Time series look at the past to tell you how much will be sold. The LP model takes that and decides how much is best to sell and buy to earn more without breaking business rules. They do not compete: one predicts and the other decides.
The forecast is the weather report. LP is deciding what to do with that weather.
The weather tells you it will probably rain tomorrow. It doesn't tell you whether it's best to cancel the delivery, change the route, or bring more umbrellas to the store: that depends on your trucks, your costs, and your inventory. That second part is what the LP model solves.
"How much are we going to sell?"
It looks for the trend and seasonality in what already happened and projects them forward (moving average, regression, ARIMA). It delivers a figure per month and how reliable it is: the band opens up because the further out, the less certainty.
it doesn't know how much is in the warehouse or how much margin each unit leaves
"How much is it best for us to sell?"
It takes demand as a ceiling and combines everything at once: margin, inventory, purchases, delivery times, budget, and targets. It delivers decisions: what to sell, where, when, and how much to buy, and which rule limits profitability.
it needs to know how much the market is asking for: that's where the forecast comes in
Cream A: the forecast says 70 units and leaves $80 each. Serum B: the forecast says 60 and leaves $50. There are 100 units in the warehouse.
| How the plan is built | Cream A | Serum B | Is it feasible? | Margin |
|---|---|---|---|---|
| Copy the forecast as is | 70 | 60 | No: 30 units missing | — |
| Cut the forecast proportionally | 54 | 46 | Yes | $6,620 |
| LP model | 70 | 30 | Yes | $7,100 |
With the forecast alone, the plan is either impossible or leaves money on the table. LP uses the same forecast and chooses the distribution that yields the most.
| Time series | LP model | |
|---|---|---|
| The question | How much will be sold? | How much is best to sell and buy? |
| What it looks at | Only the history of the variable (and its seasonality) | Everything together: demand, margin, inventory, purchases, deliveries, budget, targets |
| What it delivers | A figure per month with its confidence range | Decisions by product, zone, and month, and the value of each constraint |
| Business rules | It doesn't know them | It respects them all; a plan that breaks a rule does not exist |
| Money | It doesn't distinguish what leaves more margin | It prioritizes what leaves the most margin |
| If something changes | It is adjusted again with new data | The data is changed (price, cost, warehouse) and it recalculates in seconds |
| Its weak point | It assumes the future resembles the past | It is only as good as the demand we give it |
| In Infinity Workflow | Data analytics › Case design (trends, ARIMA, and ARMA) | Planning › LP sales optimization |