Out-of-stocks cost the retail industry an estimated $1 trillion in lost sales every year. The global average out-of-stocks rate across grocery retail sits at 8.3%—which means for every 13 items on a shopper's list, one isn't there.
The first thing GPC brands do when those numbers come up is to look at the supply chain, but the supply chain is only part of the problem. Research shows that between 25% and 60% of out-of-stocks events happen when product is already in the store—sitting in the backroom while the shelf sits empty.
That gap between what's in the building and what's on the shelf is where most CPG brands have the least visibility. And it's the gap that a field rep can actually close during a store visit—if they find out about it while they're still there.
An out-of-stock occurs when a product a shopper wants to buy is not available on the shelf at the moment they arrive. It's the gap between what a shopper expects to find and what's actually there—an empty shelf position, a missing SKU, or a facing that has dropped to zero visible units.
Most CPG teams treat that gap as a single problem, but in practice, "out-of-stock" describes three distinct situations with different causes, different owners, and different fixes. Treating them as one leads to investing in solutions that address the wrong layer.
The product doesn't exist in the supply chain at the store level. Nothing was ordered, the delivery failed, or the distribution center ran out. No amount of in-store effort fixes this. It requires a supply chain intervention—adjusting the replenishment order, expediting a delivery, or shifting inventory from another location.
Product has been delivered to the store but inventory records are inaccurate. The system shows units available, but those units may not actually be there—or they're not where the system thinks they are. This is often caused by theft that wasn't logged, damage that wasn't recorded, or counting errors during receiving. The product may be physically absent, or it may exist in the store but be miscounted.
Product is physically in the store—in the backroom, in a cage, on a pallet waiting to be worked—but it isn't on the shelf. The inventory system shows it as in stock. A shopper looking at that shelf sees an empty position. This is a pure execution failure, and it's the type a field rep can fix during the current visit.
This is also the most common type. Industry research puts 25–40% of all out-of-stocks events in this category. Some studies of specific retail formats put it as high as 60%.
The reason this three-way split matters is simple: the right fix depends entirely on which type you're dealing with. A forecasting improvement won't help a backroom-to-shelf failure. An in-store execution program won't help a distribution gap. Before investing in any out-of-stocks detection or prevention solution, a category manager or field execution director needs to know which layer is generating the majority of their failures.
A void occurs when a SKU is authorized for a store but has no shelf tag and no inventory—the product was never properly set up at that location. Unlike an out-of-stocks, which means the product ran out, a void means it was never introduced at that store in the first place. The fix is different: a void requires a new store setup, not a restock. Field reps who don't distinguish between the two can spend time looking for backroom stock that was never ordered.
The question sounds basic, but it's the most important diagnostic in out-of-stocks management. A product that's in the backroom is not an out-of-stock from the supply chain's perspective. It's an execution failure—specifically, a restock gap.
The restock gap is the time and process between stock arriving at the store and that stock landing on the shelf.
In a busy grocery store with a small team managing hundreds of SKUs, products can sit in the backroom for hours or even an entire shift before a store associate gets to them. Meanwhile, a shopper arrives, sees an empty shelf, and switches brands.
Understanding where the gap comes from determines which fix applies.
Distribution and store out-of-stocks trace back to supply chain failures that a field rep can't resolve during a visit. Shelf out-of-stocks trace back to in-store execution failures that a field rep can resolve—if they know about them while they're still in the store. Here's what drives each:
That last four causes are all fixable by a field rep during a standard store visit. The challenge is detecting them before the rep leaves the building.
Most out-of-stocks programs don't catch these until a rep does a manual visual check, which by definition only happens during the visit. But there are detection methods that surface them earlier—and that's what the next sections cover.
The commercial damage from an out-of-stock depends on whether a shopper has a strong reason to seek out that specific product—or whether they'll just reach for whatever is next on the shelf.
When a shopper can't find an established brand with strong category presence, they typically substitute within the category. They reach for the brand sitting next to where yours should be. The sale is lost, but the shopper doesn't leave the aisle. For a category leader, out-of-stocks is expensive but it mostly benefits the brand two slots to the right.
For a challenger brand or a product still building distribution and trial, there's no ingrained habit driving the shopper to look for it. If it's not visible on the shelf at the moment they're in the aisle, the consideration never happens. The shopper doesn't substitute—they just buy whatever is familiar. That's a lost first impression that may not come back.
|
"At a previous CPG brand I worked at, a third of our first-time consumers only engaged with the brand because they saw it on a display in a store. Off-shelf execution was directly driving trial—not advertising, not pricing. If the display wasn't there, that first interaction never happened." — Steven Bussiere, VP Customer Success, Vision Group |
This distinction is key when setting out-of-stocks detection priorities.
For a category leader, an out-of-stock on a hero SKU is a revenue problem measured in dollars per day. For a brand still building awareness, an out-of-stock during a promotional push means the marketing spend that drove the shopper to the aisle generated zero return—because the shelf wasn't set when they arrived.
Either way, the commercial cost is real. The size and nature of it varies enough that which SKUs and which stores you prioritize for detection should reflect where your brand sits in the category.
This is the phantom inventory question—and it's one of the most frustrating gaps in out-of-stocks management because everything looks fine from the back office while shoppers are encountering empty shelves.
Here's what's happening: your ERP or inventory management system tracks stock movements—deliveries received, units sold at POS, adjustments made. But it doesn't have eyes on the physical shelf. It records what happened, not what's there right now.
Phantom inventory builds up over time through a handful of specific failure points:
The practical result: your system shows a healthy in-stock percentage. Your field rep visits the store and finds an empty shelf. The system thinks there's nothing to fix. The shopper found out before anyone else did.
Phantom inventory is the primary reason why inventory-based detection consistently overstates actual shelf availability. The only way to know what's on the shelf is to look at the shelf—either through a field rep's manual check or through image recognition that reads the shelf directly during the visit.
That leads to the next question: which detection methods actually see the shelf, and which ones are just reading the system's version of events?
Not all detection methods are looking at the same thing.
Some read the inventory system. Some read POS data. Some physically look at the shelf. The method you use determines which type of out-of-stocks you'll catch—and which ones will slip through.
Inventory system analysis uses ERP or WMS data to infer whether a product is available. Because every delivery, sale, and adjustment is logged in the system, a category manager can check the reported stock level for any SKU across every store from a single dashboard.
What it misses: everything related to phantom inventory. The system shows available units whether they're on the shelf or in the backroom. It can't distinguish between "in stock" and "on shelf." For shelf out-of-stocks caused by execution failures, inventory analysis produces false confidence.
POS sales velocity analysis uses point-of-sale transaction data to infer shelf availability. Instead of checking what the inventory system reports, it checks what's actually selling—and uses an unexpected drop to zero as a signal that something has gone wrong at the shelf.
When a high-velocity SKU shows zero sales for 24–48 consecutive hours while the inventory system shows units available, that pattern is a high-confidence signal of a shelf out-of-stocks. The system thinks the product is there. Nothing is selling. The most likely explanation: the product is in the backroom, not on the shelf.
This is the earliest warning signal available to a CPG brand without any additional technology. A category manager who sets up an automated alert for any top-20 SKU showing zero sales for 48+ hours in a store where it normally moves 20–30 units per day gets a daily list of probable shelf out-of-stocks events—before a rep visit is even scheduled.
What it misses: slow movers. A SKU doing 5 units per week can be out of stock for two weeks before the sales signal disappears. It also can't tell you why the product is gone—supply chain or execution.
A manual shelf audit means a field rep physically walks the shelf and checks each position during a store visit. It's the most direct method available to CPG brands auditing stores they don't own—and the one that's been the industry standard for decades.
What it misses: it's a snapshot. A rep visiting at 9am confirms availability at 9am. By 2pm, the shelf can be empty with nobody catching it until the next visit. Manual audits also can't reliably detect a facing that's been reduced to zero units pushed to the back—the shelf looks occupied, but the SKU has effectively disappeared from shopper view.
RFID (Radio Frequency Identification) uses electronic tags attached to individual product units or cases. As tagged units move through the supply chain—received at the DC, delivered to the store, scanned at POS—the system logs each movement. When a unit is in the store but hasn't moved to the shelf, RFID can detect that it's somewhere in the building.
What it misses: RFID answers "is this product somewhere in this store?"—not "is this specific SKU in this specific shelf position right now?" It can't tell you that a product is pushed to the back of the shelf with no front-facing unit. It can't distinguish between a correctly positioned product and one misplaced in the wrong aisle. And it doesn't catch execution failures like a facing count that dropped below minimum.
RFID is useful for tracking movement through the supply chain and backroom. For shelf-level out-of-stocks detection—the type a CPG brand needs to catch execution failures—it doesn't see the right layer.
AI-powered image recognition works by having a field rep photograph a shelf section during a normal store visit. A computer vision model reads the image, identifies every visible SKU, counts facings per position, detects empty shelf slots, and compares the full picture against the planogram for that specific store. Within 90 seconds, the rep receives a gap list on their phone—before they leave the aisle.
Image recognition is the only method that catches both fully empty positions and near-out-of-stocks situations—SKUs that still have 1–2 units but have dropped below the minimum facing threshold. A product with one unit pushed to the back looks empty to a shopper. Manual audits often miss it. Image recognition catches it because it counts facings, not just presence.
It also distinguishes between three different shelf situations that look identical on a manual check:
Out-of-stocks: The correct product has sold through. Fix: restock from backroom or escalate to store manager.
Planogram non-compliance: The wrong product is in that position. The correct SKU is elsewhere on the shelf. Fix: reposition the products to match the planogram.
Void: The SKU was never set up at this store. No stock was ordered. Fix: initiate a store-level setup, not a restock.
A manual audit catches that a shelf position is empty. Image recognition technology tells you which of those three situations you're actually looking at—and therefore which fix to apply.
The zero-sales anomaly is a shelf detection signal derived from POS data: when a product that normally sells every day suddenly shows zero sales for two or more consecutive days—while the inventory system still shows units available—the most likely explanation is that the product is in the backroom rather than on the shelf.
Every CPG brand with access to store-level POS data has this signal available. Most don't have automated alerts configured to use it.
Here's how it works in practice: take any SKU in your top 20% by sales velocity. In a grocery store where that SKU normally sells 25 units per day, two consecutive days of zero sales is a shelf signal. Either the product isn't on the shelf, or it's been positioned somewhere a shopper can't find it.
When zero sales coincide with the inventory system showing units available, the probability of a shelf out-of-stocks exceeds 90%. The system thinks the product is there, but nothing is selling. The most likely explanation is that the backroom stock hasn't been worked to the shelf—the restock gap in action.
You don't need new technology to use this signal. You need a POS data feed and a simple alert rule:
The result is a prioritized list of probable out-of-stocks events delivered to a field manager before rep routes are even set. Instead of reps discovering out-of-stocks events by chance during a visual check, they arrive at flagged stores already knowing which SKUs to verify.
The zero-sales anomaly only works for fast movers. A SKU doing 5–8 units per week in a store can be out of stock for two full weeks before the absence shows clearly in POS data. Slow movers, new product launches with no velocity baseline, and seasonal items all have low-signal periods where this approach fails.
That's the gap image recognition fills. It reads the shelf directly rather than inferring from sales patterns—so it catches slow-mover out-of-stocks events, near-out-of-stocks situations, and planogram compliance failures that the zero-sales signal never sees.
Used together—POS anomaly detection for fast movers and image recognition for shelf-level verification—a CPG brand gets coverage across the full velocity range of their catalog. That combination is what changes out-of-stocks management from reactive to proactive.
Most out-of-stocks programs are reactive by design: a rep visits a store, finds a gap, reports it, and a correction happens—either during that visit if the backroom has stock, or on a follow-up trip if it doesn't. The average correction lag across a typical CPG field operation runs 4–7 days from the time an out-of-stocks opens to the time it gets fixed.
On a high-velocity SKU in a high-traffic grocery account, that's significant. A product doing $50,000 in weekly sales at a single store has an out-of-stocks cost of roughly $7,000 per day. A 5-day correction lag is a $35,000 revenue event in one store. Across 500 stores with the same SKU-level gap, the math changes the business case for doing this faster.
The last meter refers to the physical distance between a store's backroom and the shelf position where a product belongs. It's the final step in the replenishment process—and the one most likely to fail without specific direction.
Even when a store has backroom stock, getting it to the right shelf position requires a store associate to identify the correct aisle, section, and slot—and then determine whether they're looking at an out-of-stocks, a planogram compliance failure, or a void. In a large-format grocery store, that process without guidance takes significantly longer and introduces diagnosis errors.
With image recognition, the field rep's phone becomes the direction system. The rep photographs the shelf, receives a prioritized gap list with exact SKU locations, and tells the store associate precisely which product to pull and where to put it. The associate walks straight to the backroom, pulls the right SKU, and fills the right position. No full-aisle walk or diagnosis at the shelf needed. The last meter gets covered in minutes rather than being left to chance.
|
Detection method |
Typical correction lag |
|
Inventory system / ERP |
Days to weeks (if ever—phantom inventory creates permanent blind spots) |
|
POS anomaly (zero-sales alert) |
12–48 hours to route to a rep, then next visit to verify and fix |
|
Manual audit—next scheduled visit |
4–7 days average from out-of-stocks opening to correction |
|
Manual audit—same visit |
Minutes, but only catches what the rep notices at that moment |
|
AI image recognition—same visit |
Under 90 seconds to detection, correction before rep leaves the aisle |
The commercial difference between the top and bottom of that table is the correction window. Every day a shelf out-of-stocks persists on a high-velocity SKU is revenue that doesn't recover. The fix doesn't restore the shopper who switched brands yesterday.
When image recognition reads a shelf section, it reads the whole category—not just your brand's positions. That creates a detection capability most CPG teams overlook: knowing when a competitor goes out-of-stocks.
An empty position in a competitor's allocated shelf space is a live opportunity. A category manager who gets that signal during a field visit can direct the rep to request a temporary secondary display, negotiate expanded facings from the store manager, or at minimum document the competitive gap to support the next buyer conversation. The intelligence is free—it's a byproduct of every shelf photo the rep takes.
Without image recognition, this information surfaces in syndicated data four weeks later, after the window has closed. With it, a field execution director gets the signal the same day the gap opens.
Most image recognition platforms detect out-of-stocks events and route them to a dashboard. Store360 is built to close the gap during the visit—which means the rep gets the finding before they leave the aisle, not after they've left the store.
Three things that make a difference for out-of-stocks detection specifically:
Store360 counts facings per SKU and flags positions that drop below the minimum defined in the planogram—not just positions that are completely empty. A product with one unit pushed to the back shows as a near-out-of-stocks before it becomes a full gap. That gives the rep a window to restock before the shopper encounter ever happens.
Most IR tools require an official planogram to benchmark against. No planogram means no out-of-stocks data for that store—a structural blind spot across networks where planogram coverage is incomplete or outdated. Store360 benchmarks against category norms and competitor positions even without a planogram file, so every store generates out-of-stocks detection data regardless of coverage.
Store360 doesn't just list every gap. It ranks them by revenue velocity, so the rep fixes the highest-impact out-of-stocks first. A missing facing on the top-selling SKU in the category gets addressed before a minor facing deviation on a slow mover. Results are on the rep's phone within 90 seconds of the shelf photo—before they move to the next aisle.
Proof point:
L'Oréal deployed Store360 at Walmart locations where out-of-stocks were a persistent problem. Before Store360, their reps were working from audit data that was 2–4 weeks old. With Store360, reps saw exactly which SKUs were trending toward out-of-stocks during each visit, showed store managers the photographic evidence of the gap, and drove replenishment action on the spot.
Result: $50,000+ in replenishment orders across 10 Walmart stores in two weeks.
"Inventory levels are going back up, sales are going back up on these out-of-stock items, and it's really moving the needle."
— Barbara Kline, Head of Retail Execution, L'Oréal
One world-leading beverage company used Store360 to move from delayed reports to live shelf feedback and cut out-of-stocks across thousands of stores in weeks.
Store360 is live in 55+ countries, runs on the device a field rep already carries, and most clients go live in under 30 days—no new hardware, no retailer permission required.
→ Book a 20-minute walkthrough of Vision Group's Store360 here: shelf photo to out-of-stocks detection to correction task, in under 90 seconds.
An out-of-stock occurs when a product a shopper wants to buy is not available on the shelf at the moment they arrive. This includes fully empty shelf positions and situations where a product is technically present but has no visible front-facing units. The global average out-of-stock rate in grocery retail is approximately 8.3%.
Store out-of-stocks means the product isn't in the building at all—nothing was delivered, or the store has genuinely run out of inventory. Shelf out-of-stocks means the product is physically in the store—usually in the backroom—but hasn't been moved to the shelf. Shelf out-of-stocks accounts for 25–60% of all out-of-stocks events depending on the retail format. It's fixable during a field visit. Store out-of-stocks requires a supply chain response.
The restock gap is the time and process between stock arriving at the store and that stock landing on the shelf. In a busy retail environment, products can sit in the backroom for hours or a full shift before a store associate works the pallet. During that time, the inventory system shows the product as available while the shelf shows it as empty. The restock gap is one of the largest contributors to shelf out-of-stocks events.
This is the phantom inventory problem. Your ERP tracks stock movements—deliveries received, units sold at POS, manual adjustments—but it doesn't physically verify the shelf. Units in the backroom, theft not logged, damage not recorded, and receiving errors all create discrepancies between what the system shows and what's actually on the shelf. The only way to close this gap is shelf-level verification through a physical audit or image recognition.
Causes split across two layers. Supply chain causes: inaccurate demand forecasting, late deliveries, and promotional volume spikes that weren't modeled. In-store execution causes: backroom stock not moved to the shelf, reset errors that left products in the wrong position, facing counts dropped below minimum, and missing shelf tags. The execution causes are fixable during a field visit. The supply chain causes require upstream intervention.
Yes. When a shopper can't find an established brand, they typically substitute within the category—the sale transfers to a competitor but the shopper doesn't leave. For a challenger brand or product still building trial, there's no habit driving the shopper to look for it. If it's not visible on the shelf when they arrive, the consideration doesn't happen. The shopper doesn't substitute—they just buy what's familiar. For brands building distribution and awareness, an out-of-stocks during a promotional push can mean the marketing spend that drove the shopper to the aisle generated zero return.
The zero-sales anomaly is when a high-velocity SKU shows zero POS sales for 24–48 consecutive hours in a store where it normally sells consistently. When this coincides with the inventory system showing units available, the probability of a shelf out-of-stocks exceeds 90%. The product is likely in the backroom and hasn't been worked to the shelf. CPG brands can set up automated alerts for this pattern using existing POS data—no additional technology required.
RFID answers 'is this product somewhere in this store?'—not 'is this specific SKU in this specific shelf position right now?' RFID can confirm that a unit is in the building, but it can't tell you whether it's on the shelf, pushed to the back, misplaced in a different aisle, or positioned incorrectly against the planogram. For shelf-level out-of-stocks detection—the type that catches execution failures—RFID doesn't see the right layer. Image recognition reads the actual shelf position.
Yes. Image recognition counts facings per SKU and compares that count against the minimum defined in the planogram. When a SKU drops to 1–2 remaining facings—units pushed to the back with no front-facing product visible to a shopper—the system flags it as a near-out-of-stocks. This gives a field rep the opportunity to restock before the position goes fully empty, before a shopper encounters the gap.
An empty shelf position can mean three different things: the product sold out, the wrong product is in that position (planogram non-compliance), or the product was never set up at that store (void). Manual audits catch that a position is empty—they typically can't distinguish which situation you're dealing with. Image recognition identifies which SKU is actually in each position, detects what's missing, and tells the rep whether they need to restock, reposition existing products, or escalate a void to the category team.
Image recognition platforms like Store360 rank detected gaps by commercial priority—revenue velocity, category importance, and deviation severity. The rep's phone shows the highest-impact out-of-stock at the top of the list, so a missing facing on the top-selling SKU in the category gets addressed before a minor deviation on a slow mover. The rep doesn't have to decide which gap matters most—the system has already ranked them.
Without real-time shelf detection, the average correction lag runs 4–7 days from when an out-of-stocks opens to when it gets fixed. That assumes a rep visits within that window, notices the gap, and either fixes it during the visit or triggers a follow-up. With image recognition-based detection during the visit, the correction window closes to the time it takes the rep to retrieve backroom stock and fill the position—usually minutes.
The global average out-of-stocks rate in grocery retail is approximately 8.3%, according to research by Gruen and Corsten. That figure has remained remarkably stable for decades. Fast-moving categories and promotional periods consistently show higher rates. During major promotional events, out-of-stocks rates in affected categories can spike 20–40 percentage points above normal.
POS anomaly detection flags probable out-of-stocks events 12–48 hours after they begin. Manual audits catch them at the moment of the visit—but only what the rep notices. Inventory system alerts miss them entirely when phantom inventory makes the shelf look fine from the back office.
Image recognition is the method that closes the correction loop during the visit. The rep photographs the shelf section, the system returns an out-of-stocks score and a prioritized gap list in 90 seconds, and the correction happens before the rep leaves the aisle. For high-velocity SKUs, that timing difference is the gap between a recovered sale and a lost one.
The zero-sales anomaly gives you a head start. Image recognition gives you the confirmation and the fix. Used together, they turn out-of-stocks management from a reactive program into a proactive one.
→ Book a walkthrough of Vision Group's Store360 here.