Skip to main content

What are you looking for?

Explore our services and discover how we can help you achieve your goals

Client Workspace

When does a digital twin of physical assets pay back?

A digital twin pays back when it changes a frequent, costly decision about physical assets and the data to feed it already exists. This Netbase guide applies a five-question filter: which decision changes, what an error costs, how often it is made, which data exists, and who acts. Most proposals pass as dashboards first.

Book a solution review See the related solution

By Netbase Commerce and Platform Engineering Desk · Reviewed by David (CEO) · Updated · 9 min read

star

This guide is for operations, maintenance and technology leads of asset-heavy mid-market companies who have been offered a "digital twin". It sits under Netbase's IoT development service, which connects devices to the business systems that act on their data.

In this guide

What is a digital twin, and what is it not?

A digital twin is a software model of a physical asset, line or site that is kept in step with the real thing through data, so that people can see its current state, how it got there and what may happen next. NIST describes digital twins as electronic representations of real-world entities that let users view the states of those entities and the transitions between them (NIST IR 8356, accessed 2026-11-19).

A 3D model with no live data is a visualisation, a dashboard of sensor values is monitoring, and a one-off simulation study is analysis. Each can be the right purchase, and none needs a model kept in step with an asset for years.

The useful test is the loop: data flows from the asset into the model, the model produces a state or a forecast, and a person or a system changes something on the asset because of it. Without that loop, a twin is a picture.

Which use cases pass the payback filter?

A proposed twin passes when all five questions below get a yes. Ask them before any platform discussion.

  1. Decision: Does the model change a named decision, such as when to service a pump, how to sequence a line or where to move stock?
  2. Cost of being wrong: Is a wrong decision today expensive enough (downtime, scrap, missed delivery, safety risk) to pay for a model that is maintained for years?
  3. Frequency: Is the decision made often, daily or weekly, rather than once a year?
  4. Data that already exists: Are the signals the decision needs already captured, with history, or can they be captured without a hardware programme?
  5. Owner who acts: Is there a team with the time and authority to act on what the model says?

A proposal with five yes answers is a twin candidate. A proposal that fails question 4 is a data project first. A proposal that fails question 1 or 5 should not be built at all, however impressive the demo. Gradion, a manufacturing software vendor, warns that a twin bought before the data foundation is in place is a simulation tool rather than a live operational asset (Gradion, accessed 2026-11-19).

The use cases that most often pass are condition monitoring of a few critical assets, predictive maintenance where failure history exists, and planning a line whose layout changes often. Whole-site twins rarely pass first time: question 2 holds for a few assets only.

What data does each use case need at minimum?

The minimum is the smallest set of data that lets the first version change the decision. Everything else can wait for a later release.

  • Condition monitoring

    Minimum data for a first version
    Live readings for the few signals that show health (for example vibration, temperature, current), an asset register with one ID per asset, alert thresholds agreed with maintenance
    Can wait
    3D views, links to the ERP, forecasts
  • Predictive maintenance

    Minimum data for a first version
    The monitoring data above with months of history, a work-order history that records failures and their causes, the same asset IDs in the sensor data and the maintenance system
    Can wait
    Spare-parts optimisation, automatic work-order creation
  • Line or site planning (what-if)

    Minimum data for a first version
    Routings, cycle times, layouts and capacities, production orders from the MES or ERP, a baseline period that operations agree is representative
    Can wait
    Live sensor feeds, energy modelling

Two data problems stop more twins than missing sensors. The first is identity: if the sensor platform, maintenance system and ERP name an asset differently, no model can join their data. Informatica notes that when asset records differ across systems, the replica becomes unreliable (Informatica, accessed 2026-11-19). The second is failure history: a predictive model needs failures recorded with causes, and many maintenance logs record only that a job was done.

How do the alternatives compare?

Option Choose it when Do not choose it when
Dashboard with alerts The decision is "is something wrong now?" and thresholds are known The decision needs a forecast or a what-if
Maintenance system with condition triggers Readings should open work orders automatically and planners already use a CMMS Failure causes are not recorded, so triggers cannot be tuned
One-off simulation study A layout or capacity decision is made once and the model can be discarded The same question returns every week
Digital twin All five filter questions pass and the model will be kept in step with the asset Data identity is unresolved or no team will act on the output

Ending on the first or second row is a good outcome: a CMMS with condition triggers or a dashboard on clean asset data is also the foundation a later twin needs.

How should a first digital twin be scoped?

  1. Name one decision and one asset group

    Write down the decision, its owner, its frequency and what a wrong call costs; pick the assets where that cost is highest.

  2. Audit the data before choosing a platform

    List each signal the decision needs, where it is stored, how much history exists and whether the asset IDs match across systems.

  3. Fix identity and history first

    Agree one asset register and record failures with causes.

  4. Build the loop at its simplest

    Ingest the signals, show the state, raise an alert or a forecast, and record what the team did. Keep the model as simple as the decision allows.

  5. Review against the decision, not the demo

    After an agreed period, check whether the decision changed and whether the change mattered. Expand to more assets only if it did.

  6. Plan the upkeep

    Sensors drift and processes change; name who keeps the model in step and review its security.

Where AI fits

AI is useful inside a twin where patterns are too many for fixed rules: spotting abnormal readings across many signals, estimating time to failure from history, and summarising alerts for the person on shift. Engineers still set thresholds with maintenance, check every model against recorded failures and decide what acts automatically. Netbase works with the major commercial and open-source AI tools and models, chosen per project.

What delivery record exists, and what does not

Netbase names AI with IoT among its AI capability areas, and IoT and mobile among the delivery trends it covers. Netbase is investing in IoT with AI at the edge; this is a direction, not a delivered result.

Netbase has delivered projects for manufacturing and logistics clients, who are not named. The anonymous manufacturing and logistics platform record describes an operations and integration platform built as consultant, custom developer and integrator, which is the kind of data foundation a twin depends on.

Netbase does not publish an IoT or digital twin case study, a payback period, a downtime figure or any result for this kind of work. This guide explains a filter; it is not a case study. Most Netbase projects are delivered on fixed-price contracts, and a first twin scoped with the steps above suits that model because its decision and asset group are fixed up front.

Alternatives: other ways to improve asset decisions

Not every asset problem needs software built for it. Selection criteria:

  • Off-the-shelf monitoring: when the assets are standard and a vendor platform reads them out of the box, buy it and integrate it.
  • Workflow automation: when the delay is in the paperwork around the asset rather than in the asset, score the workflows first with the manufacturing and logistics workflow automation guide.
  • Better stock data: when the physical asset is inventory, a custom WMS module or an off-the-shelf WMS usually pays back sooner than a twin of the building.
  • Guidance on the equipment: when people need instructions shown on the asset rather than a model of it, choose a browser or native AR path instead.

Limits of this guide

  • It is written for asset-heavy mid-market companies; city-scale and infrastructure twins have governance needs this filter does not cover.
  • Third-party payback periods are vendor estimates and vary by plant; treat them as context, not targets.
  • It does not choose a platform or a sensor; those follow from the data audit.

Plan the next step with a Netbase consultant

Frequently asked questions

It depends on the use case and on how actively the team acts on the output. Gradion, a vendor, estimates 12 to 18 months for predictive maintenance twins and 2 to 4 years for facility-level simulation, accessed 2026-11-19. A twin that changes no decision never pays back.

Not always. Many plants already capture the signals in controllers, historians or the MES; add sensors only for a signal the decision needs that nothing records today.

No. A dashboard shows current values. A twin also models how the asset behaves, so it can explain a state or forecast the next one. A dashboard is often the right first release and the base a later twin is built on.

Data that cannot be joined: the same asset carries different IDs in the sensor platform, the maintenance system and the ERP, and failures are logged without causes. Fixing that comes before any model.

Next step

If a twin has been proposed for your plant, fleet or site, tell us about the assets and the decision it should change and we will reply with a proposal for a data audit and a first scope.

IoT development with AI analytics that connects devices to the business 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
line
Machine learning and predictive AI tied to a business metric Machine learning and predictive AI tied to a business metric

Netbase builds machine learning models that predict something a business acts on: which products a customer will likely want, how demand will move, which orders or accounts need attention. Each model is tied to one agreed business metric, validated on your own data and monitored after launch. Beyond the recommendation engine delivered for 4over4, machine learning is a growing capability.

Learn More
line
Enterprise integration hub ready for AI agents: one monitored layer for commerce, ERP and CRM data 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
line
CMMS and maintenance management software: assets, work orders, spare parts and production on one record CMMS and maintenance management software: assets, work orders, spare parts and production on one record

A CMMS from Netbase is one system where maintenance teams plan preventive work, run work orders and keep spare parts in stock, linked to production schedules and purchasing. We design and build it around your assets, shifts and approval rules, connect it to sensors and the ERP, and add AI that predicts failures while engineers decide.

Learn More
line

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.

Contact Netbase

Discuss a project

Netbase JSC helps organizations design, build, modernize, and operate digital products and AI-enabled business systems.
Project enquiries

project@netbasejsc.com

WhatsApp

+84 937 869 689

Office address

91 Nguyen Chi Thanh, Dong Da, Hanoi, Vietnam

Get in touch

Tell us what you want to build, modernize, or operate.

Tell us what you want to build, modernize, or operate.

Contact Netbase