| Cata Peña | 28 | 3.35 | 1.13h | 0.93 | 94.9% | 0 | 77.6 |
| Isabella Viracacháthis profile | 0 | 0 | 1.6h | 0.63 | 100% | 0 | 70.7 |
| Mari Mora | 12 | 2 | 0.66h | 1.08 | 86.7% | 0 | 68.3 |
| Amanda Najera | 22 | 2.04 | 1.06h | 1 | 84.8% | 0 | 67 |
The cohort is every Forge user who owned work in the window, tagged Creative Operator or not, so the baseline does not move when Harvest tags change. Speed index is mix-adjusted against the same product + complexity cells, so a lighter queue does not inflate it.
Monthly history
| Month | Claimed | Completed | Weighted units | Active days | Production median | First-pass QA | Omissions | SLA first delivery |
|---|---|---|---|---|---|---|---|---|
| 2026-06 | 1 | 0 | 0 | 1 | 1.6h | 100% | 0 | 66.7% |
| 2026-07 | 72 | 65 | 78.6 | 19 | 1.03h | 92.8% | 2 | 97.1% |
How every number here is defined
- task
- A client-created container (Forge `projects` row). One task holds N deliverables.
- deliverable
- One individual asset an operator claims, produces and bills (`deliverables` row).
- output units
- Deliverables expanded by bundle size — a 4-size banner bundle is 1 deliverable, 4 outputs.
- weighted units
- Output units × complexity weight (AD1 1.0 / AD2 1.5 / AD3 2.0, × 0.7 for adaptations).
- production h
- First in_production → first internal QA submission. The operator production clock.
- qc h
- First internal QA → first ready_for_review.
- first delivery h
- First in_production → first ready_for_review (production + QA loop).
- queue h
- Created → claimed. Queue time, not operator time — context only.
- qa return
- A QC bounce: internal_qc → in_production, with the reviewer reason attached.
- first pass qa
- Share of deliverables that entered internal QA and never bounced back.
- confirmed omission
- A round an Assemble ADMIN ruled our fault ("assemble-fix"): either classified internally (internal QA path) or flagged by the client and confirmed afterwards. Pending flags, rejected flags and no_agreement outcomes are excluded.
- sla first delivery
- First ready_for_review vs the TAT deadline from the portal deadline function (2h/4h/6h = created + N h; overnight = next 8am Sydney; 2nd/3rd day = +1/+2 on that ladder).
- sla revision
- Each revision_requested → next ready_for_review vs a fresh overnight deadline. Reported separately from first delivery — the two are never blended.
- speed index
- Median of (cohort median hours for the same product+AD ÷ own hours) per deliverable. 1.00 = cohort par, >1 = faster on comparable work.
- throughput index
- Own weighted units per active day ÷ the cohort median rate. 1.00 = par.
- active day
- An AU-local day with at least one upload or status transition by this operator.