DMF Stock Intelligence

Reorder recommendations for David Musson Fencing, computed from OrderWise data.

Order-now queue · data cut-off 2026-06-16

17 lines flagged

5,838 units below reorder point across those lines 444,349 of annual throughput). Historical snapshot — becomes a live daily view once DMF enable live-DB access.

17
Order now
4
Order soon
74
OK
5
Need data
How to use this alongside your current process

For the first few weeks, treat this as a second pair of eyes — your existing OrderWise routine stays primary. Once you see where the system agrees with you (and where it usefully disagrees), the trust naturally moves over.

  1. Open the dashboard alongside your normal OrderWise review. The “Order now” list is the same conversation you’d have looking at OrderWise min-level alerts — except it’s already filtered to lines where our view of cover + lead-time + seasonal demand says you should be acting this week.
  2. Mentally tag each line: agree, disagree, or surprise. “Agree” lines confirm your judgement (and ours). “Surprise” lines — flagged by us but not on your radar — are where the value usually shows up. Disagreements are the most useful: they’re what tells us how your judgement differs from the model so we can tune it.
  3. Place orders the way you always do, in OrderWise. Nothing here writes back. We watch what you actually order and learn from the gap between our suggestion and your decision — without you having to type anything extra.
Live mode: OFFsource: sample

Viewing the 2026-06-16snapshot — the recommendations the engine would have produced on that date. When the live OrderWise connection is switched on, the same engine runs every night and this becomes a genuine “today” view, with a daily “what changed since yesterday” delta replacing the historical Nov→Jun comparison shown above.

Model performance (Phase 1 forecasting backtest on sample variants)

Each forecast was trained on earlier years and scored against a year it never saw. Per-line accuracy is inherently noisy for individual timber lines; the service-level reorder math above doesn’t require accurate per-line point forecasts — it routes volatile lines to deeper safety stock instead.

Weighted accuracy
54%

1 − WAPE, volume-weighted

Median line accuracy
49%

typical line, 1 − WAPE

Interval coverage
63%

actuals inside ~90% band

Variants
100

years 2023–2024–2025–2026