Shop-floor board
Eight KPIs, each with a target.
Measured by the lab itself every 15 minutes and pushed out as aggregates. If the lab is asleep or offline, the last snapshot stays up with its timestamp.
Loading measurements…
How each number is measured
| KPI | Source | Granularity |
|---|---|---|
| Uptime | Uptime monitor, average of the core services (DNS, smart home, photos, documents, files, proxy, sign-in, memory) | 30-day %, daily trend |
| Backup streak | Push monitors of the nightly backup and replication jobs; a day counts when no job reported a failure | days |
| AI renders | Finished jobs in the history of each local image-generation node that answers; asleep nodes keep their last count | per day, 7 days |
| PDCA cycles | The lab's failure log: every gotcha gets a number, a cause and a countermeasure | total, per week |
| Ideas | A ledger of every distinct idea raised since July 2026 (a new capability or project, not a task), each with the note or commit that proves when it was raised and when it first worked; implementation rate = implemented ÷ raised. First count, under review | per month, by category |
| 3D-printed fixes | A ledger in the site's repository: print cost against the price of the alternative | per entry |
| Make vs buy (3D) | Per printed part: hours from the order (the part is asked for) to the end of the final good print, split into design/render (order → first print start) and print + rework (first start → final good end), against the days to buy it (about 21 from China, assumption) plus the 33.5 % import duty on parcels from non-treaty countries. Cost to make = filament of every attempt (failed ones included) + power; MXN saved shows "pending" until the price of the bought alternative is known, and pending parts stay out of the total | per part, median hours |
| RFT — Right First Time | The printer's own job history: parts whose first attempt finished good ÷ parts attempted. Prints aborted before the first layer are no-starts and do not count. A job that "completed" with a failed part is corrected by hand from the lab log. The whole-lab line = implemented ideas without a documented redo ÷ implemented ideas (first count, under review) | totals since the first print |
| COPQ — Cost of Poor Quality | Internal-failure cost of failed or cancelled prints: scrapped filament × MXN/kg + their energy (print hours × average power × MXN/kWh). Machine hours lost are shown as hours. "Assumptions pending" while a price is still a proposal; filament without a price is shown in grams | MXN per month, per part |
Privacy: only the whitelisted aggregate fields above are published. No hostnames, addresses, service URLs, names or hour-of-day activity. Idea titles appear only for implemented ideas marked public, printed parts only when marked public; personal ones are counted, never named.
Want a board like this on your line?
The same KPIs, targets and PDCA rhythm work on a production floor. Zenilify Consulting sets them up with you.
Book a diagnostic call