Why We Don’t Use Device Meters
When the internet first arrived, meters – small bits of software installed on a user’s machine – were conceived as a way in which to track the entirety of an individual’s online behaviors. In those days, when the vast majority of the global internet population were sitting inside the United States on a PC desktop using Internet Explorer or Netscape browser, these tools provided a logical solution for publishers and media buyers to get an accurate view of their audience and reach.
When it comes to today’s global internet population, considering the complex nature of modern online behaviors, passive metering software has become utterly ineffective at helping you to truly understand your audience.
At GlobalWebIndex, we don’t incorporate any metered data – a strategic decision that continues to raise questions. Yet, having revisited this a number of times, the same fundamental barriers remain in place. Here’s whyâ€¦
They are impossible to scale.
Building a genuine opt-in metered panel is impossible to scale, and in reality, there are simply not enough people out there willing to share their online activity with third parties. Even more of a challenge is ensuring that they remain opted in for a year, making it impossible to build up a longitudinal annualized data set.
Back when the majority of the world’s internet population resided in the United States, this proved less of a challenge, but for today’s global internet population, massive scale is absolutely imperative to provide accurate measurement.
Even if you did manage to find a number of users who are willing to opt in, due to the level of data you are gathering, the costs of building a metered panel are utterly prohibitive. According to our sources, the typical cost to get a person to install one is around $50 and around $7.50 a month in incentives. If you then hope to measure across devices, this cost scales significantly beyond that.
This poses a challenge in a single market, but for us to implement on a global scale, the costs are beyond any business sense. In order to match our 350,000 annual interviews, this would cost a minimum of $49m annually (yes you read that right), assuming 100% retention and only measuring a single device. Incorporating realistic churn of three months and measurement across three devices, this scales to a ludicrous $164m annually – before any technology costs are added.
A new entrant recently made the claim that they had a measurement panel with 50m opt-in consumers. Using basic maths, you can assume this to be either untrue, or that the consumers in question here are not truly opted in. This is where metered measurement becomes spyware, and the cost for a 50m genuinely opted in panel is an estimated (wait for itâ€¦.) $7.9bn dollars a year! Even if you were to cut the incentives to $10 (a highly unfair incentive for the level of data shared), it would still equate to an annual cost of $500m, a steep price for the NSA or Google, let alone a start-up with series A funding.
If you want to do things ethically, with genuine consumer participation and a fair value exchange, the maths simply do not add up.
They raise serious privacy concerns.
There are two core privacy concerns when it comes to device meters… Firstly, it is questionable whether the most â€˜opted in’ users fully understand what they have opted in for. It has been well documented and reported in a number of class action legal cases that some past providers had been bundling meters in with other downloads. For example, you might opt in to download the latest anti-virus software and – having failed to read the small print – unwittingly opt in for the meter at the same time. This, while technically classed as an â€˜opt-in’, is highly questionable from an ethical standpoint.
As the world becomes increasingly privacy-conscious and consumer power continues to rise, it’s clear that this approach will no longer hold up.
Secondly, it is absolutely apparent that genuinely opted in consumers are unaware of the level of data they are sharing. Meters track everything from online banking and credit card payments to private browsing and private online chat. This is easily matched to third party data, particularly in the U.S., where there is a vast pool of this Personally Identifiable Information (PII) available and open to misuse.
They don’t capture online behavior accurately.
When metered site data first emerged in the late 1990s, it made sense given the somewhat predictable behavior of internet users.
Today, people utilise an average of three internet devices and seven apps monthly and it is not unusual for them to use two or more browsers.
Linking all of these devices, browsers and apps accurately to one individual is impossible through the use of these meters. This means the devices measured are not providing a complete and accurate measurement of the websites and content that panellists use and consume. This would be acceptable if we knew the level of accuracy, but the core issue here is that we don’t. We could be capturing 5% of internet behavior or 95%… nobody has any idea.
This is a challenge that is only going to escalate as an increasing number of our devices get connected – from the fridge to the TV to the games console and the car.
They can’t measure distributed reach.
Today, content is portable and it is likely that the majority of those consuming a publisher’s web content do not do so on their own websites. This was recently flagged by Business Insider who were increasingly frustrated with the data they were getting from metered measurement, which represented less than a third of their total audience as it only measured PC visits to their site. They estimate (and we have too) that they have over 200 million additional users across Mobile Apps, Facebook, Flipboard, Apple News and all the other platforms through which they now distribute content.
Distributed content is the future of publishing, but meters are unable to measure it accurately.
They measure devices, not people.
Device sharing is continuously on the rise, with 33% of internet users sharing their mobile devices, while for PCs this number rises to 51%. In a world where audience-centric marketing is king, this is a massive issue which is hugely impacting the quality of third party data, to the point where gender and age data is well under 50% accurate. As a case in point, I recently checked the third party data recorded against my machine and I was classed as both male and female simultaneously.
The issue for meters is even more pressing. Meters work under the core philosophy that a device is a person. As outlined in my previous post, Why the Future of Web Analytics is People, this is no longer true and with video and lean back content becoming more and more central to the internet experience, this is becoming all the more evident.
They don’t provide the answers you need.
Even if you could fix the many issues outlined above and deliver accurate metered data at scale, you still face the fundamental issue of need. Mirroring changes in digital behavior, the marketing world has transformed significantly. In the late 90s, marketers allocating spend to â€˜online’ would most likely have spent this on banner ads. To execute this spend, you might have referred to your sales rep from Portal Y and negotiated a rate based on the latest industry measurement numbers sourced from metered software. While this was a digital buy, the way of doing it was entirely analogue.
Today, the opportunities and process for online marketing spend are much broader and far more complex. Firstly, you have to determine whether you are going to invest in, not just paid advertising, but social and content. This demands far more rigor, with strategy that needs to be built firmly on audience insight.
If you decide to spend money on online advertising today, you’re not likely to ring up a sales rep and book a homepage takeover. Instead, you’re likely to execute the campaign programmatically across multiple publishers, choosing an audience and the right mix of relevant content, or you might use one of the rapidly growing self-serve advertising solutions offered by the likes of Facebook and Google.
What really matters in today’s marketing world is campaign analytics; the ability to understand the success of your ads – how they resonate with real people, and the right audience.
What this tells us is that what we need to know is changing. We no longer care how many of our consumers have visited Google last month. Instead, we need to understand our consumers’ behaviors, actions, engagements, attitudes and perceptions in granular detail, to provide real guidance on strategy. This is why we need detailed audience-centric data to drive strategy along with campaign analytics to measure ROI. In all of these scenarios, there is absolutely no need to understand the data produced from meters.
The future lies in real people analytics.
Even back in 2008, before I founded GlobalWebIndex, I had absolutely no use for metered data to help drive strategy or planning. Today, I can imagine almost no real use cases. For this reason and for those outlined above, we are not investing in meters to supplement our data set.
Our vision for the future is different – It’s built firmly around the concept of active data at massive scale. Firstly, we are looking to scale our reach of data coverage in innovative new ways and our aggregated 100% opted in panel has already grown to 18 million people.
Our solutions are re-engineering market research as we know it – not only in the way you collect data, but in the consumer value exchange and how you recruit and scale an audience panel.
Our goal for this product is to scale real people coverage to 50 million and beyond and to put real people at the core of marketing. The power of our â€˜active and massive’ approach lies in the fact that it provides a foundation for real people analytics. This, coupling both the rich data of accurate profiling with consumer attitudes, needs, wants, perceptions and behaviors, is what now drives strategy that works and offers the scale to deliver incredibly rich and accurately measured campaigns and measurement across all types of content, from news to video to TV.
Based on our latest client wins and the incredible growth we’re witnessing, it’s clear that this is something the global marketing industry desperately needs.
Want to know more? Download ourresearch and methodology for a closer look at what’s driving audience-centric marketing.