New business models for industrial IoT startups
Internet of Things

New business models for industrial IoT startups

Across the globe, tiny sensors are silently changing the way supply chain and manufacturing works. Industrial Internet of Things (IoT) devices are creating value everywhere, but for the startups making them, one challenge is building a business model around connected hardware.

Consumer IoT startups rely on selling hardware at massive scale, but in the industrial sector, where equipment is expensive and sales happen less often, companies providing the sensors must consider other ways to grow revenue.

Petasense, a hardware startup which exited stealth mode in June 2017, is blending complementary revenue streams to build a business model around its connected industrial sensors.

The company produces tiny sensors called Motes that attach to industrial equipment and measure the frequency of their vibrations. It uses these sensors to solve a problem that can cost industrial companies hundreds of thousands of dollars: damaged machinery and production downtime stemming from poor maintenance.

“We think of ourselves not just as a hardware product but as an integrated end-user experience combining hardware and software,” explains co-founder Arun Santhebennur. “A lot of the intellectual property and the end customer value comes from analyzing the data that our sensors generate.”

The Motes feed their IoT data wirelessly to the company’s cloud-based software, which then applies machine-learning algorithms to identify vibrational patterns and infer what’s happening to the equipment.

“Vibration is a little like speech recognition. You’re looking for certain patterns which you then transform into a frequency domain,” Santhebennur explains.

Vibrational frequencies match known patterns that indicate specific problems. One frequency profile may point to a bearing that will shortly fail, for example. Another could indicate a gear that is beginning to slip. Knowing one from the other gives companies the warning they need to target an issue before it becomes a problem.

The company currently makes two varieties of its devices. One, selling for $399, identifies late-stage defects, while a $599 version can detect flaws earlier.

Petasense makes an undisclosed margin on these sensors, selling them in packs of between 50 and 250 devices. On average, customers typically install five on a single industrial machine to measure vibrations across different parts.

Selling hardware generates some welcome upfront capital, but subscription-based revenues will eventually outstrip sales. Companies pay a monthly fee for every Mote they use to analyze the vibration data it produces to make maintenance predictions in the cloud.

Customers pay a $10 per month fee for each sensor, assuming they buy enough upfront. That sees the subscription-based revenue surpass the device’s original cost in 3.3 years, and Motes operate far longer than that. “The lifetime value comes from software,” Santhebennur says.

Another proposed business model involves aggregating and selling IoT data to third parties, but this can be problematic, warns Duncan Stewart, director of research for technology, media and telecommunications at Deloitte.

“I have talked to hundreds of companies in the last five years,” he says, adding that they often point to the value of their amassed IoT data, but admit that they haven’t made money from selling it to third parties. “Not a lot of people seem to actually be making coin from it.”

Privacy and competitive advantage is a concern for many industrial firms. Many of Petasense’s customers want their data used only for training their own equipment, explains Santhebennur. This brings another advantage, by making Petasense increasingly valuable to the customer over time as it trains its analytics model on more of the customer’s specific IoT data.

“Especially as you uncover real problems, failure data is also training those models. It’s a very rich data set that was not there before,” he says. “It becomes extremely valuable.”

Once a company’s machinery is outfitted with vibration sensors, how does the technology provider expand its revenue from that customer? Santhebennur sees two opportunities.

The first involves bringing predictive maintenance to a whole class of smaller companies that couldn’t access that data before. The second will see Petasense launch other types of sensors to measure equipment in different ways.

“There are everything from valves, steam traps and electrical panels. There are all kinds of machines that a maintenance organization is responsible for,” Santhebennur says.

Advice for hardware startups

His advice to other industrial IoT companies trying to monetize hardware is to plan several hardware cycles ahead. “Getting hardware right takes several iterations, so plan on having several versions of the product before it becomes commercially available,” he says.

Santhebennur also warns that cloud-based sales are challenging for industrial customers not used to buying solutions under that model. “That’s the change you need to bring. It’s a matter of them getting more comfortable and seeing the benefits.”

That comes from helping industrial customers to quantify the potential ROI from the subscription-based service. “What’s your unanticipated repair cost?” he asks. “What’s the cost of keeping spares? They quickly get the business value of predictive maintenance.”

As companies continue to mix service and capital sales models, Stewart sees processor technology pushing the models still further, moving intelligence from the cloud to the edge.

While Petasense’s business focuses on detecting trends in equipment performance, Stewart envisages machine learning models embedded directly in sensors. These could expand service offerings into new applications such as real-time fault detection.

“Their advantage is that they’re low latency. If I can put machine learning on a device, I can stop a turbine whirling in a thousandth of a second,” he says, adding that this wouldn’t be possible if you sent data to the cloud and back for analysis. “That round-trip latency could be more than enough for a turbine to tear itself to pieces.”

As the technology underpinning the IoT continues to evolve, no doubt the business models will, too. In the meantime, companies like Petasense have their work cut out for them in persuading customers to embrace a new, cloud-centric way of working with all the security and privacy concerns that brings. Success guarantees a recurring revenue model for many years to come.

Related Stories