File: 2026.10.26 - Lumenwire Audio - 02 Challenge Notes.md
Lumenwire Audio: Challenge Notes from Stage 2
From Kewpie to Gary and Adam. 26-Oct-26. Sample data. Full evidence is in the KPI Calculation Pack. Kewpie has not added, dropped or reprioritised any KPI and has not changed a target. Each note is open until Gary or Adam answers it.
Corrections to Gary's draft measures
| # | KPI | Before (Gary's draft) | After (Kewpie) | Why |
|---|---|---|---|---|
| A1 | Total revenue (K1) | Filters: "Hardware = 4000, 4050, 4200; software = 4100, 4110 (Kewpie to confirm)" | Hardware = 4000 and 4050 (hardware sales less volume rebates). Software = 4100 and 4110. Other = 4200 and 4210 (developer kits and freight recovered). All six accounts stay in total revenue. | Developer kits are not hardware sales, and the board plan and the forecasts treat them as 'other'. The total is unchanged and reproduces A$118.4m (FY25) and A$80.6m (FY26). |
| A2 | New devices activated (K2) | Filters: "First activation only: one row per device_serial"; OEM split from oem_id | First activation only: keep the earliest first_seen_ts per serial (5,040 repeat rows dropped). Fill a blank oem_id from the OEM that owns the same manufacturer_code (22,024 rows recovered; 85,186 stay unassigned). | Repeat serials are devices re-joining a network. The OEM backfill changes only the split by OEM. The 'exclude beta' rule has no effect in the window. |
| A3 | Months of stock cover at the OEMs that report (K2.1) | Divisor left open: "average monthly units activated or invoiced (Kewpie to confirm)"; stock "SUM(units_on_hand, reels x 500)" at reporting OEMs | Divisor = average monthly units invoiced to the same OEMs in the same quarter (erp_invoice_lines.qty, chip, module, card). Stock = units_on_hand, reels x 500, from OEMs that reported in that quarter; names cleaned to match oem_master. | Invoiced units are the same unit as the stock, full scale and carry an OEM on every hardware line. Activations are a 1-in-10 sample with 7% of devices unassigned. Result in 2026-Q3: 10.9 months (Gary's draft said 9.3). |
| A4 | Software revenue growth (K3) | Formula (v1.1 draft): "leaving out any single licence worth more than A$1m (threshold to agree with the CFO)" with the calculation -1 * SUM(gl_monthly.amount_aud) for accounts 4100 and 4110, R12M over prior R12M, minus 1 (no exclusion in the calculation) | Calculation now includes the exclusion: (software R12M less licences over A$1m started in R12M) / (software prior R12M less licences over A$1m started in prior R12M) - 1. Growth including them is shown alongside. | The calculation had to match the plain English. Result -15.6% (1 licence removed). |
| A5 | Design wins that ship (K4) | Calculation: "COUNT(opp where stage_entered_shipping not blank) / COUNT(opp where stage_entered_design_win <= today - 18 months)"; Sources include "erp_invoice_lines.oem_id, sku (cross-check)" | Same ratio, with 'today' fixed at the end of the data (30-Sep-26), shipped defined by the shipping date, and rows with no stage dates excluded and counted. The invoice cross-check is dropped for lack of a key. | A stage field without dates cannot be placed in a cohort. OEM-level invoices cannot confirm a product shipped: nearly every OEM buys every year. |
| A6 | Months from design win to first shipment (K4.1) | Calculation: "MEDIAN(months_between(stage_entered_design_win, stage_entered_shipping))" for opportunities that shipped | Same median, restricted to design wins at least 18 months old. | Younger cohorts only show their fastest shippers. Median 15.0 months for mature cohorts against 14.3 for all. |
| A7 | Forecast accuracy (K5) | Calculation: "ABS(actual_q - forecast_q) / actual_q; forecast_q = forecast_snapshots.forecast_amount from the snapshot nearest the quarter start" | error = (forecast - actual) / actual, reported as a signed figure and an absolute figure. Forecast = latest snapshot on or before the first day of the quarter, converted to AUD, summed over hardware, software and other. Actual = hardware (4000, 4050), software (4100, 4110), other (4200, 4210). | 'Nearest' could pick a snapshot taken after the quarter began. The sign shows bias. The family mapping matches the board plan. |
Challenge notes
| # | KPI | Issue | Evidence | Suggested fix | Status |
|---|---|---|---|---|---|
| C1 | Total revenue (K1) | The FY27 board plan is below the straight-line path in Gary's thresholds. Plan A$92.7m (BT-02, Finance, 15-Jun-26) against A$97.8m on the straight line at Jun-27: -A$5.1m (-5.2%). On Gary's thresholds, plan delivery would already be Red against the path. | board_targets, board_plan_fy27 | CFO to explain how the A$115m and FY27 numbers fit together (Q-1, Q-2). Gary to re-check the straight-line path with the CFO. | Noted in v1.1 (target rationale and threshold note) |
| C2 | Total revenue (K1) | The FY27 plan repeats the one-off licence. Every plan line is FY26 x 1.15 (checked line by line: every ratio is 1.15), so Mar-27 software is A$9.9m, which carries A$8.5m for the Mar-26 licence. Without it the plan is A$84.2m. Actual Jul to Sep 26 is A$19.4m against a plan of A$22.4m. | board_plan_fy27, software_licences | Gary to carry this into the target rationale. CFO to confirm whether the plan should repeat it. | Noted in v1.1 (target rationale) |
| C3 | Total revenue (K1) | Hardware revenue can be pulled into quarter-end months by volume rebates. In FY25, 46% of hardware revenue landed in quarter-end months (33% if even), and rebates appear only in those months. Gary's draft has no check on this. | erp_invoice_lines, gl_monthly | Gary to consider a rebate-rate or margin counter-metric on the page. Kewpie has used Gross margin % as the counter-metric and noted the rebate rate. | Open |
| C4 | New devices activated (K2) | The measure is only as good as the telemetry's coverage. Units invoiced per device activated were 1.91 to Jun-25 and 1.04 after. Either OEMs held large stocks, or the telemetry sees only part of the installed base. Both cannot be separated. | device_activations, erp_invoice_lines | Head of Platform Engineering to answer Q-4 (coverage, phone-home, firmware 4.5.0 from Jul-25). Keep K2 off page one until then, as Marcus asked. | Open |
| C5 | Software revenue growth (K3) | The target fails the Attainable test. Growth is +19.7% only because of one A$7.4m licence (-15.6% without it). To reach +20% for FY27 software needs A$2.78m a month for 9 months, against A$1.48m a month underneath. | gl_monthly, software_licences | Gary, VP Sales and CFO to set the target on repeat revenue, or to show growth with and without large single licences (Q-3). | Resolved in v1.1 (K3 now excludes one-off licences, target 0%) |
| C6 | Software revenue growth (K3) | Annual subscriptions appear to be booked in full at the contract start date: the licence table (by start date) equals ledger software revenue within A$27k a month. If so, software revenue is not spread over the term, and 'repeatable' needs a different measure. | software_licences, gl_monthly | CFO to confirm revenue recognition for subscriptions and for the Mar-26 licence. | Open |
| C7 | Share of software revenue from the largest licences (K3.1) | The licence is 29% of the year's started contract value and sits Red until it leaves the 12-month window in Mar-27, which fixes the figure without any change in behaviour. The customer is an OEM; naming is for the CFO and legal (Q-7). | software_licences | Gary to add 'with and without the largest licence' to the view. Customer name stays off the board page. | Noted in v1.1 (shown with and without the licence) |
| C8 | Gross margin % (K3.2) | Gary's target is 77%. The board's own FY27 gross margin target is 76% (BT-03, Finance). Also, 12-month margin includes the licence: 76.5% with, 74.1% without, so the 77% hold-the-line target is Red without it. | board_targets, gl_monthly | Gary and CFO to align the target with BT-03 and agree whether one-offs are shown. | Resolved in v1.1 (target 76%, with and without one-offs) |
| C9 | Design wins that ship (K4) | The target fails the Attainable test. Baseline 31.5%; pace needed +5.4 pts a month against a run rate of +0.08 pts. Marcus's 80% was a belief, never measured. The curve for the oldest cohort levels off near a third. | crm_opportunities | Gary and the VP Sales to reset the target once the Q-11 window and the CRM clean-up are agreed. Kewpie has not changed the target. | Resolved in v1.1 (target 35%) |
| C10 | Design wins that ship (K4) | The data cannot fully support the measure. 2,932 of 8,990 CRM rows (33%) have no stage dates; 33% of design-win-or-later records cannot be placed in a cohort, so the baseline is a Working Hypothesis. The draft's 'confirm shipping against invoices' has no product key between the CRM and the ERP. | crm_opportunities, erp_invoice_lines | VP Sales and Finance to add a shared product or SKU key; VP Sales to back-fill dates. | Actioned in v1.1 (data notes: date back-fill, shared product key) |
| C11 | Design wins that ship (K4) | Reps are paid A$400 per design win with no clawback. FY26: 1,180 wins, about A$472k (assuming every dated win was paid). The counter-metric (share lost after design win) is in place, but the data holds no reason codes, so a win logged only for the bonus cannot be told apart. | sales_incentive_plans, crm_opportunities | Gary to add a 'reason lost' field request to the Requirements. The incentive design is for the VP Sales, CFO and the people team, not Stage 2. | Actioned in v1.1 (reason lost request; bonus referred out) |
| C12 | Forecast accuracy (K5) | The measure is not yet one metric (Specific). Finance and Sales forecast differently (19% apart in the median quarter) and the CFO has not said which counts (Q-5). Neither has been within 5% since FY24-Q4, so even the 10% line is far from history: to reach the new 10% line Finance needs -2.4 pts a quarter. | forecast_snapshots, gl_monthly | CFO to choose the forecast and the tolerance (Q-5). Gary to keep K5 off page one until she does. | Partly resolved in v1.1 (10% and 20% bands; CFO still to choose the forecast) |
| C13 | Forecast accuracy (K5) | Data problems in the forecast table: 12 Finance rows in AUD thousands (fixed); the Sales snapshot for FY25-Q3 is missing; one Sales revision (13-Oct-25) falls after its quarter began. And Finance's 26-Jun-26 hardware forecast (A$7.1m) is 52% below Finance's own 15-Jun-26 plan (A$14.8m). | forecast_snapshots, board_plan_fy27 | Add an 'amount_unit' check to the data load. CFO to explain the gap. | Open |
| C14 | OEM stock on hand (K2.1) | Gary's title says top 40 OEMs. Only 35 report, covering 46% of FY26 hardware revenue, and 235 spellings of the names had to be cleaned. The measure has no target or thresholds yet. | oem_inventory_reports, oem_master | Gary to set thresholds from the baseline, for example on months of cover (10.9 at the reporters now). That would be a new measure, so it is Gary's call. | Resolved in v1.1 (replaced by Months of stock cover, 6 and 9 month lines) |
| C15 | Design wins per quarter (K4.2) | Design wins have risen from 230 to 273 a quarter since the bonus, with smaller estimated deal size and no change in the shipping rate. It remains a sub-KPI as Gary proposed. | crm_opportunities | No change. Read it with Design wins that ship. | Logged |