Changing Retail Execution with AI Vision Models

Written by Jane Eriksen | ConnectGroup

The hardest part of retail has never been deciding what needs to happen, it’s knowing what actually did.

Head office teams plan promotions, lock in listings, allocate shelf space and map category layouts, decisions that need to hold across varied sized stores, shifting regional demand and constant operational pressure.

But once plans leave head office, there is a lot of guesswork with regards to its execution. Stores are busy, staff are stretched, day to day priorities shift and no one really knows what’s happening on shelf at scale from a retailer and manufacturers perspective.

Most of the industry has worked around this and done their best to get the visibility they need to make better decisions. Reps are sent into stores, audits are completed and reports from merchandising teams are sent through to stakeholders.

But these are all snapshots of a moment in time. By the time the report lands on a desk, seasons have changed, promotions are switched up and there is never really a consistent view of the full picture. Each part of the value chain has accepted their own version of the truth, until now.

The impact of AI technology in the retail industry is impressive but also very practical. Issues are picked up earlier, teams focus on stores that need attention, follow-ups are based on evidence, not assumption and performance is easier to manage across regions. These gaps that have always defined retail becomes harder to ignore.

Through QikScan AI, technology developed by ConnectGroup, on-shelf image recognition is being used to create a consistent, real-time view of execution across large store networks. Most importantly, the cost of the technology is removing the barrier to entry for companies who aren’t willing to pay a dollar per image.

Instead of relying on periodic audits, store-level data is now captured continuously through image recognition. Share of shelf, product availability and out of stocks are recorded and processed in near real time. Retail execution is becoming measurable in a way it never has been before, and that changes how decisions get made at every level of the value chain.

The shift is not just technological. It changes how teams operate.

ConnectGroup rolled out QikScanAI with Nestlé across 1,500 retail stores, spanning five distinct business units. Each unit had its own priorities, market shares, product ranges, competitors and execution standards, which historically made consistent national visibility difficult to achieve.

What changed was not just visibility, but how quickly teams could act on it.

Issues that would previously sit unnoticed between reporting cycles were now flagged within days. Store visits became more targeted. Conversations between manufacturers and retailers shifted from debate to evidence.

At scale, this created a single reference point for execution. Instead of fragmented reports and subjective assessments, teams could compare performance across regions, formats and categories using the same lens.

This does not eliminate complexity. Different stores still behave differently. Regional dynamics still matter. But it removes ambiguity. Gaps that were once assumed can now be measured, prioritised and addressed.

That has practical consequences.

Execution is no longer a periodic check. It becomes part of the daily operating rhythm. Teams know where to focus. Follow-ups are based on what is actually happening, not what is believed to be happening.

Over time, this starts to influence how performance is defined. Conversations around compliance, ranging and promotions become more grounded and focused. Expectations are clearer because they are based on consistent data, not interpretation.

None of this requires a fundamental change in retail strategy. The same plans still need to be made. The same pressures still exist in store.

What changes is the feedback loop.

When execution is visible at scale and in near real time, decisions are no longer delayed by incomplete information. The gap between plan and reality does not disappear, but it becomes easier to manage.

And that, more than the technology itself, is where the real shift is happening.