Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.firstresonance.io/llms.txt

Use this file to discover all available pages before exploring further.

Overview

ION’s dashboards summarize data from across the platform: builds, inventory, runs, quality events, Autoplan output, and more. When something looks wrong on a dashboard, the cause is almost always one of three things:
  • The underlying data hasn’t refreshed yet (you’re looking at a cached or batched view).
  • The data exists but the dashboard’s filter is hiding it.
  • The pipeline that produces this data is genuinely lagging or broken.
This page maps the common symptoms to the actual cause and a concrete fix.

Data refresh cadences

Dashboards in ION are not always real-time. Each dashboard pulls from one or more underlying data sources, and each source has its own refresh cadence:
Dashboard / data sourceTypical cadenceNotes
Build progress / run stateReal-timeComes from the live operational tables; reflects the current state of runs and steps
Inventory levelsReal-timeInventory updates are committed transactionally; dashboards reflect them within seconds
Production analytics (cycle time, throughput)~5–15 minutesAggregated from a rollup that updates on a short cadence
Quality analytics (issue rate, disposition mix)~15–30 minutesIssue-driven aggregates run less frequently than production analytics
Autoplan analytics (anticipated demand, schedule projections)~hourlyAutoplan runs on a longer schedule because its inputs (BOMs, schedules, lead times) change less often
If a dashboard looks “empty” or “stale” within these cadences, that’s expected — wait for the next refresh window before assuming a bug.

Why dashboards show empty data

When a dashboard shows nothing or far less than you expect, work through these in order:

1. Filter is hiding the data

The most common cause. Filters persist between sessions and across users — a filter set yesterday by someone else can hide today’s data. Fix: Click Reset filters (or clear them manually). If the data appears, narrow the filter back down to what you actually want.

2. Date range is wrong

Many dashboards default to “last 7 days” or “last 30 days”. If you’re investigating an event from 3 months ago, the default range hides it. Fix: Expand the date range to cover the period in question.

3. Permissions

Some dashboards show only data the current user has access to. If you’re investigating from a sandbox or test account, you may genuinely have no rows. Fix: Check what role you’re using and what scopes that role has access to.

4. The data hasn’t been generated yet

If the dashboard reads from a periodically-refreshed source (production analytics, Autoplan), and the underlying event happened within the cadence window, the data won’t appear yet. Fix: Wait for the next refresh. If the cadence is hourly and the event happened 5 minutes ago, give it an hour.

5. The source pipeline is lagging or broken

Rare, but happens. If multiple users see empty data on the same dashboard at the same time, and the cadence has clearly passed, it’s likely a pipeline issue. Fix: Open a support ticket — see Diagnostics below.

Autoplan analytics latency

Autoplan-driven dashboards (anticipated demand, schedule projections, supply gap analysis) are the most common source of “is this data wrong?” tickets. Two things to know:
  • Autoplan runs on an hourly cadence by default. A change to a BOM or schedule won’t show in Autoplan analytics for up to an hour.
  • Autoplan is deterministic for its inputs. If the BOM, lead times, and existing inventory are correct, Autoplan’s output is correct. “Wrong” Autoplan output usually means an input is wrong — most often a stale lead time or a mis-tagged inventory unit.
If Autoplan output looks wrong:
  1. Check the inputs first — BOMs, lead times, on-hand inventory at the relevant locations.
  2. Wait one Autoplan cycle after fixing inputs.
  3. If still wrong, surface to support with a snapshot of the inputs and the expected vs actual output.

Snowflake read-only connection

For deeper analytics, ION makes a read-only Snowflake schema available. The schema mirrors selected operational tables (parts, part inventory, runs, run steps, issues, build requirements, anticipated demand) on a near-real-time basis.

Requesting access

Email support@firstresonance.io with:
  • Which Snowflake account you’ll connect from (or whether you need credentials provisioned in our shared account).
  • Which users/service accounts need read access.
  • The use case (BI dashboard, data science notebook, ETL into a downstream warehouse).
You’ll receive connection details + a list of available views.

Available views

The most-used views (subject to change as the schema evolves):
ViewWhat it contains
partsPart catalog — one row per part revision
part_inventoryInventory units — one row per serial / lot / count-at-location
runsRun records with status, timestamps, target part inventory
run_stepsStep-level records — fields, measurements, sign-offs
issuesQuality issues with disposition + resolution
build_requirementsaBOM line items
anticipated_demandAutoplan output — projected demand by part, time bucket, source
Document any custom views you build on top of the read-only schema in your team’s wiki — when the schema evolves, you’ll want a list of what to revalidate.

Connection latency

The Snowflake mirror lags the operational tables by a few minutes. If you need real-time data, query the API directly; Snowflake is for analytics and BI workloads where seconds-fresh isn’t required.

Diagnostics

When opening a support ticket about a dashboard or analytics issue, include:
  • Which dashboard — full URL or screenshot.
  • What you expected vs what you see — be specific (e.g. “I expected ~50 issues from last week; I see 3”).
  • Filters and date range in effect.
  • Time of observation — and whether the issue is reproducible.
  • For Autoplan: the input change you made, when you made it, and when you ran the dashboard relative to the next Autoplan cycle.
  • For Snowflake: the exact query, the result, and the expected result.
A complete diagnostic ticket gets routed faster than an “analytics is broken” report.

Tips

  • Default to assuming the filter is wrong. Reset filters before reaching for any other explanation. It’s the cause maybe 80% of the time.
  • Treat cadence as a contract. If you tell a downstream consumer that anticipated demand updates hourly, build that into their expectations — don’t apologize for “stale” data within the cadence.
  • Keep an eye on Autoplan inputs, not just outputs. When Autoplan output looks wrong, the input is the lever.