Grocery analytics at scale: Bueno Analytics brings lessons from 1,300 Woolworths stores to US retailers

“No one does what Woolworths does,” said Hugh Amoyal, CEO of Bueno Analytics, when describing the Australian grocery giant’s data-driven facilities program. As someone who’s spent years analyzing smart building technologies, I was intrigued—and admittedly a bit skeptical. In theory, every supermarket operator wants to cut energy costs and prevent equipment failures, but Woolworths Group (nicknamed “Woolies”) has actually done it at an unprecedented scale. 

They’ve connected 1,300 supermarkets, 178,000 devices and 2 million data points, streaming 5-minute interval data into one platform. The result? Analytics that go far beyond the typical alarm systems most grocers use today, unlocking proactive maintenance (yes, actually proactive) and massive savings. 

The Woolworths Energy Management Centre (EMC) used the Bueno platform to prioritise energy and cost savings measures, including data-driven maintenance, capital projects, and HVAC optimisation. The result was a 27% reduction in carbon emissions, a 37% drop in refrigerant leaks, and $80 million per year in energy savings—all at a cost of only a few thousand dollars per store per year. This isn’t a pitch for Bueno; it’s the real-world story facility managers in grocery and retail need to know about, especially given the challenges hitting U.S. supermarkets right now.

Let’s set the scene. Grocery facility teams are facing a perfect storm: rising refrigerant costs, stricter AIM Act regulations, and a shrinking maintenance workforce. As we covered in our earlier piece, this has many retailers urgently rethinking how they manage refrigeration. Woolworths’ approach—using software and analytics to predict and prevent issues—offers a compelling path forward.

What makes Woolworths unique? In Hugh Amoyal’s words, it’s three things: 

• Scale: Few grocers can bring all their data into one platform. Woolworths’ supermarkets consume about 1% of all electricity in Australia, so optimizing their portfolio has an outsized impact. 
• Moving beyond alarms: Most grocers still rely on BAS and refrigeration alarms and think that means they’re doing “analytics.” Woolworths realized early on that alarm-based energy management wasn’t enough; alarms only tell you something’s broken after the fact and often produce so many alerts that important issues get overlooked. 
• Failure prediction: Woolworths’ program uses cloud-based fault detection and diagnostics (FDD) to find inefficiencies and problems before they trigger alarms, or even before performance degrades. In Hugh’s view, this is “the first example of true predictive analytics” in our industry—and it’s hard to disagree when you see the results. With nearly a decade of data from thousands of units, the system can now tell when a refrigeration asset is likely to fail days in advance, enabling planned fixes in lieu of emergency repairs. 

It’s time to tell the story of how Woolworths and Bueno accomplished this, and what lessons U.S. grocery facility managers can take from it. The journey unfolded in phases (mirroring an “energy management hierarchy of needs” approach). Here’s a walkthrough of those phases, followed by an analysis of why it worked and practical takeaways for retailers in North America.

Phase 1: Proof of Concept with Meter Analytics (2015–2016)

Woolworths’ first step was simply collecting data and visualizing energy patterns. In 2015–2016, they kicked off a project with Bueno to pull together energy consumption data in dashboards. At this stage, it was basic—mainly main meter data and some manual data dumps—but it revealed enough waste and anomalies to get leadership’s attention. 

The facility team realized they needed more granular data (device-level, not just whole-store meter) to pinpoint root causes. This initial project built the business case for a larger investment. It demonstrated that data-driven insights could lead to savings, justifying a significant capital expenditure to go further. 

Phase 2: Upgrading Infrastructure for Scale (2017-2019)

With a compelling business case in hand, Woolworths moved to lay the groundwork for enterprise-scale analytics. Phase 2 was all about infrastructure: deploying a converged OT network and integrating with legacy controllers to gather equipment data from every store. 

This meant enabling connectivity to thousands of controllers and installing edge gateway devices (e.g. Tridium JACEs) across all sites. Afterwards, every refrigeration rack, HVAC unit, submeter, and major load could talk to a central system. 

This infrastructure investment was a significant CapEx project and took about 18 months. This step isn’t a requirement in getting started, Hugh noted, but it paid dividends. By the end of 2019, Woolworths had something few retailers anywhere can claim: a converged OT network covering all energy-intensive assets across the entire portfolio. 

All stores were “digitally ready,” with a pathway for data to flow upwards. Most retailers lack this unified network—they’ve grown by acquisition and still have siloed systems per region or brand. Woolworths’ leadership realized that without investing in connectivity up front, any analytics would stall at pilot scale. 

Phase 3: Rapid Rollout of Analytics (2019-2020)

With data streaming in from all corners of the portfolio, Woolworths and Bueno launched full-scale analytics across the fleet. Phase 3 involved onboarding roughly 1,000 stores in one go onto Bueno’s cloud analytics platform. At its peak, the team was connecting about 70 stores per month—essentially commissioning two new stores every day. 

They built out a standardized data model enabling the application of a single fault detection rule library to all that data. Initially, many rules were generic (looking for HVAC faults, lighting schedules, etc.), but then the team developed refrigeration-specific analytics rules that targeted grocery use cases. By 2019, the platform was monitoring everything—refrigeration racks, display cases, compressors, HVAC units, lighting circuits, you name it—across over a thousand supermarkets in real time. 

This is when the savings really started to roll in: energy managers at Woolworths’ central EMC could see, for example, which stores had abnormal compressor runtimes, or which refrigeration cases weren’t maintaining setpoint, and then direct technicians where needed. 

Over the first few years, Woolworths saw energy consumption drop significantly and began seeing improvements in equipment uptime and fewer after-hours emergencies, thanks to the ability to catch the issues early.

Phase 4: Sustaining Data Quality (2020–2021)

Phase 3 proved that scaling up FDD across an enterprise can yield dramatic results. But it also taught Woolworths a hard lesson about scale: more data points mean more chances for data to go bad.

By 2020, about three years into running fleet-wide analytics, Woolworths noticed the quality of its data was degrading. Every time a store did a remodel, a controls upgrade, or even swapped out a case, points would change or disappear. With ~2 million data points, even a small percentage changing adds up—Bueno’s team estimated 5–7% of points were being renamed, removed, or replaced each year due to on-site changes. Over a few years, that compounded to perhaps 20% of the system out of sync, which also degraded analytics accuracy. 

Woolworths realized that to sustain the value of the system, they needed to continuously commission and cleanse the data at scale. So together they built a custom tool, nicknamed “Synchro,” to automatically detect and reconcile changes in the store data across the portfolio. 

Essentially, Synchro compares the live data coming from each store’s controllers to the cloud database model; if it finds a new point, a missing point, or a renamed point, it flags or fixes it. Hugh explained that about 80% of point changes could be resolved automatically (following predefined rules for common naming changes and device swaps). For the rest, the team established processes to update the model whenever store renovations occurred. By running the Synchro tool every couple of weeks across all sites, Woolworths was able to maintain roughly 95% data consistency in its platform. 

By finding the missing ~15% of data points, Woolworths was able to optimize the consumption of the previously invisible equipment. Without such automation, enterprise analytics can drown in a sea of point names and database mismatches. With it, the system stays resilient and trustworthy even as the fleet evolves.

Phase 5: Data-Driven Predictive Maintenance (2021–Present)

With a stable, high-quality data foundation in place, Woolworths spent the last two years pushing into the holy grail of facilities management: predictive maintenance. Phase 5 brought Woolworths’ maintenance data (work orders, service logs, etc.) together with the operational data to find patterns that precede equipment failures. 

By correlating seven years of work-order history with system trends, Bueno learned to spot early warning signs of issues like refrigerant leaks or ice build-up. 

The big success story involved the bane of every grocer’s maintenance team: ice build-up on refrigeration coils. It’s a common problem: coils ice over, temperatures rise, and eventually a high-temperature alarm triggers in the middle of the night, demanding an urgent (and costly) fix. 

Woolworths used FDD to change this story. They identified subtle clues in the data (like a gradual increase in defrost cycle times and case temperatures) that indicate an “ice-up” condition days before the freezer goes into alarm. Initially, their algorithms could predict these ice-up failures about 3–4 days in advance with ~65% accuracy. After iterating and feeding in more examples, the models improved to 5–7 days in advance at 95–98% accuracy. 

Woolworths can now catch a freezer trending toward failure a week ahead, create a planned work order to fix it during normal hours or at night, and avoid the emergency call-out entirely. The maintenance team can defrost the unit or replace a part proactively, before food is in danger—a huge cost savings and operational win. As Hugh put it, “We can now tell when something’s going to break down before there’s even degradation in performance… before the customer or store staff even notice anything wrong”.

Another high-impact example is refrigerant leak detection. Traditional leak detection in supermarkets often relies on fixed gas sensors (“sniffers”) or periodic manual inspections with handheld detectors. Those methods either catch the leak very late or miss it entirely. Woolworths’ analytics, by contrast, monitor pressure, temperature, and compressor patterns to infer leaks as they begin. 

In many cases, the system flags a likely refrigerant leak days or weeks before legacy methods would notice. It monitors 10 distinct parts of the refrigeration loop, looking for changes in relationships between the variables that predict a refrigerant leak, enabling a fix before product is put at risk or a major leak occurs. 

This predictive approach has contributed to Woolworths cutting its refrigerant leak rate. It also prevents catastrophic failures: losing refrigerant can burn out compressors, but catching a leak early avoids that damage. The financial stakes are huge—the average supermarket leak costs not just $20–50k in refrigerant, but potentially another $15k in emergency labor, spoilage, and downtime if it becomes a crisis. 

Grocers I talk to believe their existing EMS alarms and service contractor calls are enough. Woolworths has shown how much that approach leaves on the table. Alarms typically indicate acute failures (like a case is already warm or a compressor is off). They don’t catch inefficiencies (like suboptimal setpoints or compressors short-cycling) and they certainly don’t predict problems in advance. Alarms are also typically fragmented across the portfolio, meaning certain alarms might mean a different thing when looking from store to store.. 

Woolworths’ story proves that a dedicated analytics system can drastically outperform an alarm-based approach, finding issues that would never trigger an alarm at all. A supermarket might get an alarm when a freezer hits a high temperature, but Woolworths’ FDD can warn of an upcoming issue days prior. 

If you’re a facility manager and your organization says “we already have monitoring via alarms,” consider that a starting point, not the finish line. True analytics means looking at trends, correlations, and subtle indicators—something humans and basic alarm logic can’t do amidst thousands of data points. Cloud analytics go far beyond alarms, moving operations from firefighting into optimization and prevention.

Hugh shared that while the program started driven by energy savings, this predictive maintenance phase has become “the biggest cost savings of the whole program” for Woolworths, because they’re avoiding expensive failures and overtime repairs on a broad scale.

What American grocers should know about this case study 

When I asked Hugh Amoyal what he thinks American grocers should know about this case study, he said: it’s not as hard as you might think, your data is freer than you think, and it doesn’t cost as much as you think. 

Hugh says that traditional leak detection hardware (infrared “sniffer” sensors, etc.) can be useful, but deploying them chain-wide is expensive and still leaves blind spots. Software can monitor every rack, every case continuously via existing telemetry—something no team of technicians or scattered sensors could economically achieve. This doesn’t mean you eliminate physical detectors (a hybrid approach can work well, as we’ve discussed previously), but analytics will catch many issues that fixed sensors miss (or catch too late).

Even if your refrigeration controllers are locked down by OEMs, you can usually tap into them. “You do not need to go to Emerson or Danfoss to get data out of your fridge controllers,” Hugh said—drivers and connectors exist for virtually any brand. In other words, your data is freer than you think. Don’t let vendor lock-in scare you away; agnostic solutions can often pull data via open protocols or interface panels that are already in your stores.

Perhaps most impressively, grocers can replicate all this at a cost of roughly $2,500 per store per year—a rounding error in a typical store’s operating budget. 

All of the above suggests that the Woolworths case is very much replicable in the U.S., and in fact some American retailers are starting down this path. The main barriers are mindset and awareness—which is why we’re writing this piece! 

Next stop for Woolworths: AI 

So where does this all go next? The latest phase of their partnership with Bueno involves integrating AI in the form of a conversational assistant—essentially an LLM (large language model) tailored to their building data. 

The idea is to make the rich trove of analytics even more accessible to a wider range of people in the organization. Instead of only analysts using the system, maintenance techs, store managers, even senior executives could simply ask questions in plain English and get insights. 

Woolworths’ goal with this “Analytics AI” is to pull in additional stakeholders and get them engaged with facility data. For example, a refrigeration technician (“fridgie”) out in the field could speak into a mobile app, asking something like, “Hey, what’s going on with Store #205’s freezers right now?” and get a synthesized answer drawing from alarms, analytics, and work order history. 

By lowering the barrier to interacting with the system, Woolworths hopes to uncover even more opportunities—the collective wisdom of more eyes on the data. Many solution providers are experimenting with AI assistants for building operations, recognizing that data is only as good as the decisions it informs. Woolworths’ approach (and another recent Nexus Labs article) suggests the next evolution is to make complex analytics human-friendly for busy facility teams. 

If a smart assistant can translate data into natural language recommendations, it empowers the existing workforce (especially useful given how stretched thin they are). It also helps break down silos—sustainability teams, maintenance crews, and finance leaders could all query the same system for the answers they need. 

In Woolworths’ case, they foresee more “citizen users” leveraging analytics via AI, which in turn will drive further optimization and innovation. It opens up the toolset beyond the energy nerds (like me!) to anyone with a question about operations.