Transforma logo

Why IoT is not ‘all about the data’

JUL 14, 2022 | Matt Hatton
region: ALL vertical: ALL Internet of ThingsArtificial IntelligenceEdge Computing

We industry analysts are probably more exposed than most to cliches surrounding technology. Things like “people buy solutions not technology”, “digital transformation is not about technology it’s about people and processes” or “think big, start small, move fast” . So far, so banal, but mostly harmless. There is, however, one which is actually actively unhelpful: “IoT is all about the data”. Here’s some good reasons why that’s not the case, and thinking that it is might be bad for your project.

The main role of most IoT devices is not to produce data

The main function of most IoT devices is to act as a trigger, sending alerts that are of the most basic kind, for instance if a manhole cover is moved or the temperature of a refrigerated container rises above a particular level, or a baby monitor is activated. Of course, this is data, even the binary alarm trigger. But it’s not data in the meaning of something that can be stored and analysed. It’s here-today-gone-tomorrow (or more accurately here-now-gone-in-seconds) stimulus for other things.

In quite a few cases it doesn’t even really generate much data of its own, being used as a mechanism for controlling a device. Connected traffic lights for instance don’t gather much data. They are more there to control the flow of traffic. Basic electronic shelf labels are another great example. There are also instances where further processing of data wouldn’t be appropriate or desirable, for instance in the case of child trackers.

In plenty of IoT use cases there will also be the opportunity to gain some benefit from exhaust data, for instance a washing machine manufacturer adapting its design to reflect usage patterns, or a car manufacturer trading aggregated vehicle position information on a data exchange. The clue here, however, is in the fact that this is exhaust data. It’s a corollary to the main reason for deploying the devices, i.e. to do something.

Media consumption isn’t there to be analysed

Another big category of IoT devices involves consumer media consumption. That might be connected car in-vehicle infotainment, media players, connected TVs, connected cameras and much more besides. In the same way that shelf labels or traffic lights are recipients of largely one-way traffic, so too are media devices. Sure, someone somewhere is probably tracking the media consumption of a particular user, but that is an extremely marginal element of the utility associated with, for instance, a connected TV.

These devices account for around 30% of all IoT devices, and certainly a far larger proportion of ‘data’. But this isn’t data that’s there to be analysed.

Added up, less than half of IoT devices generate meaningful ‘data’

Based on a segmentation of IoT use cases in Transforma Insights’ IoT Forecast Database, we estimate that less than half (46%) of IoT devices really generate data that is worth analysing. The remainder is either not really generating data that can or should be analysed (15%) or deals with the consumption of media (39%). The proportion creating analysable data will creep up over our forecast period to 51%.

And it’s worth bearing in mind that the half of IoT devices producing data that can be analysed isn’t necessarily going to deliver a huge amount of value through being analysed. Take the cases of applications like smart watches, home weather stations, printers, ATMs, card payment terminals, assisted living solutions, bike sharing schemes, industrial monitoring or any number of other applications. They all produce analysable data, but it’s a moot point how valuable it will actually be to do so. It’s also questionable how valuable the data is for analytics, given that there may well be substitutable data that can also be used for the same purpose. To be truly valuable, IoT data needs to be unique, and often it isn’t. For the most part their value is also ostensibly in the main use case, not in some nebulous data analytics layered onto them.

Automation negates the need for ‘data’

The increasing prevalence of AI and edge computing, means that much data will not make it out of the device at all, meaning no potential for exhaust-data analysis off-device. In an effort to speed up processing time, many IoT deployments will increasingly rely on AI (and other data processing) performed on the device itself with minimal requirement for delivering large volumes of data back to a server. This also, of course, helps keep down the cost of connectivity where a device relies on a public network to connect, e.g. a mobile network.

With the advent of edge computing and AI, IoT devices are increasingly becoming closed-loop systems where the device receives input from sensors, processes information based on certain rules (potentially using machine learning) and acts accordingly. In most perfectly functioning IoT systems there will be very little ‘data’.

Why is it actively unhelpful to think the value lies in the data?

Why would it be a problem for organisations to believe that the value of IoT lies overwhelmingly in the data? The main issue is that it will potentially hold back deployment. The implication is that success in IoT depends on actively doing something with data, in terms of ingestion, storage, management and monetisation. This can be quite intimidating and may well delay deployments.

The obsession with data may also result in adopters making wrong choices about technologies. If you believe that all the value of IoT is in the data you would tend to opt for delivering all data from your IoT device back into the cloud (or similar). This is inefficient in many ways. It’s costly in data traffic and storage charges and leads to much higher latency in applications, meaning probably a lower performance.

It also encourages the adoption of sub-optimal technologies. We have previously discussed the benefits of ‘Thin IoT’, variously networking technologies, protocols, operating systems and so forth that are aimed at optimally supporting IoT deployments in constrained environments. These tend to be cheap and effective ways of connecting IoT devices. But seeking to process and/or transport much larger volumes of data means a different choice of technologies, overwhelmingly more expensive ones. It also generates more data to be processed and stored (and backed-up) ‘in the cloud’, which is not good from a sustainability perspective.

In most IoT use cases, the majority of the value derives from the mere act of connecting the thing. Doing that cheaply and efficiently has a much better ROI than trying to find mechanisms for monetising large volumes of data that can only be harvested using technologies that push up the price of the deployment.

All Blog Posts