The gate is the only point in the whole container loop where every unit is guaranteed to stop, be looked at, and be handled by someone whose job includes recording things. That makes it the cheapest identification point available and, at most plants, the least used one — because gate capture is scoped around the vehicle and the paperwork rather than around the assets on board. A truck arrives, the delivery is matched, the barrier lifts, and forty returnable racks pass through with no record that they specifically were the racks that arrived. The consequence shows up months later as a pool nobody can reconcile and a loss rate nobody can attribute. Getting this right is mostly a matter of deciding what to capture, choosing a method whose failure mode you can live with, and having a rule for the damaged label that does not amount to waving it through. Talk to a solutions engineer to design capture for your own gate.
2026 GUIDE · GATE CAPTURE
Container Identification at the Plant Gate
Label and marking capture designed around a chosen failure mode, damaged labels handled by rule rather than by judgement, and arriving units reconciled against what you were expecting back.
Transport label
Licence plateOrg code + serial
Part numberSupplier reference
QuantityPer handling unit
Batch / dateOne per unit only
Single barcode field, scanned once
The label tells you what arrived. It does not tell you whether it was the unit you were owed.
What to Capture, and Why
Six fields. The first two are what most gates already record; the rest are what turn a delivery record into an asset record.
Licence plate identifierOrganisation code plus serial number, unique within a year, readable from one barcode field. The anchor for everything downstream.EnablesReceipt matching without relabelling
Part number and quantityPer handling unit. Must match the despatch advice and the delivery note — mismatch here is an exception, not a note.EnablesQuantity variance detection before unload
Container type codeWhich physical container standard this is. Frequently omitted, and without it pool counts cannot be split by type.EnablesPer-type pool sizing and cycle time
Asset identifier, where it existsA persistent marking on the container itself, separate from the shipment label. The only field that survives unpacking.EnablesCycle time, loss attribution, stagnation detection
Direction and stateInbound full, inbound empty, outbound empty, outbound full. Four states, and gates that record only "arrived" lose half the loop.EnablesReturn reconciliation and net drift per lane
Condition flagServiceable, damaged, label illegible. A single field at the barrier prevents a damaged unit re-entering circulation silently.EnablesPool quality control and repair routing
The field that changes the most
Direction and state. Almost every gate records inbound arrivals; far fewer record outbound empties leaving on the collecting vehicle. Without that second event you can see containers coming in and never going out, which means the pool position drifts and the drift looks like loss. Capturing both directions at one barrier is usually a configuration change rather than a project.
Capture Methods and Their Failure Modes
Choose the method whose failure you can tolerate at your volumes, not the one with the best headline read rate. Every method fails; what differs is how, and how visibly.
← Swipe to see all columns →
| Method | Speed at the barrier | Fails when | Failure is |
|---|---|---|---|
| Manual keying | Slowest | Always, at a rate proportional to volume and shift length | Silent — a plausible wrong number looks identical to a right one |
| 1D barcode scan | Fast | Label damaged, dirty, wet, obscured by load or film | Visible — the scan does not complete |
| 2D barcode scan | Fast | Same physical causes, but tolerates partial damage better | Visible, and less often triggered |
| RFID portal | Fastest, hands-free | Tag damaged, shielded by metal or liquid, or obscured by the load | Silent — an unread tag is indistinguishable from an absent unit |
| Fixed vision or OCR | Fast, hands-free | Poor light, weather, non-standard label placement | Visible if confidence scoring is enabled, silent if not |
| Handheld fallback | Slow but reliable | Rarely, if the operator is available | Visible, and it is the fallback the other rows depend on |
The distinction that matters is the last column. A method that fails visibly costs you seconds and an exception record. A method that fails silently costs you a container you will spend a quarter looking for. An RFID portal missing three units in a load of forty is the most expensive failure on this table precisely because nothing at the barrier indicates it happened — which is why portal reads need a count reconciliation against the despatch advice rather than being trusted on their own.
Design the capture
A solutions engineer walks your barrier, volumes, label compliance rate and container mix, and recommends a method whose failure mode fits your operation — plus the reconciliation that catches what it misses.
Talk to a Solutions Engineer
Try it first
Start with handheld capture at the barrier on your existing devices, record both directions, and see what your reconciliation gap actually is before specifying any fixed infrastructure.
Start Free Trial
The Damaged Label Rule
Every gate has an informal answer to this and it is usually "wave it through and sort it out later". Later never arrives, and the unit enters circulation with no identity. These four rungs replace the informal answer.
R1
Readable by fallbackThe barcode fails but the human-readable text is legible. Capture manually, flag as a label defect against the supplier, admit normally. Seconds of cost.
R2
Recoverable from the adviceLabel unreadable, but the despatch advice plus position in the load identifies it unambiguously. Capture by inference, mark the identity as derived rather than scanned.
R3
AmbiguousCannot be resolved at the barrier. Divert to the exception lane rather than queueing behind normal flow, and resolve before staging. This is the rung most gates skip.
R4
UnidentifiableAssign a temporary identity, relabel to standard before the unit enters the pool, and record the event against the supplier. Never admit an unlabelled unit into circulation.
Why label quality is a supplier metric
Not every supplier has the back-end systems to produce compliant labels to specification consistently — which is a known and acknowledged problem in the industry, to the point that free generator tools exist specifically to help smaller suppliers produce conformant labels from a simple data file. Treat label defects as a scored supplier issue with a route to a fix, rather than as a gate annoyance to absorb. The gate is where you find it; the supplier is where it gets solved.
Reconciling Against Expected Returns
Capture on its own produces a list. Reconciliation against what you were expecting produces a decision. Four outcomes, each with a different action.
Matched
Unit arrived, was expected, identity and condition confirmed. Close the open expectation and update the loop state.
ActionNone. This should be the overwhelming majority and is the baseline everything else is measured against.
Unexpected arrival
A unit arrived that was not on any expectation — commonly a container drifting back from another site, or one written off prematurely.
ActionAccept, reinstate into the pool count, and check whether it was previously written off. Recovered units are cheaper than purchased ones.
Expected, not arrived
The collection was scheduled and the unit did not appear. The most useful signal on this list, and the one that has no physical trigger.
ActionOpen a stagnation record against the last known holder. Absence must generate an event or it generates nothing.
Identity mismatch
The unit that arrived is not the unit the record says arrived — a keying error, a duplicated serial, or a label applied to the wrong container.
ActionDivert and resolve before staging. A mismatch admitted into the pool corrupts every subsequent reconciliation on that unit.
Design Rules for Gate Capture
Seven rules that determine whether capture survives a wet Tuesday with a queue building on the access road.
1Capture must not extend the gate transaction meaningfully. A method that adds a minute per truck will be bypassed within a fortnight at any real volume.
2Every automated method needs a manual fallback that is faster to use than to avoid.
3Reconcile counts against the despatch advice, always. Hands-free methods fail silently and only a count check exposes it.
4Record both directions at the same barrier. One-way capture guarantees an unexplainable pool position.
5Mark derived identities as derived. An inferred identifier and a scanned one should never look the same in the record.
6Divert exceptions out of the queue rather than resolving them in it. One ambiguous unit should not hold the barrier.
7Label placement is a supplier requirement, not a hope. Obscured labels are the commonest cause of read failure and the easiest to specify away.
Start at the barrier
Bring your gate layout, transaction times, label compliance rate and container mix. We will design the capture method and the reconciliation that catches its failures, and show what both directions of capture would reveal about your current pool position.
What to Measure
Six figures. Read rate is the obvious one and the least informative on its own. Our analytics and reporting module carries them.
First-pass read rateUnits identified without fallback, by method and by supplier. Split by supplier or you cannot tell a label problem from a portal problem.
Silent-miss estimateGap between despatch advice count and captured count. The only visibility you get into hands-free failures.
Derived identity shareUnits identified by inference rather than by scan. Rising values mean label quality is degrading somewhere specific.
Added transaction timeSeconds per truck attributable to capture. The number that predicts whether the process survives peak.
Reconciliation outcome mixMatched, unexpected, not arrived, mismatch. The shape of this distribution is the health of the whole loop.
Label defects per supplierScored and fed to supplier logistics. Cheap to collect at the gate and the only route to actually fixing it.
Frequently Asked Questions
Is RFID the right choice for gate capture?
It is the fastest and hands-free, which is genuinely valuable at a busy barrier — but its failure mode is the worst on the list. A tag that is shielded by metal, obscured by the load or damaged simply does not read, and an unread tag is indistinguishable from a unit that was never there. That makes count reconciliation against the despatch advice mandatory rather than optional with RFID. Where you can pair a portal with an advice-count check, it works well; where you cannot, a 2D scan with a visible failure is often the safer choice.
What should happen when a label is unreadable?
Follow a rule rather than a judgement, because the informal answer is always to wave it through. If human-readable text is legible, capture manually and flag a label defect. If the despatch advice plus load position identifies it unambiguously, capture by inference and mark the identity as derived. If it is ambiguous, divert to an exception lane rather than resolving it in the queue. If it is unidentifiable, assign a temporary identity and relabel to standard before the unit enters the pool — never admit an unlabelled unit into circulation.
Why record outbound as well as inbound?
Because a pool position needs both sides. If you capture containers arriving and not containers leaving on the collecting vehicle, your records show units accumulating indefinitely and the drift eventually gets interpreted as loss — while the units in question are sitting perfectly safely at a supplier. Capturing both directions at the same barrier is usually a configuration change rather than new infrastructure, and it is the single change that makes reconciliation possible at all.
Should the gate identify the shipment or the container?
Both, and it is worth keeping them separate in the record. The transport label carries a licence plate — organisation code plus serial number, unique within a year — that identifies the handling unit and enables continuous tracking through the transport chain, which is exactly what receipt matching needs. But it ends at unpacking, so if you want cycle time and loss attribution you also need a persistent asset identifier on the container itself. Capture both fields where available and never let one stand in for the other.
How much time can capture reasonably add?
Very little, and this constraint drives most of the design. A gate transaction has a working range measured in a few minutes, and any capture step that pushes meaningfully beyond it will be bypassed under pressure — which produces worse data than having no process. Hands-free methods add almost nothing but fail silently; scanning adds seconds and fails visibly. The practical answer is usually the faster method plus a reconciliation check, rather than the slower method plus trust.
What if suppliers cannot produce compliant labels?
It is a recognised gap rather than an unusual one — not all suppliers have back-end systems capable of producing labels to the correct specification consistently, and free tools exist that generate conformant labels from a simple data file precisely to close it for smaller suppliers. Score label defects per supplier at the gate, publish the figure, and point non-compliant suppliers at the available tooling. The data content agreed in the logistics agreement remains their responsibility, so the fix is contractual with a technical assist rather than something the gate should absorb.
Where should we start?
Handheld capture of both directions on devices your gate team already has, for one or two container types. It costs nothing, produces a reconciliation gap figure within weeks, and that figure is what justifies — or rules out — any fixed infrastructure. Most plants find the gap is larger than expected and concentrated in outbound empties, which is both the cheapest thing to fix and the reason their pool count never balanced. Our integrations overview covers the capture and data surface.
Identify It Where It Already Stops
Six fields captured in both directions, a method chosen for its failure mode rather than its read rate, damaged labels handled by a written rule, and every arrival reconciled against what you were expecting back.
Works with GTL and Odette labelling · Runs on existing handsets · No gate hardware required to start
August 17, 2026
By Alex Rowan
All Articles
