← Blog

Unlocking The Power of Video Data: IBC Interview with Vodafone’s Klaus Heimannsfeld

At IBC 2023, Till Sudworth, CMO, member of the management board, and Head of BU Video of NPAW, chats with Klaus Heimannsfeld, Head of TV Data & Analytics, TAD-TV Architecture and Development at Vodafone. Heimannsfeld explains the role of his department within the company, the opportunities and challenges of working with video data, and how NPAW’s Video Analytics is helping Vodafone achieve its streaming goals.

Maybe you can explain to our customers or audience what exactly is your task, your job at Vodafone. 

Yeah, we are TV architecture and development, which is a group inside Vodafone group, and we are developing the TV platform for all the Vodafone markets. I myself have a new team We are concentrating on data analytics and insights for the TV platform.

Okay, so if you’re concentrating on data video analytics, what are the main challenges within these tasks that you are currently facing at Vodafone? That’s a difficult question. There’s lots of challenges. So the first challenge that we have is always data privacy. So, you know, keeping the data of our customers safe and making sure that we only use it for the intended users is very important to us. So we typically always aggregate the data, anonymize the data and then work on that kind of data for the analysis.

And the second topic is, of course, the observability. You know, we really want to understand what our customers’ experience is, okay? And the third one is, once we understand what’s happening at the customer, the challenge is always to act on it, generating insights. So that never works with the TV data alone, but it typically always means that we need to correlate with network data or other data that we have in the company.

And how does NPAW’s analytics help you with these tasks?

Well, NPAW helps us. Let’s start with the first one, data privacy. There are lots of features that make data privacy implementation very easy. We can configure which data we anonymize. We can configure which data will be deleted on reception. Or actually also later We can configure retention times. So that’s very helpful, makes things transparent for us. And yeah, that makes our life much easier when it comes to data privacy.

The second thing is we do serve a lot of different stakeholders. So we do our own analytics, our own insights, but we’re also like the service provider for all the other departments. So might it be network quality, might it be customer support, might it be customer fulfillment. So there’s a whole bunch of people working on the data. And Nice People at Work gives us a very good user interface that can be basically configured, I would say, to answer the question of those stakeholders in a very simple, non-technical way.

What are the most important KPIs that you track? Would you say there’s one single specific KPI where you say it’s by far the most important? Or is it a bunch of KPIs?

In principle, it’s all the KPIs that we collect. Of course, I mean, basically the three top ones that are the most for me, interesting are, of course, the happiness score because it gives you a good idea of what the customer experiences. And we’ve done comparisons with our NPS values and the CSAT scores, and we know that it quite closely matches. So we can use the happiness score to actually identify customers with issues and proactively address those customers.

I think the second most important one, you probably wouldn’t guess that, it’s for me the throughput and the profile changes. So the throughput is always a good indication of the quality of the network connection of the customer. And it’s like having millions of probes in our network additionally to the probes that we already have. So this is also very valuable for our network people to optimize the network and to fix incidents as soon as they occur.

The third one, as I mentioned, is profile changes because profile changes in the AVR streaming gives me an idea of the stability of the connection. And that could indicate, for example, very typically in-home problems with the customer Wi-Fi or something like that. So if you see a lot of profile switches, you basically have an idea where to search.

Very interesting. You mentioned before the customizable way of adapting your needs in terms of GDPR. How important is that for analytics tools, that they are really customizable? And how important is it for you and how strongly do you use that customization of NPAW?

OK, let me be clear about customization because there might be a misunderstanding. So customization in terms of the product from Nice People at Work is something that we don’t want to have. Because we think that the feature set itself is great and it gives us most of the things that we want. When you start customizing a product for a specific customer, it makes everything more expensive, more complicated, takes more time.

Configuration or customization, as you probably think about it, building your own dashboards, basically developing specific basically under specific requirements is, for me, one of the key differentiator of nice people at work. Because what typically happens is that somebody comes along and says, yeah, I would need to know how was the distribution, for example, in the women’s soccer world championship in Germany? Which region watched the most? You know…And those things, to develop them in a traditional way, takes a long time.

With NPAW, we just simply go in the data, we create a dashboard, and we typically take half an hour an hour to answer such a question. So it’s very important for me, it’s the key differentiator, that we’re allowed to build our own dashboards and, even more, that this can be done in a non-technical way.


When it comes to data in the tech environment, how important is it for you to have one single source of truth for a lot of various different KPIs and data? Is that something where you say it takes a lot of work off your shoulders Or is it something that you don’t need?

It would be great to have it. But it’s an organizational challenge. Because we collect data at lots of different points, we have specific tools for specific features. So there’s always the component of integrating data.


How important is it for you that several departments of Vodafone speak the same language in terms of that the one tool is used by a lot of different departments Department A says something different than Department B because they’re looking at different data. How important is that for you?

It’s very important. So building a data dictionary, having a clear definition of how you measure and what the measurements mean is key to avoiding conflicts between departments. Because you can take the same measurements, actually the same data collected, and you can interpret it in different ways. And the results will be totally different. So it’s really important for everyone in our organization to understand what we’re collecting and what it means. You know…

And to achieve that, we typically do a lot of training. And I guess it also divides a little bit, like silo thinking. If everybody has different data, it would be more challenging for departments. So what we’re actually doing, besides my team, which is now specialized in the TV area, we’re also actively building cross-functional teams with everyone with product operations, network quality, so that we discuss that. And one of the challenges is really also building dashboards and reporting and insights that are agreed between the departments. So that’s also very important.


How strongly do you share your insights in your data with external providers? I mean, do some of your external providers have access to your data?

No. We don’t share any production data, so it’s always kept very tight. Even inside Vodafone, we use a very restrictive system to actually allow people and actually only allow people access to the data that have need for that. At the same time, for those people, we try to make it as easy as possible to get to the data.


You mentioned at the very beginning that it’s super important to have actionable insights, right? To have data that you actually can use and work with. How do you implement and filter data so that you do have actionable insights? How do you work with the data that you get some insights out of it?

Yeah, that’s a difficult question. Yeah, it’s really a lot with experimentation and generating ideas and then quickly trying to validate assumptions. I can give you one example. So one of the things that we did was we matched basically the happiness scores with some more detailed Wi-Fi data we get from another tool. And the interesting outcome was the question, would we recommend a customer to use 2.4 gigahertz or 5 gigahertz in their Wi-Fi configuration?

And my immediate reaction would have been, yeah, 5 gigahertz because I have more bandwidth. It has less interference. The data proves that this is wrong. the data proves that that the coverage of 5G in the 5 gigahertz Wi-Fi in our customers is typically worse than 2.4. And we see from the quality data that we have that we would basically, if customers cannot connect through cable to fix that, we would always recommend at the moment using 2.4 gigahertz. Of course, if you have the coverage, you can go to 5. But in general, it seems that 2.4 works better.

And this is kind of the thing, where you have an idea, you look at the data, you play around, and then you generate the insights. And I think the important thing is that the tool allows you to do that quickly. You don’t want to wait another day or two days for development.


Very interesting. How did you do all that before you had NPAW? Is that something that you did by your own and internally generate data? Or how did you work with data before you implemented our solution?

We do collect a lot of different data. We still do that. But we never collected anything that was directly related to video distribution. And that was a problem because we didn’t have the visibility. So the visibility that we had was really down to manual testing and to testing with test robots, which can give you an idea of the video quality as well. But then the problem is this is always just for test setups. So it’s not really for all the customers. So you don’t really get good, I mean, you get an understanding of how well your solution performs in the perfect setting. But you don’t have an idea how it performs out in the field. Yeah, it’s only a subset of the entire customer base.


Klaus, as you are a super expert in the industry, one last question would be interesting for me. If you look into the future, let’s say like two years, because it’s two years a long time in the digital world, how do you see the video streaming market will evolve and develop? Especially what do you think will be the role of broadcasters, telco operators, like Vodafone, or other OTT players? 

That’s an interesting question. It’s a million dollar question. To be honest, for a second, let me take my personal perspective as a consumer. I’m really fed up with all the streaming services. Because the content is so fragmented, it’s so distributed, that we, as an industry, we have to find a solution to overcome that. I mean, every small content producer introduces its own streaming platform, and it’s a mess for the customers, okay?

So I think there will be a consolidation, especially on the streaming platforms, to get more aggregated offerings. We’re trying to do that on a device basis, integrating all the streaming applications on our set of boxes, but that’s a costly and difficult proposition. So maybe, if take a guess, the best thing for our customers would be in the coming years when maybe the telcos concentrate really on optimizing the delivery of the TV services, and work together with the streaming platforms to aggregate the content. That would be my wish for the future.


Related Articles