ORDOrders & serviceStop 4 of 9
Substitution
Offering the next-best product when an item is out of stock, without hurting margin, trust or the forecast.
Read atChange any time
A customer orders the 500 ml premium shampoo, but it’s out of stock. Instead of losing the sale, the company can offer an alternative: the 400 ml premium pack, or two 250 ml packs, or the 500 ml standard version. That’s substitution.
There are two kinds:
- Customer-driven: the shopper picks another product on their own when the first choice is missing, as happens on a shelf.
- Company-driven: the seller recommends or ships an alternative, as on an e-commerce page or for a distributor’s order.
A good substitute has to fit: is it compatible, will the customer accept it, how does its price and margin compare, and is the substitute itself available? Common rules include:
- Upgrade at the same price (downward substitution): give the bigger or better pack to protect the relationship, which costs margin but keeps loyalty.
- Size or pack swap: two smaller packs for one big one.
- Customer approval first for anything significant, such as different specifications or a different brand.
Substitution also affects planning. Sales that move from A to B during A’s stockout must be recorded correctly. Otherwise the forecast thinks demand for B grew and for A fell, and the next plan buys the wrong mix.
Words to use
- Substitution
- Serving demand for an unavailable product with an alternative.
- Downward substitution
- Supplying a higher-grade item in place of a lower-grade one that is out of stock.
- Demand transfer
- Sales shifting from one product to another because of a stockout or promotion.
Which substitute to offer?
Three options, with the margin each earns and the chance the customer accepts it:
| Option | Margin if accepted | Chance of acceptance | Expected margin |
|---|---|---|---|
| Upgrade: 650 ml premium at the 500 ml price | ₹50 | 95% | ₹47.50 |
| Standard 500 ml, refunding the price difference | ₹60 | 60% | ₹36.00 |
| 400 ml premium at a lower price | ₹70 | 40% | ₹28.00 |
The upgrade earns the least per unit but the most in expectation, and keeps the customer happy. Acceptance rates come from history: how often customers kept or rejected each kind of substitute.
The rules should also say when not to substitute: different specifications for a technical product, allergens in food, a different brand when the customer is loyal to one, or any substitute the customer has opted out of.
Correcting the demand history
Suppose product A was out of stock for two weeks and product B, which normally sells 1,000 a week, sold 1,150. If the history is left as it is, the next forecast sees A’s demand falling and B’s rising by 15%. Planners should:
- Flag the stockout period for A, so its zero sales aren’t read as zero demand.
- Estimate A’s true demand in those weeks from its normal rate.
- Move the substituted sales (about 150 a week here) back from B to A in the forecasting history.
Substitution as pooling
When customers will accept B in place of A, the stock of B partly covers A’s uncertainty too. Planned well, this is a form of risk pooling: less total stock is needed for the same service, especially with one-way (downward) substitution, where a higher-grade item can always replace a lower-grade one.
Modelling customer choice
How customers substitute when an item is missing can be estimated from sales data. Choice models, such as the multinomial logit, describe the probability that a customer picks each available product. Kök and Fisher (2007) showed how to estimate substitution rates from store data and use them to choose assortments and shelf space. With such a model, planners can predict how much of a stockout’s demand moves to other items and how much is lost.
Downward substitution in production
In some industries a better product can always stand in for a lesser one. Semiconductor makers sort chips by measured speed (binning), and a faster chip can be sold as a slower model if the slower one runs out. Steel and paper mills can supply a higher grade in place of a lower one. Inventory models with downward substitution (studied by Bassok, Anupindi and Akella, 1999) decide how much of each grade to stock, given that the higher grades act as a shared buffer.
Online grocery practice
Grocery delivery depends on substitution because many items run out between order and picking. Good practice includes letting customers set preferences (“no substitutes”, “same brand only”), guiding pickers with ranked suggestions learned from what customers accepted before, showing substitutions clearly at delivery with an easy way to refuse, and feeding refusals back into the suggestions.
Regulated substitution
Some substitution is controlled by law or contract. Pharmacies can switch a branded medicine for a generic only under the applicable rules; food service may need customer approval for allergen-related changes; industrial customers may require formal approval for any part change. Rules engines must respect these limits.
Interview kit
Practise it out loud
Answer each question yourself before opening the model answer.
When an item is out, substitution offers an acceptable alternative to save the sale. Rules should cover compatibility, customer acceptance, price and margin, and the substitute’s own availability, with customer approval for significant changes. I’d also make sure forecasting sees the true demand, separating sales that transferred because of a stockout, so the next plan buys the right mix.
How would you design substitution rules for an online grocery?
For each product, define acceptable substitutes ranked by similarity: same brand in a different size, then a similar product from another brand, within a price band. Let customers opt in or out per item, avoid substitutes with allergens or dietary conflicts, and track acceptance and return rates to refine the rules.
Why does substitution matter for forecasting?
Because during a stockout, sales of the substitute rise and sales of the original fall to zero, which isn’t true demand. If the history isn’t corrected, the forecast overstates the substitute and understates the original, so the next replenishment repeats the shortage.
Connected stops
Where this idea feeds in, or gets its inputs from.
Sources and further reading
- Chopra, S. & Meindl, P. Supply Chain Management: Strategy, Planning, and Operation. Pearson.
- Silver, E. A., Pyke, D. F. & Thomas, D. J. (2017). Inventory and Production Management in Supply Chains (4th ed.). CRC Press.
- Kök, A. G., & Fisher, M. L. (2007). Demand estimation and assortment optimization under substitution: Methodology and application. Operations Research, 55(6), 1001–1021.
- Bassok, Y., Anupindi, R., & Akella, R. (1999). Single-period multiproduct inventory models with substitution. Operations Research, 47(4), 632–642.