Industry research consistently shows 70-75% of IoT projects fail to reach scale. Not because of bad ideas — but because the gap between "working prototype" and "industrial production" is wider than most teams realise. Here are the five recurring failure modes we see, and how we work around them.
1. Underestimating supply chain complexity
The prototype runs on 10 development boards bought on a hobbyist site. The product needs 50,000 units, custom PCBs, sourced components in stock for 5+ years, with second sources for critical parts. The transition kills more projects than any technical issue.
What we do: design for industrial sourcing from day one. Component selection considers availability, lifecycle, MOQs, lead times, alternatives. Hardware is built around qualified parts, not exotic dev modules.
2. Ignoring field reality during prototyping
The POC works perfectly on the engineer's desk. In production, devices face vibration, temperature swings, RF interference, unreliable power, accidental disconnections, end-user abuse. Half the bugs only surface at scale.
What we do: simulate field conditions early. HALT/HASS testing, EMC pre-screening, drop tests, climate chambers. The pilot phase deploys hundreds of units on real sites — not just labs.
3. Treating security as a final-step concern
Encryption added last week. PKI infrastructure improvised. Devices that can't receive secure firmware updates. Default passwords still in production. Mirai-style botnets. Bricked fleets. Reputational damage.
What we do: security architected from day one. Hardware-rooted trust (secure elements, TPM), PKI from sandbox, signed OTA updates, threat modelling done before the first line of code is written.
4. Building cloud platforms without operational discipline
The cloud works for 1,000 devices. Then you scale to 100,000 and the database collapses, the message queue saturates, the costs explode, and no one is on call when devices stop reporting on a Saturday night.
What we do: design cloud platforms with SLO/SLA targets from the start. Capacity planning, observability, alerting, runbooks, on-call rotations — same discipline as a SaaS startup, applied to your IoT backend.
5. No clear ownership of the end-to-end system
Hardware vendor blames firmware team. Firmware team blames cloud provider. Cloud provider blames operator. Issues take weeks to triage. The customer just sees a broken product. This is the most fatal failure mode of all.
What we do: we are the single accountable team across the entire stack. One throat to choke, as the saying goes. We coordinate sub-suppliers internally so you don't have to.
Going from prototype to production?
If you recognise any of these failure patterns in your own project, let's talk. A 45-minute exploratory call is free.
Talk to engineering →