Planning Benchmark Club

What this is

Why it matters: Every planning team tracks its forecast error and bias. Almost none can say whether their number is good, because every company defines the metrics differently. The same forecast can come out as a 12% miss in one building and a 25% miss in another, only because the math differs.

What you get: Your WMAPE and forecast bias for the last 3 months, calculated on this site with one locked formula set. Your percentile against the full pool and your peer group. The cohort median and the top-quartile line. And the metric with the bigger gap named as your first fix.

WMAPE, in plain words: WMAPE is the share of your total demand that the forecast missed over the last 3 months, weighted so your big items count more than your small ones. A WMAPE of 12% means the forecast missed by about 12 units for every 100 units of demand. Lower is better.

Why it's safe: No company name is collected, and you use placeholder SKUs instead of your real UPCs, so nothing you submit identifies your products. Nothing can be traced back to your company. A contributor code is your only identity here. The full detail is in "Why this is safe" below.

How it works

Request a contributor code below.
Paste your forecast and actual units for the last 3 months, item by item, with placeholder SKUs (SKU-001, SKU-002) in place of your real UPCs. Use the same placeholder for the same item in every month. Add your industry, company size, and how you mainly sell. The site does the math and shows you the result before anything is saved.
See your standing. Standings unlock once 10 founding contributors are in for the window.

The fine print: This is a founding-cohort test. Five more metrics are already defined and locked (forecast stability, forecast value add, cover vs. target, service level, excess and obsolete share) and join in a later phase.

The current submission window is loading, the last 3 months, rolling. Definitions version v2.

The founding cohort

Loading cohort state…

Why this is safe

What the site collects: Your forecast and actual units for the last 3 months, item by item, under placeholder SKUs you choose. Your industry, your company size range, and how you mainly sell. And the two metrics the site calculates from your rows.

What the site never collects: Your company name. Your name or email. Your real UPCs or product names. Customer names. Prices, revenue, or any dollar figures.

How your identity works: A contributor code is your only identity here. Codes are stored hashed, and a submission carries nothing that links it back to a company.

How results are shared: Only as group numbers, and only once a group reaches 10 contributors, so no single member's figures can be picked out. Demo rows on this site are flagged as demo and are never counted in real standings.

Who runs the math: The site does, with locked formulas that are the same for every member. You see your calculated metrics and confirm them before anything is saved, and submitting again for the same window replaces your earlier submission.

Request a contributor code

Leave a short note so the admin can route the code back through your group. Do not include your company name.

Members

Submit your window

Industry

Company size (yearly revenue)

How you mainly sell

My actuals are shipments, not demand (orders). My WMAPE and bias results will be marked as shipment-based.

Your item rows: forecast and actual units, last 3 months

One row per item per month, in units. Before you enter anything, replace your real UPCs with placeholder SKUs (SKU-001, SKU-002, and so on), and use the same placeholder for the same item in every month. The forecast is the final consensus forecast for that item and month, frozen one month before the month it covers. The actual is demand (orders) for that month, or shipments if you ticked the box above. The site sums the errors at item level itself, so enter the row-level numbers from your forecast review rather than any totals you have already calculated.

The months in this window are loading. Every item you include needs a row for each month it was active, and the submission needs rows in all three months.

Paste rows from a spreadsheet or CSV export Four columns: Item, Month, Forecast, Actual. The first column holds your placeholder SKUs, not real UPCs. The month can be written as 2026-08 or Aug 2026. A header row is fine; it is skipped. With hundreds of items this is the fast path: swap the UPC column for placeholders in your export, then copy the block and paste it here.

Or type rows in directly

For a small set of items, rows can be typed into the grid instead of pasted. Blank rows are ignored, and rows from both places are combined into one submission.

Item	Month	Forecast units	Actual units

My results

Loading…

The launch metrics

The site computes both metrics the same way for every member, from the item-level rows each member submits. The formula, the level, and the lag are part of the metric, and because the site runs the formulas, a number cannot enter the cohort computed any other way. The window is the last 3 months, rolling, and actuals are demand (orders), never shipments, unless the shipment-based flag is set on the submission.

Five more metrics are defined and locked alongside these two: forecast stability, forecast value add, cover vs. target, service level, and excess and obsolete share. They join in a later phase, under the same definitions, and the launch asks for WMAPE and bias only so the first cohort can form quickly.

Rules that apply to every submission: item-level rows for your own portfolio only, with placeholder SKUs instead of real UPCs, and no customer names and no company name. Changing a definition starts a new series; it never rewrites history.