---
title: "Grocery Substitution Management in 2026: Why the Approval Window Closes Before the Customer Answers"
id: "26800"
type: "post"
slug: "grocery-substitution-approval-window"
published_at: "2026-09-22T15:00:00+00:00"
modified_at: "2026-09-22T12:33:50+00:00"
url: "https://locus.sh/blogs/grocery-substitution-approval-window/"
markdown_url: "https://locus.sh/blogs/grocery-substitution-approval-window.md"
excerpt: "Customer response time does not shrink when the delivery promise does. Below a 15-minute pick window, asking for substitution approval stops working."
taxonomy_category:
  - "General"
---

#### [General](https://locus.sh/blogs/category/general/)

# Grocery Substitution Management in 2026: Why the Approval Window Closes Before the Customer Answers

[Ishan Bhattacharya](/author/ishan_locus/)

Sep 22, 2026

16 mins read

Grocery substitution management is how a retailer resolves an order line that cannot be filled as ordered, in the interval between the picker finding the gap and the vehicle leaving. The accepted best practice is to ask: push the proposed replacement to the customer, let them approve it, decline it for a refund, or choose an alternative. That practice was designed for orders picked hours before dispatch, and it is being carried unchanged into express and quick commerce operations where the entire promise is shorter than the time a customer takes to read a notification. Modeling the decision across promise lengths shows the break clearly. Human response time is the one variable that does not scale with the delivery window, and platforms built to decide before the gap appears, Locus among them, are where the answer has to move.

## Key Takeaways

- Reaching a 50% substitution answer rate requires a decision window of about 12 minutes, which is longer than an entire quick commerce promise.
- The answer rate has a ceiling near 62% set by engagement, not by time. Extending the window from 45 minutes to three hours adds 1.4 points.
- Compressing the window from 180 minutes to 2.5 minutes roughly doubles the share of substituted items rejected at the door, from 9.5% to 20.3%.
- At a 2.5-minute window, 81% of customers are told about a problem and given no usable say in it, which is worse than a good default applied silently.
- Substitution preferences captured at order time covering 60% of lines beat a three-hour real-time ask, and they work identically at 12 minutes and 12 hours.
- Locus resolves the substitution against the plan before the customer is involved, so the question asked is one the customer can still answer.

## Why Substitution Management Matters: The Business Case

Basket change is not an edge case in grocery. McKinsey’s work on omnichannel grocery fulfillment finds that in a pick-from-store model, [10% to 15% of orders](https://www.mckinsey.com/industries/retail/our-insights/creating-a-competitive-edge-in-omnichannel-grocery-fulfillment)
 will experience stockouts and potentially require substitution during fulfillment, and that poor execution on out-of-stocks destroys customer trust. At one in seven to one in ten orders, substitution is a core flow rather than an exception queue.

The customer’s reaction is policy-sensitive and measurable. Research published in the Journal of Retailing on post-purchase out-of-stock in online grocery finds substitution acceptance rises from [66% for random selection](https://www.sciencedirect.com/science/article/pii/S002243592200046X)
 to 75% when the replacement follows a recommended policy matched on the category’s dominant attribute, and that stock-outs have a pervasive negative effect on assessment of the transaction and on repurchase intention. Acceptance is something the retailer engineers, not something it discovers.

| Also Read: Online Grocery Order Tracking: Item-Level Visibility 2026 |
| --- |

What has changed is the clock. Brick Meets Click data reported by Digital Commerce 360 shows [online grocery passing 19% of category sales](https://www.digitalcommerce360.com/article/online-grocery-sales/)
, with delivery in one hour or less now accounting for 18% of delivery orders. Nearly a fifth of the category has moved to a promise short enough to break the approval flow, and the flow itself was never redesigned. With last mile running [60% to 70% of total parcel delivery cost](https://www.mckinsey.com/de/publikationen/2024-10-28-ooh-delivery)
 by McKinsey’s estimate, an item rejected at the door is the expensive leg paid twice for a line the customer never wanted.

The rejected unit rarely recovers its value either. ReFED’s U.S. Food Waste Report puts [70 million tons of surplus food in 2024, roughly 29% of the domestic food supply](https://refed.org/food-waste/the-problem/)
, against a $380B surplus value. A chilled or short-dated item that comes back on the van has left its temperature chain and its shelf life has already been spent on the trip, so a declined substitution in fresh categories is usually a write-off rather than a restock. That makes the door rejection rate a margin number, not only a satisfaction number.

## How the Substitution Approval Window Closes

The model below uses explicitly illustrative inputs: 62% of customers will engage at all with a mid-order push, and among those who do, response time is lognormal with a median of 4.5 minutes and a heavy tail. Acceptance of a delivered substitution is 100% where the customer chose it, 75% where a matched policy chose it, and 66% where the selection was arbitrary, per the research cited above. These are model inputs, not measured Locus averages.

### Step 1: The decision window is a residual, not a design choice

The window is what remains after the picker finds the gap and before the vehicle leaves. On an order picked the night before, that is hours. On a 12-minute quick commerce order where picking starts at zero, the gap surfaces around 90 seconds in and the rider departs at roughly four minutes, leaving a window of about two and a half minutes. Nobody chose that number. It fell out of the promise.

### Step 2: Human response time does not scale with the promise

This is the structural break. Every other quantity in a quick commerce operation compresses proportionally: pick time, travel time, dwell, the ETA refresh interval. The time a person takes to notice a push, open it, think about whether they want oat milk instead of almond, and tap a button is a human constant. It is the same 4.5-minute median whether the promise is 12 minutes or 12 hours.

That asymmetry is worth stating plainly, because it is the only thing in the operation that does not respond to engineering. A faster picker shortens the pick. A denser store network shortens the ride. A better forecast shortens the wait. None of them shorten the interval between a phone buzzing and a person deciding whether they will accept a different brand of yogurt. Every efficiency gain elsewhere in the chain makes the mismatch worse, because it shrinks the window while leaving the response distribution exactly where it was.

### Step 3: The answer rate collapses on the steep part of the curve

| Decision window | Answer rate | Substituted items rejected at the door |
| --- | --- | --- |
| 2.5 minutes | 18.9% | 20.3% |
| 5 minutes | 33.3% | 16.7% |
| 15 minutes | 52.9% | 11.8% |
| 45 minutes | 60.6% | 9.9% |
| 180 minutes | 62.0% | 9.5% |

Compressing the window from three hours to two and a half minutes roughly doubles door rejection. The curve is flat on the right and steep on the left, and express grocery sits on the steep part.

### Step 4: More time is worth almost nothing above 30 minutes

Moving the window from 45 minutes to 180 minutes raises the answer rate by 1.4 points. Moving it from 2.5 minutes to 5 minutes raises it by 14.4 points. The ceiling is engagement, not patience, and it sits near 62%. Operations teams that respond to substitution complaints by picking earlier are buying on the flat part of the curve.

| Also Read: Grocery Route Optimization: Planning Before the Pick |
| --- |

### Step 5: The window you would need is longer than the promise

To reach a 50% answer rate, the model requires a decision window of about 12.2 minutes. A quick commerce operation promising 10 to 15 minutes cannot allocate 12 minutes to a question, because the question would consume the entire product. There is no schedule change, no picking sequence and no notification design that resolves this. The ask does not fit.

### Step 6: Asking without getting an answer is its own harm

At a 2.5-minute window, 100% of affected customers are notified and 18.9% respond. The remaining 81.1% receive a message telling them their order has a problem, containing a control they cannot use in time, followed by a bag that reflects a decision someone else made anyway. They were given the anxiety of the exception and none of the authority. A well-matched substitution applied silently produces a better outcome than a question the customer could never answer.

This runs against instinct, because transparency is normally the safe default in delivery experience and is normally correct. The exception is narrow and specific: transparency helps when it precedes a choice the recipient can still make. Once it cannot, the same message is an advance notice of a disappointment, delivered with enough time for the customer to anticipate it and not enough to change it. Measuring notification delivery rather than notification usefulness is what keeps this invisible, because on a dashboard the unanswered 81% and the answered 19% look identical.

### Step 7: Moving the decision to order time removes the dependency

A substitution preference captured at checkout, per line or per category, resolves without a real-time exchange. It works identically at 12 minutes and 12 hours because it no longer competes with the promise.

| Preset coverage of order lines | Door rejection at a 2.5-minute window |
| --- | --- |
| 0% | 20.3% |
| 40% | 12.2% |
| 60% | 8.1% |
| 80% | 4.1% |

Preset coverage of 60% at a two-and-a-half-minute window beats a three-hour real-time ask, which lands at 9.5%.

## Real-Time Approval vs Pre-Set Preferences: Key Differences

| Dimension | Real-time approval | Pre-set preferences |
| --- | --- | --- |
| When the customer decides | During picking | At checkout |
| Depends on promise length | Yes, heavily below 15 minutes | No |
| Answer or coverage ceiling | About 62%, set by engagement | Set by how many lines carry a preference |
| Works in quick commerce | Not at the observed window lengths | Yes, unchanged |
| Data required | Live pick events pushed to the customer app | Preference capture at the line or category level |
| Failure mode | Customer told, cannot respond in time | Preference is stale or the line has no preference |
| Cost of a miss | Item rejected at the door, return leg incurred | Falls back to policy-matched selection |
| Customer perception | A question they could not answer | A default they set themselves |

Neither column is complete on its own. Presets cover the lines a customer has an opinion about; real-time approval is still the right mechanism when the window genuinely exists and the substitution is unusual. The error most operations make is not choosing wrongly between the two, it is running only the right-hand column in scheduled delivery and then keeping it when the promise shortens, so the mechanism degrades without anyone changing a setting. A substitution system should treat the window as an input and select the mechanism per order, rather than treating the mechanism as a fixed product decision made once.

## What to Look for in Grocery Substitution Software

**Preference capture that does not sit in a settings screen.** A preference nobody sets covers nothing. The capture has to happen where the line is added, defaulting sensibly by category, because coverage is the only variable that moves the outcome once the window is short. Prior purchase history is the cheapest source of coverage available, since the same research finds [customers are considerably more likely to accept a product they have bought before](https://www.sciencedirect.com/science/article/pii/S002243592200046X)
, and that history already exists without asking anyone anything. Treating a previously purchased alternative as an implicit preference converts dormant transaction data into coverage on day one, with no customer action and no interface change, which is the only lever in this model that costs nothing and works immediately at every promise length.

**A matched-substitution policy, not a nearest-price rule.** The published gap between arbitrary and attribute-matched selection is nine points of acceptance, and that gap is what carries every line the customer did not pre-set and did not answer. Matching on the category’s dominant attribute is the documented mechanism, and the dominant attribute differs by category, which is why a single global rule underperforms. Size governs some categories, brand others, and dietary constraint overrides both wherever it applies.

| Also Read: Last-Mile Delivery for Quick Commerce 2026 |
| --- |

**Pick events streamed, not batched at dispatch.** A substitution known at handover can only be reported. A substitution known during picking can still be decided. Platforms that receive the final basket at dispatch are structurally limited to transparency, and the difference shows up in [what the routing layer can still change](https://locus.sh/route-planning-system/)
 before the vehicle leaves.

**A window-aware decision policy.** The system should know how much time remains before departure and change its own behavior accordingly: ask when the window supports it, apply the matched default when it does not, and never issue an approval request it cannot honor. This is the single capability most substitution flows lack.

**Payment adjustment before the statement.** The authorized amount has to reflect the delivered basket before the customer sees a charge they did not expect. Declined substitutions and short picks become a settled number rather than a dispute, which removes a contact class instead of handling it faster, and it needs [visibility that spans the order rather than the vehicle](https://locus.sh/control-tower-software/)
.

## Substitution Management in Action: Real-World Results

**A grocery brand delivering fresh and perishable orders across more than 30 cities** ran this problem on contracted third-party operators rather than an owned fleet, where a missed window means spoilage rather than a late parcel. With Locus orchestrating allocation and execution, the operation recorded [33% faster deliveries and 15% lower fulfillment cost](https://locus.sh/case-studies/grocery-carrier-orchestration/)
, alongside 25% less manual shipping time and customer support resolution 10 to 20 times faster. The support figure is the substitution-relevant one, because basket disputes are resolved against a record of what was picked and when rather than reconstructed after the fact.

**A leading North American retailer across multiple hundred stores** consolidated six legacy systems into one planning and execution layer and now [resolves exceptions in under two hours](https://locus.sh/case-studies/retailer-multimodal-logistics-automation/)
, with 99%+ on-time delivery, 95%+ route compliance and $1M+ in savings at break-even inside year one. A sub-two-hour exception cycle is what makes a decision window exist at all. Where detection is slower than the window, the question is moot before it is asked.

**The model supplies the third result.** Holding everything else constant and moving only the decision window from 180 minutes to 2.5 minutes raised door rejection from 9.5% to 20.3%, with no change in picking quality, substitution policy or notification design. The operation got worse at substitutions by getting faster at delivery, which is the trade nobody wrote down.

| Also Read: Delivery Experience Optimization for E-Grocery 2026 |
| --- |

## Common Substitution Mistakes to Avoid

**Carrying the e-grocery approval flow into express operations unchanged.** The flow assumes a window that express fulfillment does not have, and it fails silently because the notification still sends and the dashboard still records it as delivered.

**Treating a low answer rate as a notification problem.** Copy, channel and timing are tuned against a ceiling set by engagement. Below 15 minutes the binding constraint is the window, and no message design recovers it.

**Buying decision time by picking earlier.** Above 30 minutes of window, additional time returns almost nothing. The effort is better spent on preference coverage, which has no ceiling imposed by the clock.

**Asking when the answer cannot arrive.** An approval request inside a two-minute window informs the customer of a problem and hands them a control that expires before they reach it. A matched default applied without comment is the better outcome.

| Also Read: Best TMS for Grocery and Food Delivery 2026 |
| --- |

## How Locus Handles Substitutions Against a Moving Promise

Locus, the world’s first Decision-Intelligent, Agentic TMS, treats a substitution as a plan event rather than a message event. The Hub agent sees pick progress as it happens, the Dispatch agent knows how much time remains before departure for that specific order, and the Customer agent is therefore able to decide whether a question can still be answered before it asks one. Where the window supports an ask, the customer gets a real choice. Where it does not, the matched policy applies and the customer is told what arrived rather than asked about what might. That branch is the capability, and it exists because promise time and customer communication are governed in the same decision layer instead of in two systems that never compare clocks.

The platform reasons across more than 250 real-world constraints and has orchestrated 1.5B+ deliveries for 360+ enterprise customers in 30+ countries at 99.99% uptime, with $320M+ in aggregate logistics cost savings. Locus has been [recognized by Gartner for seven consecutive years](https://locus.sh/analyst-recognition/)
, including the 2026 Gartner Hype Cycle for Supply Chain Execution and Logistics Technologies and the 2026 Gartner Market Guide for Multicarrier Parcel Management Solutions, where ShipFlex is featured as a Representative Vendor. Locus holds Leader designation in the QKS SPARK Matrix for Transportation Management Systems 2025 and the #1 position for Route Planning in G2’s 2026 Best Software Awards.

In October 2025, Ingka Investments, the investment arm of Ingka Group, the world’s largest IKEA retailer, acquired Locus. Locus continues to operate independently.

Grocery substitution management breaks at short promises for one reason: customer response time is a human constant while every other quantity in the operation compresses with the delivery window. Reaching a 50% answer rate needs about 12 minutes of decision time, which a 10 to 15 minute promise cannot supply, so express operations that keep asking end up notifying four in five customers of a problem they cannot act on. The fix is to move the decision to order time through preference coverage, apply attribute-matched defaults to everything else, and let the system ask only when the window genuinely supports an answer. Locus makes that branch explicit inside the same layer that governs the promise. [Request a Locus grocery fulfillment assessment](https://locus.sh/schedule-demo/)
 to run this model against your own pick times and promise windows.

## Frequently Asked Questions

**What is grocery substitution management?**

Grocery substitution management is the process of resolving an order line that cannot be filled as ordered, covering how the replacement is selected, whether the customer is consulted, and how the charge is adjusted. It matters because grocery is the only major retail category where what the customer bought is routinely not what arrives, at a rate McKinsey puts at 10% to 15% of pick-from-store orders.

**Why do substitution approval requests fail in quick commerce?**

Because the decision window is shorter than human response time. A 12-minute promise leaves roughly two and a half minutes between the picker finding the gap and the rider departing, and in this model only 18.9% of customers respond that fast. The request is sent, recorded and largely unanswered.

**How long a window do you need for customers to actually respond?**

About 12 minutes to reach a 50% answer rate, and roughly 38 minutes to approach the 62% ceiling. Both figures exceed the total promise in quick commerce, which is why the mechanism does not transfer from scheduled-slot e-grocery.

**Does giving customers more time to respond help?**

Only below about 30 minutes. Moving the window from 2.5 to 5 minutes adds 14.4 points of answer rate, while moving from 45 minutes to three hours adds 1.4 points. Above half an hour the constraint is whether the customer engages at all, not how long they have.

**Is it better to substitute silently than to ask and get no answer?**

When the window is genuinely too short, yes. At a 2.5-minute window, 81% of customers receive a notification about a problem alongside a control that expires before they can use it. An attribute-matched substitution applied without a question carries 75% acceptance in published research and avoids handing the customer a decision they cannot make.

**What actually reduces substitution rejections in fast grocery?**

Preference coverage captured at order time. Presets covering 60% of order lines at a two-and-a-half-minute window produce lower door rejection than a three-hour real-time ask, and unlike the ask, coverage is unaffected by how short the promise gets.

MEET THE AUTHOR

Ishan Bhattacharya

Lead - Content

Ishan, a knowledge navigator at heart, has more than a decade crafting content strategies for B2B tech, with a strong focus on logistics SaaS. He blends AI with human creativity to turn complex ideas into compelling narratives.

### Related Tags:

[https://locus.sh/blogs/recovery-capacity-peak-season-reschedule/](https://locus.sh/blogs/recovery-capacity-peak-season-reschedule/)
#### [General](https://locus.sh/blogs/category/general/)

## [Recovery Capacity Planning in 2026: Why the Reschedule Offer Runs Out Three Weeks Before Christmas](https://locus.sh/blogs/recovery-capacity-peak-season-reschedule/)

[Aseem Sinha](https://locus.sh/blogs/author/aseem_locus/)

Sep 22, 2026

Every exception playbook assumes a second chance is free. Recovery capacity is finite, and it runs out at 92% utilization, not 100%.

[Read more](https://locus.sh/blogs/recovery-capacity-peak-season-reschedule/)

## Grocery Substitution Management in 2026: Why the Approval Window Closes Before the Customer Answers

- Share
- [Print](javascript:window.print())
- [Download](#)
- [Schedule a Demo](https://locus.sh/schedule-demo/)

### Is your team spending more time on fixing logistics plan than running the operation?

- Agentic transportation management from order intake to freight settlement
- Route optimization built on 250+ real-world constraints
- AI-driven dispatch with automatic execution handling

20%Cost Reduction

66%Faster Planning Cycles

[Schedule a demo](/schedule-demo/)

Insights Worth Your Time

#### [General](https://locus.sh/blogs/category/general/)

## [Locus 2026 UK Consumer Survey: Why Returns Visibility is Now the Conversion Engine for AI-Driven Shopping in UK Retail](https://locus.sh/blogs/returns-visibility-conversion-engine-ai-shopping-uk-retail-locus-q2-2026-consumer-survey/)

[Aseem Sinha](https://locus.sh/blogs/author/aseem_locus/)

May 29, 2026

#### [General](https://locus.sh/blogs/category/general/)

## [Locus 2026 US Consumer Survey: Generative AI isn’t Just Changing How Consumers Shop, it’s Breaking the Demand Patterns US Retail Was Built On](https://locus.sh/blogs/generative-ai-shopping-effect-retail-fulfillment-operations-locus-q2-2026-consumer-survey/)

[Ishan Bhattacharya](https://locus.sh/blogs/author/ishan_locus/)

May 29, 2026

#### [General](https://locus.sh/blogs/category/general/)

## [Embedded vs Bolted-On AI: The Architecture Question European Logistics Buyers Are Asking](https://locus.sh/blogs/embedded-vs-bolted-on-ai-european-logistics-platform-architecture-business-benefits/)

[Aseem Sinha](https://locus.sh/blogs/author/aseem_locus/)

May 21, 2026

#### [General](https://locus.sh/blogs/category/general/)

## [Hybrid Fleet Management: How Owned, 3PL, Gig, ICE, and EV Capacity Actually Operate at Most Enterprises](https://locus.sh/blogs/three-workforce-fleet-reality-owned-3pl-gig-drivers/)

[Aseem Sinha](https://locus.sh/blogs/author/aseem_locus/)

May 7, 2026

#### [General](https://locus.sh/blogs/category/general/)

## [US Returns Hit $850 Billion in 2025: Why US Retailers Are Restructuring Reverse Logistics in 2026](https://locus.sh/blogs/850-billion-us-returns-ai-routing-reverse-logistics-2026/)

[Ishan Bhattacharya](https://locus.sh/blogs/author/ishan_locus/)

May 7, 2026