Frequently asked questions about 4PL/5PL and supply chain orchestration

4PL stands for fourth-party logistics: a party that orchestrates a client’s transport and logistics chain without carrying it out itself. A 4PL selects and manages carriers and service providers, designs the network, manages the data and reports on performance. A defining feature is that a 4PL is asset-light: no trucks or warehouses of its own, so the choice of an executing party is not influenced by an interest in filling its own capacity. Noventra works this way for shippers in the Netherlands and across Europe. See also: What we do

A 3PL carries out logistics with its own resources — trucks, warehouses, staff. A 4PL carries out nothing itself but orchestrates: selecting and managing service providers, managing the chain and reporting on performance. So the difference is not in how much you outsource, but in what: with a 3PL you outsource execution, with a 4PL you outsource decision-making. The two are not mutually exclusive. A 4PL does not replace your 3PLs, but decides who does what, on which terms, and whether that is still the best choice next quarter.

The ‘party logistics’ ladder describes how much logistics a company outsources and who is in charge of orchestration. With 1PL, the company transports goods itself, with its own vehicles and drivers. With 2PL, transport capacity is bought from a carrier or shipping line, while planning and orchestration stay in-house. With 3PL, a logistics service provider takes over execution, usually transport and storage, with its own resources. With 4PL, a neutral orchestrator manages those executing parties and manages the chain and the data. 5PL is the least clearly defined step: orchestration of several chains at once, strongly platform- and data-driven.

Supply chain orchestration is the central management and monitoring of every link in a logistics chain, rather than arranging each link separately. It covers designing the network, selecting and contracting carriers, day-to-day planning and monitoring, intervening when deviations occur and reporting on costs, performance and emissions. The aim is one coherent picture and one set of management figures, so that decisions are based on the whole chain rather than on individual routes. Orchestration pays off most when several carriers, countries or modes are involved. See also: A transparent, manageable transport chain

A logistics control tower is a central point where all transport flows, systems and parties involved come together. From there, shipments are planned and tracked, deviations are flagged and dealt with, and accountability is provided through reports. A defining feature is that status information comes in in real time and that action is taken proactively: an impending delay is reported before it has an impact, rather than someone phoning round afterwards. The term comes from aviation, where a control tower does the same for aircraft movements. See also: OrangeChain

A freight forwarder arranges transport for a client and buys capacity for it, usually per shipment or per route, and earns on the difference between the buying and selling price of that transport. A 4PL works one level higher and with a different business model: it takes on orchestration of the entire chain, contracts and manages several carriers and forwarders, and is paid for the orchestration rather than for a margin on the transport. The practical consequence is that a 4PL has no interest in a higher transport rate.

Outbound logistics is the outgoing flow to your customers; inbound logistics is the supply from your suppliers to your own locations. At many companies, outbound is well organised while inbound stays out of sight, because suppliers arrange the transport and there are only loose commitments about arrival dates. That costs visibility and negotiating room: you do not know exactly what arrives when, and you pay for the transport indirectly through the purchase price. Bringing both flows under the same orchestration means they are measured with the same definitions and trips can be combined.

5PL stands for fifth-party logistics. The term is usually used for a highly digital and automated model in which several logistics chains and networks are optimised through data and technology.

Where 4PL mainly orchestrates the chain of a single client, 5PL focuses more on platforms, data, scale and extensive automation. The term is less clearly defined than 3PL and 4PL.

A 4PL focuses on orchestrating the logistics chain: managing carriers, measuring performance, organising capacity and continuously improving the chain.

With 5PL, the emphasis is more on digital platforms, data and extensive automation across several logistics flows or networks. In practice, the terms are sometimes used interchangeably.

AI is not the same as 5PL, but it can be an important technology in the move towards it. With AI, supply chain orchestration shifts from reporting and reacting to predicting, advising and acting automatically.

AI agents can take over part of the day-to-day orchestration, while people set the goals, boundaries and exceptions. See also: Our own AI agents

Outsourcing pays off once transport becomes complex enough to need daily management, while it is not your core business. Concrete signs: you work with several carriers, countries or modes; the overview lives in mailboxes and spreadsheets; you lack a reliable overall picture of costs and performance because each carrier uses its own definitions; or your volumes fluctuate sharply due to season or growth. If you have one stable flow with one regular carrier and fixed rates, an orchestration layer mainly adds an extra link.

A 4PL usually charges a fee for orchestration rather than a margin on the transport, sometimes supplemented by a performance-based element tied to demonstrable savings. A fixed rate per shipment can rarely be given in advance: the costs depend on the number of lanes, the modes, the volumes, the extent of peaks and the system integrations required. A serious proposal therefore starts with an analysis of your actual shipments and rates. At Noventra, your lanes, volumes and requirements are mapped first, followed by a concrete cost proposal for your flows.

Choose a 3PL if you mainly need operational capacity: transport and storage for a defined flow, while you handle the orchestration yourself. Choose a 4PL if coordination itself is the problem — many carriers, fragmented data, figures that cannot be compared or insufficient in-house capacity for daily management. In practice they go together: the 4PL orchestrates, the 3PLs drive and store. So the question is not which of the two, but at what level you want the decision to sit.

A switch starts with data, not with a contract: without visibility of current flows, rates and performance, there is nothing to design. Next come the network design, tendering and contracting carriers, and the technical set-up — building integrations, getting master data in order and going live flow by flow rather than all at once. At Noventra this follows three steps: introduction, in which your current situation, ambitions and challenges are discussed; insight and proposal, in which we map your relevant flows and translate them into a suitable proposal; and moving forward together, in which you decide whether and how we continue. See also: Plan an introduction

Because a service provider with its own trucks always has a second interest: those trucks need to be on the road and full. In times of scarcity, peak pressure or a commercially unattractive trip, that interest can clash with what is best for the customer’s cargo. An orchestrator without capacity of its own chooses purely on suitability, availability and measured performance, and can switch carriers without losing out. When assessing a 4PL, the first question is therefore whether that party transports goods itself — and so whether the reporting on carrier performance also covers itself. Noventra does not run its own trucks. See also: Independent orchestration

The most widely used transport KPIs are on-time performance, OTIF, cost per shipment or per tonne-kilometre, load factor, lead time, invoice deviation and CO₂ per tonne-kilometre. More important than the choice of indicators is that the definitions are the same for all carriers — otherwise figures cannot be added up or compared. Noventra does not fix that set in advance: which indicators count and with which targets is decided per client, in line with what that organisation wants to manage. Common examples are on-time arrival within the booked time slot (90%), on-date arrival (98%) and order data in line with the booking (98%). The agreed set is reported periodically and re-examined at each review. See also: You decide what gets measured

OTIF stands for on time in full: the share of shipments delivered both on time and complete. A shipment only counts if both conditions are met — late or incomplete means no OTIF. That makes it a stricter measure than delivery reliability alone, and the most widely used headline indicator of supply chain performance. When comparing OTIF figures, always check the definition: from which moment it is measured, with what tolerance in hours, and from the perspective of the booked time slot or of the end customer. See also: You decide what gets measured

By agreeing targets in advance, measuring them with the same definitions and discussing them at fixed moments. A workable cycle is monthly reporting, a quarterly review in which results and points for improvement are recorded, and, in the case of structural underperformance, an automatic improvement plan with reassessment. Without an agreed consequence, a KPI remains non-committal. Crucially, there must be an alternative: if you cannot switch carriers, you have no means of control. That is easier when the orchestrator has no capacity of its own to fill. See also: You decide what gets measured

That depends on exactly what is being measured. Arrival within a specific time slot is sensitive to traffic jams, waiting time at the loading and unloading docks and congestion on site; targets of around 90% are usual there. At day level the bar is higher, typically around 98%, because missing a day is a real disruption while missing an hour often is not. A 100% target at slot level sounds strict, but in practice mainly leads to discussions about force majeure rather than to process improvement.

The acceptance rate, also known as tender acceptance, is the share of offered trips that a carrier actually accepts under the contract agreements made. A low acceptance rate means falling back on spot capacity: more expensive, less predictable and with more variation in quality. Structural refusal is usually a sign that the rate, the volume or the trip profile no longer suits that carrier. The indicator is therefore viewed together with cost development: a sharp contract rate that nobody accepts is, in practice, more expensive than a realistic one.

A transport management system (TMS) is software in which transport orders are created, planned, tracked and settled. A good TMS brings orders, statuses, documents and performance figures together in one place, connects to the shipper’s ERP or planning system and to carriers’ systems, and makes figures traceable down to shipment level. Without such a system, information gets scattered across mailboxes, carrier portals and spreadsheets, making totals and comparisons unreliable. Noventra uses its own platform, OrangeChain, for this. See also: OrangeChain

Usually through an API integration: the ERP, WMS or planning system forwards orders automatically and receives status updates in return, without double entry. Older integrations use EDI messages; working through a dashboard without an integration is also possible, but takes manual work. The advantage of connecting to an orchestrator rather than to each carrier separately is that you maintain one integration: if a carrier is added, nothing changes on your side and your IT department does not need to build anything. See also: OrangeChain

Management by exception, also known as exception management, means you only look at shipments where something is wrong, rather than going through the entire order list. The system marks which orders need action based on incoming status information; the rest runs without intervention. The effect is that attention goes to the ten shipments that need it rather than the thousand that do not. The precondition is that status updates arrive quickly and completely; otherwise issues are flagged too late or wrongly. See also: OrangeChain

Transport falls under Scope 3 of the GHG Protocol and appears in CSRD reporting through the ESRS climate standards. The practical requirement is that figures are traceable and verifiable: an auditor wants to be able to see how a number was built up. That cannot be done with a single estimate after the fact, but it can with registration per shipment according to a recognised method such as ISO 14083, stating the data quality of each figure. Noventra records emissions per shipment and keeps them traceable down to shipment level. See also: Sustainability

ISO 14083 is the international standard for quantifying and reporting greenhouse gas emissions from transport chains, published in 2023. The standard describes how emissions should be calculated and allocated per shipment and per link, so that figures from different carriers and modes are comparable with one another. Without such a common method, every party calculates differently and adding up becomes meaningless. For shippers, the practical consequence is that the figures can be used directly in their own sustainability reporting, rather than first having to be converted or justified.

That varies widely, and the reliability should be stated alongside the figure. There are roughly four levels: default values per mode and vehicle type, averages of the carrier concerned, a calculation based on the actual route and load of that specific trip, and measured fuel or energy consumption. Only the last two say anything about your specific shipment. A figure that does not state its level is therefore hard to use in reporting. Ask every supplier how a number is arrived at and at what level it sits. See also: Sustainability

That is a contractual question, and it is wise to settle it in advance. Put in writing that your shipment and performance history remains yours, that it is only used for carrying out and administering your transport, and that you can request the data in a usable format. That last point is the most important: supply chain orchestration provides a great deal of insight but also creates dependency, and anyone who can take their own history with them keeps the freedom to switch. In addition, the processing of personal data is subject to the GDPR.

In the event of damage or loss, a claim is submitted to the party that carried out the transport, within the time limits and liability limits of the applicable transport law — for road transport usually the CMR Convention internationally and the Dutch AVC conditions domestically. That liability is tied to weight, not to the value of the cargo, so additional insurance is customary for high-value goods. In practice, the most important thing is that claims are recorded and tracked in one file and reflected in reporting, so it becomes visible whether certain carriers, lanes or cargo types structurally lead to damage.

ISO 27001 is the international standard for information security. For a logistics service provider, working to this standard means, among other things, encrypted data exchange, personal user accounts instead of shared login details, documented procedures and reporting security incidents without undue delay. Note the difference between being certified and ‘working according to the system’ — that difference is significant and the two are often confused in the market. Noventra is setting up its information security in line with ISO 27001; the certification process is under way and has not yet been completed.

AI is increasingly turning a TMS into a predictive system. By combining historical transport data with current market information, such as traffic, weather, congestion, capacity and carrier performance, AI can predict what is likely to happen.

Think of flagging delays, capacity problems or rising transport costs early. This shifts transport management from recording and reacting to predicting and steering ahead. AI can warn a planner, propose an alternative or carry out an action independently within predefined limits. See also: OrangeChain

Yes. AI can combine historical shipments with current information on, for example, traffic, weather, congestion and a carrier’s performance. This allows the system to estimate which shipments are at increased risk of delay.

The advantage is that you do not have to wait until a shipment is actually late. You can intervene earlier and, for example, adjust capacity, planning or the carrier. See also: Proactive management

An AI agent goes a step further than a chatbot or conventional automation. An agent can independently gather information, assess a situation and carry out an action within predefined limits.

In transport, an agent can, for example, continuously monitor a shipment, recognise an impending delay, look for alternative capacity and make a proposal to the planner. People stay in charge of the boundaries and exceptions. See also: Our own AI agents

Automation works with fixed rules: if A happens, do B. AI can also recognise patterns and predict what is likely to happen. An AI agent can then also carry out several steps in response.

For example: automation processes a delivery status, AI predicts a delay and an AI agent then investigates which solution is available.

That depends on the application and the data sources available. Think of traffic information, weather conditions, congestion, border situations, available capacity, fuel prices, market prices and historical carrier performance.

It is precisely the combination with your own historical data that makes this information valuable. This way AI can not only see what is happening, but predict what it is likely to mean for your transport.

Traditionally, a TMS mainly looks at what is happening or what has already happened. Predictive transport management adds an expectation: what is likely to happen?

By combining historical and current data, the system can, for example, predict which shipments are at risk, where capacity is becoming scarce or which carrier is likely to fall short of the agreed performance. That lets you adjust sooner.

AI works with valuable business data: orders, rates, routes, customer information and carrier performance. With your own or a private AI environment, you keep more control over where this data is processed and which data is available to AI.

Moreover, having your own AI server does not mean you have to develop an AI model yourself. Existing models can also be used within a controlled environment.

With cloud AI, the computing power and often the AI model as well are supplied by an external provider. With your own or a private AI server, processing takes place within an environment that you manage yourself or that is managed on your behalf.

The main difference lies in control over data, security, infrastructure and processing. Which solution fits best depends on the sensitivity of the data, the usage and the scalability required.

Your own AI environment can be attractive when control over business data, privacy and security is important. It can also offer advantages when large volumes of transport data need to be analysed continuously.

A hybrid solution is also possible: sensitive operational data stays within your own environment, while external models are used for generic AI applications.

Technically, an AI agent can carry out more and more actions independently. In a transport environment, the key is to determine in advance what an agent may and may not do.

An agent can, for example, independently flag a risk or look for alternative capacity. For important decisions, such as a large deviation from a contract rate, human approval can remain mandatory. Moreover, every action must be traceable afterwards. See also: Independent orchestration