The Systems Agree the Order Exists. Who Gets It Moving?
Connected warehouses can still lose time, labor and margin between an exception being detected and someone being authorized to resolve it. Responses from ITF Group, Da Vinci and Armada expose three different weaknesses behind the same stalled shipment.
A warehouse can know that an order exists, where its inventory is recorded and when the inbound truck should arrive, yet still struggle to move the goods. A changed appointment forces a new dock decision. Stock marked available may still need handling before it can ship. When the systems report these conditions without settling what happens next, the queue moves from the software onto the floor.

Written responses to Warehouse Insider from ITF Group, warehouse software provider Da Vinci and Armada describe why that gap survives investment in connectivity. Their explanations overlap, but they are not interchangeable: ITF points to decision rights, Da Vinci to inefficient execution, and Armada to fragmented decisions and inadequate checks. Together, they suggest that a connected warehouse needs more than synchronized records. It needs a defined route from changed conditions to an executable instruction.
The investment backdrop makes that distinction consequential. The 2025 MHI Annual Industry Report, produced with Deloitte and based on responses from more than 700 supply chain leaders collected at the end of 2024, found that 55% had increased their technology and innovation budgets, according to MHI’s account of the findings. That is a measure of investment across supply chains, rather than warehouse performance. The operational question is what those investments change once an order encounters a problem.
The exception needs an owner
For Sam Agyemang, Vice President, Business Development & Sales at ITF Group, the failure begins when access to information is mistaken for authority to use it. A delay can be visible to every participant while nobody has a clear mandate to change the plan. Escalation then becomes another handoff, consuming time while leaving the underlying problem untouched.
“Warehouses rarely break down because they're missing data. Things tend to fall apart because having the data and knowing who is empowered to act on it are two different things,” Agyemang said. “Once the order, inventory and inbound information are available, execution usually stalls on exceptions: the truck is late, inventory is ‘available’ but not operationally ready, the appointment changed, or the dock suddenly has a different priority.”
“The bigger problem comes when the warehouse, carrier and customer are each working from different definitions of ‘on time,’ ‘ready’ or ‘delivered.’ Three systems can contain accurate data and still produce three different versions of the truth. Then the freight sits while people reconcile rules instead of moving product,” he said. “Before AI or automation can truly execute, companies have to agree on the source of truth, event definitions, decision rights, service thresholds, exception ownership and escalation rules.”
That diagnosis changes what an integration project has to deliver. Transmitting a status successfully establishes that information arrived; it does not establish that the receiving party interprets it the same way. If ready means physically present to one team and cleared for picking to another, faster updates can reproduce the disagreement more quickly. The practical requirement is to connect each consequential status change to an owner, an allowed action and a condition for escalation.
An order can ship and still lose money
Dan Cavanaugh, CEO & President of Da Vinci, identifies a second problem: completing the transaction can conceal the cost of completing it badly. An order eventually leaving the building says little about unnecessary travel, repeated handling or the labor required to prepare it. Those losses deserve attention even when there is no obvious system outage or missed shipment.
“It stalls with manual or inefficient processes. Every warehouse gets inventory in and shipped out ultimately,” Cavanaugh said. “The hidden cost is when you're not labeling or putting away optimally, or batching outbound orders based on order profile and packing modes that fit the order mix that maximize throughput and minimize labor. Excess travel time, touching inventory too many times, or lack of visibility into labor performance are all hidden costs inside a warehouse operation. These hidden costs can be significant margin leakage where stronger modern cloud systems can materially improve a business's performance beyond simply getting orders in and out.”
He also described what happens when shipment rules and systems diverge. “You'll have mismatched expectations and a mismatch in data that relies on fragile manual processes to keep all systems in sync. Updates to orders or required handling rules risk missed execution when systems aren't connected and working in concert.”
The distinction matters when evaluating a WMS or automation project. A dashboard showing completed orders may miss the repeated touches that made those completions expensive. Cavanaugh’s broader response calls for standardized operating procedures and complete operational data, including labor, before automation or AI can deliver reliable results. That places process design and measurement inside the implementation itself, rather than leaving them for a later improvement program.
The record must survive contact with the floor
The data problem also appeared in Warehouse Insider’s earlier coverage, “Warehouse Technology Is No Longer the Constraint. Execution Is.” In that article, Oana Jinga, Co-Founder and Chief Commercial & Product Officer at Dexory, said: “Before organisations ask what AI can optimise, they should first ask whether they trust the operational data feeding those systems.” Her observation connects the authority problem described by ITF with Da Vinci’s emphasis on operational discipline. A clear decision process still needs records that accurately describe the goods and conditions it is acting on.
Kris Jensen, Vice President, Warehouse Operations and Client Support at Armada, adds an important qualification to the debate about disconnected systems. In her account, routine transactional exchanges can work well; the weakness becomes more visible when conditions change and each participant responds independently. Sharing enough data to complete an individual transaction does not necessarily provide enough context to coordinate the response across the shipment’s journey.
“Where things tend to break down is when something changes: new products or pack-type, new customers, new carriers, or new suppliers. When conditions change unpredictably (think ‘disruption’), each party in the chain will make independent decisions to optimize their part of the whole,” Jensen said. “Since the customer, carrier, and warehouse systems are sharing only the information necessary to complete the transaction in their node of the supply chain, none of them have a view of the whole. Each will make decisions based on the limited information they have. This often results in suboptimal outcomes for everyone involved. It's less about different rules and more about disjointed decisions.”
Her description of autonomous equipment shows the consequences at floor level. “The autonomous forklift robots follow the system as directed. If the robot finds the wrong product in a location, it stops until we fix it.” That interruption makes the discrepancy impossible to leave unresolved while expecting the equipment to continue its assigned work.
Catch the error before the next receiving dock
Jensen said Armada uses computer vision at its dock doors to flag pallets moving onto or off the wrong truck. He described visual alerts that prompt dock personnel to review exceptions, with images and video also available to support feedback to suppliers and distribution centers. According to her response, the system can help identify wrong or damaged products, mixed lot codes and pallets being loaded against the wrong order. These are Armada’s descriptions of its operation; the response did not provide a measured reduction in errors or handling time.
The relevant distinction is where the check occurs. Catching a discrepancy during loading or unloading creates an opportunity to correct it before another facility inherits the problem. It also gives the receiving team evidence to investigate the cause, rather than simply recording that something went wrong. An alert still needs an assigned reviewer and a resolution, bringing the process back to Agyemang’s question of authority.
Measure the time between detection and resolution
Taken together, the responses suggest a practical way to assess whether connectivity is improving execution: trace a real exception from its first signal through to resolution. Record who recognized it, who could authorize a change, how many manual interventions followed and whether the correction reached the other affected systems. Alongside throughput, measures such as exception age, repeated handling and recurrence can expose work that completed-order counts leave hidden. This is an operational implication of the responses, rather than a performance benchmark supplied by the companies.
Each source identifies a different place to intervene. ITF makes the case for agreed meanings and decision rights; Da Vinci for efficient, standardized work supported by labor data; Armada for physical verification and coordinated responses when conditions change. A warehouse has made progress when the next late truck, changed order or incorrect pallet can move through that process without starting a fresh negotiation. The shipment needs a clear next action, and somebody or something authorized to carry it out.
