NB-IoT vs LoRaWAN for industrial projects in Colombia: when to choose each one

If you’re scoping an industrial IoT project in Colombia —well monitoring, tank telemetry, agricultural sensors, smart metering— the most expensive decision you will make is not the sensor, it’s the network. A wrong choice between NB-IoT and LoRaWAN can multiply operating cost by five, or leave devices offline in six months.

This guide summarizes when to choose each one, based on current Colombian coverage, real cost of ownership, and the cases where the difference is decisive.

TL;DR — what to choose by project type

Case	Recommendation
Few devices (<100), urban areas, no network ops	NB-IoT (Claro, Tigo, Movistar)
Many devices (>500), single site, cost-per-message control	Private LoRaWAN
Rural areas with cellular coverage	NB-IoT, with care on per-point coverage
Rural areas without cellular coverage	LoRaWAN (own gateway) or satellite
Regulatory compliance requiring operator SLA	NB-IoT (operator owns the network)
Frequent readings (>1/min) and large payload	NB-IoT
Low frequency (hourly or less), small payload	LoRaWAN

That table covers 80% of cases. If you have five minutes, read on.

The two technologies in one sentence

• NB-IoT (Narrowband-IoT) is a cellular network operated by Claro, Tigo and Movistar. You buy a SIM, pay per device/month, and forget about infrastructure.
• LoRaWAN is a low-power network you deploy yourself (gateway + network server) on the 915 MHz unlicensed band. No monthly per-device fee; you own the gateway TCO.

That alone defines the first axis: do you want to pay OPEX to the operator or invest CAPEX in your own network?

Real coverage in Colombia (2026)

The datasheet lies. Real coverage depends on operator and zone, and it shifts.

NB-IoT: Claro, Tigo and Movistar cover the main capitals and industrial corridors (Bogotá, Medellín, Cali, Cartagena, Barranquilla, the coffee region, Mamonal, Yumbo). Outside that, run a point-check pilot before committing hundreds of devices. We’ve seen sites 30 km from a major city where NB-IoT won’t attach due to terrain blockage, even though 4G LTE works.

LoRaWAN: there is no reliable public Colombian network covering the country (no equivalent to The Things Network with national coverage). In practice, LoRaWAN in Colombia is operated as a private network: install 1-3 gateways on your site and cover the field.

Practical rule: for a rural project without infrastructure, always run a 2-week coverage pilot with real sensors before ordering the full batch.

Cost: the calculation almost nobody does well

Compare 3-year TCO, not sensor price.

NB-IoT — realistic example for 100 devices

• Sensor with NB-IoT module: USD 60–120 each
• Colombian operator IoT plan: ~USD 0.50–1.50 per device/month
• Cloud server: depends on provider (AWS IoT Core, Azure IoT Hub)
• 3-year TCO, 100 devices: sensors ~USD 8,000 + connectivity ~USD 1,800–5,400 + cloud ~USD 600/year

LoRaWAN — same scenario, 100 devices on one site

• LoRa sensor: USD 40–90 each
• Industrial LoRaWAN gateway: USD 600–1,500 (one is usually enough)
• Network server (open-source ChirpStack or The Things Stack): self-hosted or ~USD 1–2 per device/month
• 3-year TCO, 100 devices: sensors ~USD 6,000 + gateway ~USD 1,000 + server ~USD 600

At small scale they look similar. The gap explodes as projects grow: 1,000 devices on NB-IoT means 1,000 SIMs billed forever; on LoRaWAN, the marginal cost of device 1,001 is just the sensor.

Battery life

Both technologies aim for years of battery operation, but reality:

• LoRaWAN: 3–7 years typical with hourly uplinks. Extremely low power.
• NB-IoT: 2–5 years typical. Good with PSM/eDRX well configured; very bad with marginal coverage because the module burns energy searching for the network.

If the device will sit where cellular signal is borderline, disqualify NB-IoT or budget for more frequent battery swaps.

Case by case

Oil & Gas: tank and remote well monitoring

Recommendation: NB-IoT for urban service stations; LoRaWAN or satellite for remote fields.

A service station in Cartagena has NB-IoT coverage and the SIM cost dissolves into the margin. A field in Casanare may lack reliable coverage from any operator. There, private LoRaWAN with a solar-powered gateway wins — and for the most critical points, satellite as backup. If the sensor is in an ATEX zone, the certified hardware catalog also constrains the choice.

Energy and utilities: smart metering and substations

Recommendation: NB-IoT for residential/commercial meters; mixed for substations.

Mass smart metering in populated areas is the canonical NB-IoT use case — the SIM cost dissolves across thousands of meters and the operator owns the SLA. For distributed substations, the balance depends on whether the substation already has an IP link (usually yes): if it does, monitoring goes over that link and NB-IoT/LoRa is only a fallback.

Agribusiness: field sensors, silos, irrigation

Recommendation: LoRaWAN, almost always.

A farm covers hundreds of hectares with unpredictable cellular coverage. Installing a LoRaWAN gateway at the highest point covers the whole property and removes recurring SIM costs. This is where LoRaWAN clearly wins: cheap devices, mass deployment, few daily readings, no operator SLA needed.

Logistics and fleets

Recommendation: NB-IoT or LTE-M, not LoRaWAN.

Assets move. If the device is not fixed, it needs a public network. Private LoRaWAN stops making sense the moment the container leaves your property. NB-IoT with roaming or LTE Cat M1 are the realistic options.

Data centers and technical rooms

Recommendation: neither NB-IoT nor LoRaWAN — use the IP network you already have.

The most common mistake: pushing LPWAN where Ethernet or industrial Wi-Fi already covers the space. If you’re monitoring HVAC and UPS inside a building with its own network, connect over Modbus TCP / SNMP on IP. LPWAN is meant for distributed assets without infrastructure, not to replace existing cabling.

Typical mistakes we avoid on real projects

1. Ordering 500 sensors without a coverage pilot. If NB-IoT fails to attach on 50 of them at install time, you already burned the budget.
2. Assuming LoRaWAN is free. The gateway, network server and platform integration do cost.
3. Forcing LoRaWAN for large payloads. 915 MHz duty cycle caps how much each device can transmit per hour. Need to push lots of data? NB-IoT.
4. No backup plan. Any network can fail. Edge buffering on gateway or sensor prevents data loss.
5. Forgetting platform integration. Your SCADA, ERP or BMS does not speak LoRa or NB-IoT — someone has to translate.

How we decide at COL0

On every project we follow this order:

1. Operator coverage map for the exact points where devices will sit. No valid decision without this.
2. Realistic 3-year device count (not 6 months) to project TCO.
3. Sample rate and payload size — decides if LoRa fits.
4. Fixed site or moving assets? Decides whether private LoRa makes sense.
5. Regulatory requirements — if the client needs operator SLA, NB-IoT wins automatically.

If your project doesn’t fit cleanly into one tech, the answer is usually a hybrid architecture: gateway with private LoRa for the bulk of fixed sensors, NB-IoT as backup or for moving assets. We’ve shipped this several times and it scales best.

---

Have an industrial IoT project stuck on this decision? Get in touch or chat with us on WhatsApp. We’ll tell you on one call whether NB-IoT, LoRaWAN or something else is right for your case.