“You can’t manage what you don’t measure!”
This quote from Peter Drucker is now more true than ever before. Big data, machine learning and a lot of other buzzwords are in everyone’s mouth and are brought up in many situations. I am not even sure if people really know what they are actually discussing. What is true is that the important questions out there are not solved yet and a lot of industrial companies are not ready to use the full potential of their data or do not even have the data in a structured way. This really becomes evident when talking to industrial companies and is proven by a study of Commerzbank according to which only 8% of German SMEs are already collecting, analysing and using its data systematically.
But the topic is on the agenda of almost everyone (97.2% of executives are eg. investing in Big Data and AI — Link)– because it should be a core capability of every company and is relevant for every other aspect of the business as well. Being in possession of and understanding your own data, and making smarter decisions based on this data is a crucial element of a successful company in the 21st century. This also opens up a lot of opportunities for industrial companies by making processes more efficient and more predictable, thereby saving cost or enabling new service models that would not have been possible in the past.
Let’s put this a bit into a structured format — I really like the overview McKinsey did and want to focus on two aspects of Big Data: the data itself and the analytics part. The latter does not work without the data part.
And the big question that industrial companies should ask themselves is: how can data bring value to the table and solve real problems?
Source: McKinsey & Company
In our view, these two pillars highlighted by McKinsey can also be illustrated as a three-way pyramid to get to real value creation. All those steps along this pyramid are opening up a lot of potential collaboration opportunities between startups and industrial companies (of course, we have to say that), although most startups are not yet able to work with a lot of industrial companies as their datasets are not suitable or not existing at all. But this is changing fast, also among the Mittelstand.
Let’s have a look at those three steps in more detail and how we see this topic at Speedinvest. In a lot of startups all those 3 steps are intertwined into one product.
1. Collect the data
In this step, data gets generated and collected via different sensors (RFID, BLE, UWB, etc.), Industrial IOT applications, ERP data, machinery data and more. New kind of sensors in combination with increasing computing power make it cheaper than ever before to track both at larger distances and also shorter distances, eg. within a factory floor. As a consequence, the cost reduction of sensors and computing power in recent years has been enabling the collection of data and industrial IOT applications that have not been feasible until recently. However, collecting data with sensors is not completely new, different parts of the processes have already been tracking data for a longer time, and even manual processes are supported with software and provide data. So, there’s a problem coming up which you can summarize with: you do not see the forest for the trees. There is just so much data available that you need a clear data strategy — especially when working with partners, eg. startups.
This leads me to the relevant questions for you when thinking about working with startups in that field:
- How do they track data?
- What technology does fit to my business (eg. Automotive vs. FMCG)?
- How reliable and scalable is the technology? From a hardware and software perspective.
- Which data do I want to track? You need to have a clear goal in mind already beforehand.
- What data is already available within my company?
- How combinable is the technology with my existing systems?
- What do I actually need to do in order to collaborate properly?
2. Make the data usable
The problem is that in most cases the data from step one is not structured in a way that it can be used immediately. These datasets normally have to be structured and enriched to be actually usable in a meaningful way. Especially working together with startups requires this data preparation step, as most of them are working with what is available and not preparing what is not there.
Part two is sometimes difficult due to the fact that of course eg. machinery producers and others want to implement their own standards (as becoming a digital service provider is an interesting growth opportunity for them). Therefore, there is no market standard for machine protocols and this makes it difficult to combine different standards and machines. Creating eg. a European standard would help solving this issue.
So, the big unsolved puzzle is how to make data usable across different machineries or whole factories, and how to work across different protocols and datasets. At the same time this opens up opportunities for software providers and startups, because there will be one or more companies solving this problem and become the middleware of industrial IOT applications. Those solutions then also enable the use of third parties for the last step — the intelligent insights or also just the visualization and reporting of data.
The relevant questions for you when thinking about working with startups in that field are:
- What data is already available and how structured is my data?
- How do we prepare the data and make it useful?
- Which platforms do I use? Get an overview of the IT stack in a complex industrial environment.
- What are the technical specifications for each of them?
- How easy is it for others to integrate with the existing systems? Can they use existing APIs, etc.?
3. Use the data for intelligent insights
The last step is the sexy one everybody talks about. Let’s assume you’ve mastered step one and two of this process and now you really want to work on specific use-cases. This is where data enrichment is relevant and where still a lot of human intelligence that understands the respective industry is necessary. Furthermore, it is a prerequisite to hire the right analytics talent that can work on these problems effectively, because, these people also serve as an interface between data analytics and business. Without understanding the business side, the data analytics side won’t add any value.
With the data you get from the first two steps you can work on real problems like the optimization of delivery times, predictive maintenance, demand planning etc. by adding an intelligence layer on top.
The next logical step then would be to adapt and design processes according to those new insights and therefore improving them massively. In a perfect world inputs are automatically pushing actions eg. asset tracking of goods in internal warehousing is leading to automated orders and pushing them through into SAP via an API.
At the same time, you can also start building new business models on top of those insights and the data gathered over the process. This can eg. be digital service models.
What are the relevant questions for you when thinking about working with startups in that field:
- What is the insight I want to get? This can be very different eg. reducing downtime or predictive maintenance.
- What are my use-cases?
- What are my top use-cases I want to improve?
- Which pain point does it solve?
- What is the value add it brings? Measurable results eg. after a POC.
- Do I have the right people for it? If no, what people do I have to hire?
Don’t forget about data security & privacy
What needs special highlighting is the fact that along all the steps mentioned above, data security and also data privacy plays a crucial part. This also gives a lot of startups the opportunity to work with industrial companies as for them data security, data privacy and data ownership is a and will always be a critical element.
Thus, it is also important to ask the following questions:
- Is my data GDPR conform?
- What do I have to consider when operating internationally and also when cooperating with a startup?
- Where is the data stored, processed etc.?
- Who owns the data?
- What do I have to do to ensure that and how to create awareness for data security and privacy within my company?
Let’s see which startups will tackle these issues successfully and if Europe’s industrial companies will manage to digitize their processes and business fast enough. But there are already very good signs that we are on the right track as more and more interesting companies popping up in Industrial Tech. So let’s stay tuned!