This guide is for operations, product and IT leads who are preparing a connected product, a sensor rollout or a machine-data project and want proposals they can compare. It is the scoping companion to our IoT development service, which describes how Netbase builds the platform, data and integration side of the work.
In this guide
- Why do IoT projects need a layer-by-layer scope?
- The four-layer scoping sheet
- How to run the scoping work in six steps
- When should you not start an IoT build?
- Where AI fits in an IoT scope
- What delivery record exists, and what does not?
- Alternatives: who can own the scope
- Limits of this guide
- Frequently asked questions
- Next step
Why do IoT projects need a layer-by-layer scope?
An IoT project fails most often at the boundary between two layers, not inside one. A device maker delivers hardware that works, a network provider delivers coverage that works, and a software partner delivers a dashboard that works, yet nobody owns the message format, the device identity or the alert that should create a work order.
A layer sheet closes those gaps before money is committed. It turns "connect our machines" into four lists that a device maker, a network provider and a software partner can each price and test against. It also shows early which questions only the business can answer, such as which decision the data must drive and who acts on an alert.
A common reference model splits an IoT solution into device, connectivity, cloud or platform, and application layers; some models add an edge gateway or a separate security layer. Four layers are enough for scoping, as long as security and edge processing appear as questions on every sheet rather than as a fifth box that nobody owns.
The four-layer scoping sheet
| Layer | Questions to answer before quotes | Usually owned by | Most common gap |
|---|---|---|---|
| Devices | What is measured, how often, on what power, and how is firmware updated? | Device maker or hardware team | No plan for firmware updates or device replacement |
| Connectivity | Which network on each site, what happens offline, and which protocol? | Network provider and device maker | Offline buffering and reconnect behaviour left undefined |
| Cloud | How do devices prove identity, where is data stored, for how long, and who can read it? | Software partner, with the client's IT | Retention and data ownership decided after launch |
| Application | Which screen, alert or system record does each signal change, and who acts on it? | Software partner and business owner | Dashboards with no owner and alerts nobody answers |
Devices. List every device type with its sensors, reading frequency, power source and expected lifetime. Note who supplies it, who updates its firmware and how a faulty unit is replaced. If the hardware is not chosen yet, the device maker's interface documentation is the first deliverable, not the dashboard. The NIST IR 8259 guidance on foundational cybersecurity activities for device manufacturers is a useful checklist for what to ask a device supplier.
Connectivity. For each site, write the network (Wi-Fi, cellular, a low-power wide-area network or wired), the expected coverage and what a device does when the connection drops. Choose the messaging protocol with the device maker: MQTT, an OASIS standard, is a lightweight publish and subscribe protocol that many constrained devices use, and CoAP is a web transfer protocol designed for constrained nodes and networks.
Cloud. Decide how each device proves who it is, for example with X.509 certificates as AWS IoT Core documents for its device authentication. Then write where readings are stored, how long raw data is kept, what is summarised, who may export it and which rules run on arrival. Managed cloud IoT services such as AWS IoT Core or Azure IoT Hub handle much of the device connection and identity work; an own message broker gives more control and more operating duty.
Application. For each signal, name the decision it should change: a work order, a stock update, a customer notification or a forecast. Then name the person or system that acts on it. A signal without a named decision is a candidate for removal from the first phase.
How to run the scoping work in six steps
-
Start from decisions, not devices
Write the three to five business decisions the data must support, and the person who makes each one today.
-
Map signals to decisions
For each decision, list the readings it needs, how fresh they must be and how often they arrive.
-
Fill the four layer sheets
Use the table above; every row gets an owner, and every unknown gets a named person who will answer it by a date.
-
Draw the out-of-scope lines
Write what the software partner does not do: device hardware design, radio or safety certification, network installation on site and connectivity contracts are common exclusions.
-
Define the pilot
Pick a small set of real devices on one site, one end-to-end data path and one integration into a business system, with exit criteria written before work starts.
-
Price the first phase only
Ask for a fixed price for the pilot and a cost-driver list for the rollout: device types, protocols, message volume, integrations, offline needs, retention and the security review your sector needs.
Pilot exit criteria should be checkable without opinion: the devices report end to end for an agreed period, a lost connection recovers without data loss, the alert reaches the named person, and the integrated system shows the record it should. If a criterion needs a meeting to decide, rewrite it.
When should you not start an IoT build?
Do not start when nobody owns the decisions the data should drive; the result is a dashboard nobody opens. Do not start when a one-off data export from equipment you already have would answer the question. Do not build a platform when one off-the-shelf sensor kit with its own app covers a single use case, and do not start when the device itself is still undecided, because the protocol, power and identity choices all follow from it.
Where AI fits in an IoT scope
AI belongs in the cloud and application sheets as an option, not as a starting requirement. Anomaly detection can flag readings outside normal patterns, forecasting can estimate when a machine or a stock level needs attention, and computer vision can treat a camera as the sensor. Each needs historical data, so the first phase should store clean, labelled readings even if no model runs yet.
Engineers still decide the thresholds, compare a model with the existing rules on real data, and keep a person in the loop who confirms any action such as a work order or a shutdown. Netbase works with the major commercial and open-source models and chooses them per project. AI with IoT is one of Netbase's AI capability areas, and IoT with AI at the edge is an area Netbase is investing in.
- Available capability
AI with IoT, anomaly detection and forecasting
A capability area Netbase offers; not yet tied to a published IoT project.
What delivery record exists, and what does not?
No Netbase IoT project is published yet, and this guide claims none. IoT with AI at the edge is an investment direction for Netbase, alongside AI integration and security by design.
The platform and integration side of the work is on record. Netbase has delivered projects for clients in manufacturing and logistics that are not named, including an operations and integration platform built as consultant, custom developer and integrator; that record publishes no figures and involves no devices. Delivery follows a six-step lifecycle from discovery to ongoing support, with weekly reviews, KPI dashboards and a dedicated account manager and project manager. Teams range from 3 to 30 people, and work typically starts within one to two weeks after discovery.
Most Netbase projects are delivered on fixed-price contracts. Security practices include role-based access control, MFA for admin dashboards, TLS in transit and AES at rest, secure code review and vulnerability scanning. Netbase hosts on AWS, Google Cloud, DigitalOcean or Cloudflare and claims no cloud partner tier. No device count, uptime, duration or cost result is claimed here.
Alternatives: who can own the scope
| Route | Choose it when | Watch for |
|---|---|---|
| Device maker leads | The product is mainly hardware and the vendor already offers a cloud and app | The data stays in the vendor's cloud and is hard to connect to your systems |
| Managed IoT platform plus in-house team | Your IT team already runs the chosen cloud and can staff operations | Device identity, retention and alerting still need an owner |
| Software partner leads the platform | Device data must reach ERP, maintenance or order systems you already run | The partner must state what it does not do on the device and network side |
| Off-the-shelf sensor kit | One use case, one site, no integration needed | No path to more device types or business systems |
Selection criteria: who owns the data after launch, how many systems the data must reach, whether your team can operate the platform, and how many device types you expect within two years. Where device data has to update operational records, the integration design matters more than the dashboard; our guides on ERP, CRM and ecommerce integration architecture and manufacturing and logistics workflow automation cover that side, and the integration hub shows the pattern Netbase uses when many systems exchange data. IoT also appears among the trends in our technology trends 2026 guide.
Limits of this guide
- It covers scoping, not platform selection; managed services are named only as examples.
- It gives no protocol or network recommendation for a specific device; that choice belongs to the device maker and the site survey.
- It states no figure for cost, duration or device volume, because those depend on the four layer sheets.
Plan the next step with a Netbase consultant
Frequently asked questions
It should include four layer sheets for devices, connectivity, cloud and application, each with an owner and an out-of-scope line, plus the business decisions the data must drive and written exit criteria for a pilot on real devices.
The device maker or your hardware team should own the device layer, including firmware updates and replacement. The software partner should own the documented interface it receives from that layer and test against real devices early.
An IoT pilot should be small enough to finish quickly and real enough to prove the data path: a few devices on one site, one end-to-end path and one integration into a business system, judged by exit criteria agreed before work starts.
A managed service usually fits when your team already runs that cloud and wants device identity and connection handled for it. An own broker fits when you need more control over hosting or protocols and can staff its operation.
Yes, if the first phase stores clean, labelled readings with enough history. Anomaly detection and forecasting need that history, and they should run next to the existing rules before they replace any of them.
Next step
If you have a first draft of your layer sheets, book a solution review and we will mark the gaps and the questions to answer before a pilot.
Related services and solutions
IoT development with AI analytics that connects devices to the business Netbase provides IoT development for manufacturers, logistics operators and product companies to connect devices and sensor data to the business platforms that act on them, with AI that spots anomalies and forecasts what needs attention. We design the device-to-cloud data path, build the back end, dashboards and integrations, and treat security, data ownership and AI as design decisions.
Learn More
API integration services that keep your systems in agreement and ready for AI Netbase provides API integration services for commerce and operations teams that need e-commerce, ERP, CRM, shipping and payment systems to agree, so data moves once and AI services can act on it, with retries, monitoring and reconciliation on every API and webhook. According to its published client record, Netbase has worked with 500+ businesses across 80 countries since 2012.
Learn More
AI and data services, from readiness check to production feature Netbase provides AI and data services for companies that want to move from an AI idea to a feature that works inside a real product or process: readiness assessment, data preparation, model and automation development, and production support. Delivered AI includes 4over4's recommendation engine, plus RAG, document AI and MLOps projects for unnamed clients; the rest is growing.
Learn More
Enterprise integration hub ready for AI agents: one monitored layer for commerce, ERP and CRM data An enterprise integration hub is a central layer that moves orders, customers, stock and invoices between your commerce, ERP and CRM systems, with monitoring, retries and replay in one place, and scoped APIs that AI agents can call safely. Netbase builds it as a custom layer on the stack you already run, starting from the flows that break most often.
Learn More
Work with Netbase remotely
Every Netbase engagement is delivered remotely, so clients in any country and any industry can work with us; Netbase has worked with clients in 80 countries. Contracts and invoices are issued by Netbase JSC, and payment is made by bank transfer to Netbase JSC's company account or PayPal Business. Talk to the CEO, Huy Nguyen (David), on WhatsApp or by email.
Discuss a project
Netbase JSC helps organizations design, build, modernize, and operate digital products and AI-enabled business systems.project@netbasejsc.com
+84 937 869 689
91 Nguyen Chi Thanh, Dong Da, Hanoi, Vietnam
Get in touch
Tell us what you want to build, modernize, or operate.