For CPG supply chain leaders
Worked Fine.
Then Broke.
The six checks before you sign with a platform
pixelsandclicks.com
swipe →
The surprising minimum
Oracle requires zero years of history
NetSuite's demand-planning module states no minimum. What matters is continuity, not depth.
Oracle NetSuite documentation
02 / 10
But continuity still matters →
The real cause
Your vendor didn’t sell a broken algorithm
Netstock, o9, and Blue Yonder all run defensible statistical methods.
A platform can run that calculation on bad data and still be wrong.
Six checks decide whether that’s your outcome.
03 / 10
The signal you're blending
Two demand signals. One number.
Baseline
Steady-state volume from distribution and repeat purchase.
Moves slowly. This is what history should show.
Promotional Lift
Incremental volume one specific promotion generates.
Event-driven. Varies by retailer and promo type.
04 / 10
Blend them and it breaks →
The silent multiplier
Demand × Wrong UOM = 12x Off
A UOM mismatch doesn't throw an error. It looks like real demand, twelve times higher or lower than reality.
05 / 10
Standardize UOM first →
Without change control
One bad edit breaks two ways
Splits It
One continuous history becomes two shorter, noisier ones.
Blends It
Two different items get treated as one that never existed.
06 / 10
The model can't tell why →
The self-reinforcing one
A stockout looks like zero demand
Sales history records both the same way: zero units sold.
The model reads that zero as low demand, and forecasts even less next time.
07 / 10
Fastest movers stall first →
Track these separately
One MAPE hides the real miss
Blending baseline and promo accuracy hides which one's driving misses.
Baseline
Forecast accuracy with no promotion running.
Promo Lift
Accuracy by event type, tracked separately.
Bias
Consistently over- or under-forecasting?
08 / 10
This doesn't need a platform →
Before you sign anything
Three verdicts. Pick honestly.
Build now, fix the data first, or don’t buy yet.
The fix isn’t a support ticket. It's the data work you needed before you signed.
Free readiness quiz on pixelsandclicks.com
09 / 10
Pixels and Clicks
AI forecasting readiness for CPG suppliers
14-day feasibility read: build now, fix data, or don't buy
Vendor-agnostic. No referral fees from any forecasting platform
Free readiness quiz and demand-forecasting article, on our site
pixelsandclicks.com