The terms "demand sensing" and "demand forecasting" appear in the same vendor conversations, the same conference presentations, and often the same budget proposals. They're not the same thing. They solve different problems, operate on different time horizons, and feed different supply chain decisions. Conflating them — or spending on one when you actually need the other — is a common and expensive mistake in CPG planning stacks.
Demand Sensing: The 0–7 Day Horizon
Demand sensing is a short-horizon signal refinement process. Its job is to take the current-period forecast that was generated 4–12 weeks ago and adjust it in near real-time based on signals that weren't available when the original forecast was made. The sensing horizon is typically 0–7 days, rarely extending beyond 14 days.
The data inputs are fast-signal, near-real-time: daily or weekly POS scan data (if available), point-of-sale velocity by store or distribution center, current promotional lift actuals, and early-week replenishment order patterns. Demand sensing applies these inputs to revise this week's or next week's demand estimate.
The operational decisions demand sensing serves are within its time horizon: distribution center replenishment priorities, same-week production schedule adjustments, and transportation routing when demand is running materially above or below the weekly plan. These are execution-layer decisions, not planning-layer decisions.
Demand sensing doesn't change your purchasing commitments. It doesn't affect your 8-week production schedule. It can't move inventory that hasn't been built yet. Its value is in reducing short-horizon execution errors — the gap between what the weekly plan said and what demand is actually doing this week.
Demand Forecasting: The 4–16 Week Horizon
Demand forecasting operates on the S&OP planning horizon: typically 8–16 weeks, with 12 weeks being the standard for most CPG procurement and production cycles. Its job is to predict demand far enough in advance that supply chain decisions can respond: production runs can be scheduled, raw material can be purchased, DC inventory can be pre-positioned, and buyer orders can be submitted with the right quantities.
The supply chain decisions that demand forecasting serves are commitment-layer decisions. A production order placed today for a 6-week lead-time item is a commitment — you can't undo it if the forecast turns out to be wrong. Safety stock levels set for a seasonal window are commitments. Club buyer order submissions are commitments. These decisions require a forecast that's accurate far enough out that there's still time to act on it.
This is precisely why external signals matter so much at the forecasting horizon. By the time your 7-day demand sensing layer confirms that a trend is real, your lead time has already consumed most of your response window. The manufacturing order that would have serviced the peak demand needs to have been placed 5–6 weeks ago. Only a 4–12 week forecast model that incorporates pre-POS signals can place that order on time.
The Confusion Point: "Real-Time Data" Doesn't Mean "Better S&OP"
The demand sensing category has been marketed aggressively over the past several years, particularly by tools that offer daily or even intra-day POS data integration. The pitch is compelling: "stop running your business on last week's scan data." And within the demand sensing scope — improving this week's execution — that pitch is accurate.
The confusion happens when demand sensing tools are positioned as S&OP planning solutions. The argument goes: "with real-time signals, we can revise the 12-week forecast continuously." This is technically true but misses the operational constraint. Revising a 12-week forecast in week 10 is interesting analytics. It doesn't help you replenish the inventory that you should have built in week 4. The forecast revision arrives after the commitment window has already closed.
We're not saying demand sensing tools are useless. For CPG brands with short lead times (private label, simple formulations, proximity manufacturing), demand sensing with daily POS can meaningfully reduce weekly execution errors. But that's a different value proposition than improving 12-week S&OP accuracy — and the two should be evaluated and budgeted separately.
Where Signal Fusion Sits in This Framework
Heatvelo operates in the demand forecasting layer, not the demand sensing layer. This is a deliberate positioning decision, not a limitation.
The external signals we fuse — weather patterns, social trend velocity, macro economic indicators — are specifically useful because they provide lead time before demand materializes in POS. A social trend signal that builds for 2 weeks before retail demand shows up gives a 4–6 week lead time advantage when combined with your production lead time math. A weather pattern forming 3 weeks before a seasonal event gives a replenishment pre-positioning window. Macro indicators building for 6–8 weeks before trade-down behavior hits POS allow supply teams to pre-build mid-tier SKUs ahead of the shift.
None of this is useful in a demand sensing context. You don't need a 6-week social trend lead to make this week's distribution center replenishment decision. Demand sensing needs fast POS data and fast fulfillment — that's a different tool set entirely.
Matching Tool Type to Planning Decision Type
A useful way to audit your current planning tool stack:
| Planning decision | Time horizon needed | Tool type needed | Key signal inputs |
|---|---|---|---|
| DC replenishment prioritization | 0–7 days | Demand sensing | Daily POS velocity, real-time inventory |
| Production run scheduling | 4–8 weeks | Demand forecasting | POS baseline + external signals |
| Raw material purchasing | 6–12 weeks | Demand forecasting | POS baseline + macro + weather |
| Club buyer order submission | 8–12 weeks | Demand forecasting | POS baseline + social trend velocity |
| S&OP quarterly planning | 12–16 weeks | Demand forecasting | Full signal fusion |
| This week's shipment execution | 0–3 days | Demand sensing | Same-day POS, order actuals |
The honest audit question for most CPG planning teams: which of these decisions is currently producing the most planning waste — execution errors (pointing to demand sensing gaps) or commitment errors (pointing to demand forecasting gaps)? Where you're losing revenue from stock-outs on trending SKUs, or carrying excess inventory from a production over-run, those are commitment errors. Demand sensing won't fix them.
The Combined Stack: When Both Matter
Some CPG supply chains genuinely need both. A brand with very short lead times (3–5 days from production to shelf for fresh or chilled categories) operates in a demand sensing world almost entirely — their commitment horizon is too short for 12-week forecasting to drive production decisions. A brand with 10-week manufacturing lead times and heavy promotional calendars needs accurate demand forecasting but may have limited use for daily POS sensing (they can't react within a week anyway).
Most shelf-stable CPG brands with 4–8 week lead times occupy the middle: they need demand forecasting for commitment-layer decisions and can benefit from demand sensing for execution refinement. In this scenario, the tools are complementary rather than competing. The right evaluation sequence is to fix your forecasting horizon first — because that's where the largest planning errors typically originate — and then optimize short-horizon execution once the long-horizon baseline is solid.
If your S&OP cycle is still running on ERP POS-only forecasting and your 12-week accuracy is in the 65–75% range, adding real-time demand sensing won't solve that problem. You'll get more precise visibility into an already-inaccurate plan. Fix the plan first.