AI Low Stock Alerts

Retail AI

Detecting a low shelf is only useful if the right person hears about it in time to act. Rubicon A-Eye turns shelf and stock conditions into threshold-based alerts that reach the right channel, escalate when ignored, and close the loop with a logged action — the notification and escalation layer that sits on top of detection and counting.

The business challenge

Most retailers already have some way of seeing that stock is low — a shelf walk, a report, a count. The harder problem, and the one that rarely gets solved well, is what happens after that. Who gets told? Through which channel? How urgently? And what happens if nobody acts? Detection without a reliable notification and escalation path just produces information that sits unused until someone happens to look at it.

In practice, low-stock notification in most stores is informal and inconsistent. One supervisor checks a report at the start of the shift; another does not. A WhatsApp message gets sent to a group chat and scrolls past unseen. A printed list is left on a desk. None of this is a process — it is a collection of habits that work until they do not, and the gaps in coverage are exactly when stockouts happen.

For operations managers, the result is a familiar and frustrating pattern:

• Low-stock information that exists somewhere but does not reliably reach the person who can act
• No defined threshold for what counts as “low” — it varies by shelf, shift and judgement
• Alerts sent once, with no follow-up if nobody actions them
• No record of how long stock sat low before someone responded
• Different urgency treated the same way, so a slow mover and a fast mover get identical notifications
• Accountability gaps when a stockout happens and nobody can say who should have acted

The hidden cost of an unmanaged alert

An alert that is sent but not acted on is functionally the same as no alert at all — except it creates a false sense that the problem is being handled. Retailers that treat detection as the finish line discover that the real failure point is almost always downstream: the notification reached the wrong person, reached the right person at the wrong time, or reached someone who had no clear accountability for acting on it. Fixing detection without fixing the notification and escalation path solves only half the problem, and often the easier half.

How Rubicon A-Eye solves the problem

A-Eye’s low stock alerting is a rules and escalation layer that sits on top of shelf detection and stock counting. It does not just observe that a facing or zone has crossed a threshold — it decides what to do about that fact: who to notify, through which channel, how urgently, and what happens if the first notification goes unanswered.

Thresholds are configured per shelf, zone or product line to match how that line actually behaves. A slow-moving line might alert at a relaxed threshold with a routine notification; a high-velocity promoted line might alert earlier and escalate faster, because the cost of missing it is higher. This is deliberately distinct from raw stock counting: counting tells you the number, alerting decides whether that number, in that context, requires someone’s attention right now.

Notifications route to the channel that fits your operation — staff handheld devices, in-store dashboards, supervisor messaging, or task queues inside Odoo. If an alert is not acknowledged or actioned within a defined window, it escalates automatically to a supervisor or an alternate channel, so a missed notification does not silently disappear. Every alert, acknowledgement and resolution is logged, which gives managers a record of response time and accountability that a one-off message in a group chat never could.

Key capabilities

What you can measure

• Average time from alert raised to acknowledged, by staff member and shift
• Average time from acknowledged to resolved, by shelf and category
• Escalation rate — how often alerts go unanswered long enough to escalate
• Alert volume by threshold tier, store and time period
• Stockouts that occurred despite an alert being raised, and why
• Channel effectiveness — which notification routes get the fastest response

Industry applications

Threshold-based alerting and escalation is valuable wherever low-stock detection exists but the notification path is currently informal:

Multi-store chains and franchise networks benefit most directly, because alerting and escalation rules can be standardised across every location instead of varying by individual store culture. High-velocity grocery and convenience retail benefit from priority-tiered alerting on fast movers where minutes of delay matter. Larger format stores with bigger teams benefit from clear routing so an alert reaches whichever staff member is actually positioned to act, not a general broadcast.

Business benefits

• Low-stock conditions reliably reach a specific accountable person, not a general channel that may go unread
• Faster average response time through defined escalation rather than informal follow-up
• Clear accountability and a logged record when a stockout occurs despite an alert
• Consistent alerting standards across every store in a multi-site network
• Differentiated urgency so high-velocity lines get faster treatment than slow movers
• Reduced reliance on staff remembering to check a report or message thread

How it works: the operational workflow

Example scenarios

Integration with your systems

Alerting earns its value by reaching the channels your team actually works in and feeding the systems that drive action. As a certified Odoo partner, Rubicon builds this connection into the solution directly.

Implementation and rollout

Alerting rules are configured deliberately rather than switched on all at once. A typical rollout starts with a short workshop to define thresholds, priority tiers and escalation windows for your key categories, followed by configuration of the notification channels your team actually uses — handheld devices, supervisor messaging, dashboards or Odoo task queues. During a validation period, alerts run alongside your existing process so the team can confirm response times and escalation behaviour match expectations before the new workflow fully replaces informal notification habits.

Most retailers begin with thresholds for their highest-velocity or highest-risk categories, where the cost of a missed alert is greatest, and expand coverage to the rest of the range once the escalation workflow is proven. Because the underlying detection already exists through shelf monitoring or stock counting, adding the alerting and escalation layer is typically a configuration exercise rather than a lengthy technical build, and can be live within a short timeframe.

Why structured escalation outperforms informal notification

The default alternative to a structured alerting system is the informal mix most stores already run on — a report, a message, a habit. The problem is not that these channels never work; it is that they work inconsistently, and the inconsistency is invisible until a stockout exposes it. There is no record of whether the message was seen, no defined consequence if it was not, and no way to know in advance which shelves are covered by someone’s diligence and which are not.

A rules-based alerting and escalation layer replaces inconsistency with a defined process: a threshold is crossed, a specific channel is notified, a window is set for acknowledgement, and an escalation path exists if that window passes. This does not require perfect staff behaviour to work, because the system itself carries the responsibility for making sure a low-stock condition is not silently dropped. For multi-store operators, it also means every location runs the same accountability standard, rather than relying on the strongest individual store managers to compensate for weaker processes elsewhere.

Frequently asked questions