I remember as a child turning over a large flat rock and seeing an extraordinary number of denizens of the soil scattering in all directions. Earwigs, centipedes, woodlice and un-named but fast-moving creatures scurrying to escape the light. The first view of the DeviceAtlas Discover report, revealing the traffic to the DeviceAtlas website, had the same fascination; very much like peering under a rock to see what is hidden there.
We recently released DeviceAtlas Discover as a free tool to enable businesses to get an immediate view of their website traffic from a technology perspective: not focused on the visitor and what they were doing, but on what the visitor is using to access the website. After all, device intelligence is what we do. But why is this different to every other analytics tool out there? Existing web analytics solutions seem to conceal what is deemed unimportant for their customer. It is perfectly reasonable that they focus on the visitor; ultimately the visitor is the one with the money. However, much can be learnt by exploring a little deeper, because the device governs the user experience. This entails examining signals from the HTTP Headers, JS interactions, and the device hardware.
Fascinating; but is it important?
Imagine you work as a teller in a bank, and someone walks in wearing a motorcycle helmet; you’re going to at least be keeping an eye on them. Maybe they will be asked to remove their helmet; if they decline, they may be politely asked to leave. On the web, the same principle applies: if a visitor is concealing their identity in some way, this is a signal that they should probably be watched a little more closely; and possibly asked to pass an additional authentication check. DeviceAtlas identifies when this concealment is occurring, in real time, enabling appropriate action to be taken. Another scenario, less obvious, is when someone comes in with a modified appearance. Maybe they have grown or removed a beard; maybe they have dyed their hair. In these cases, there is no signal that they should perhaps be watched a little more closely.
However, on the web, modified or inconsistent signals from the visitor device or browser are identified by DeviceAtlas, and appropriate handling of the visitor can be applied. This is where DeviceAtlas adds value: it provides identification, characterization and verification of visitor devices. DeviceAtlas Discover is an easy way to try it out on web properties, through a simple tag inclusion. What this enables is real-time comparison of characteristics in three areas: HTTP Headers, JS, and hardware. This comprises passive measurement; in addition the response to stimuli can be measured, i.e. active measurement, which is a great deal harder for a bad actor to anticipate and counter with the appropriate response. The best bit: all of this is done without any PII.
Example insights
What brands make up the mobile phone visitors?
First view, examining authentic traffic only: the result is as one might expect. Samsung dominates, with Apple close behind, but other businesses could see a reversal of this, depending on the nature of their business. Potentially more interesting are the rest of the top five: Redmi, Vivo and Oppo amounting to a fifth of mobile phone traffic visits to deviceatlas.com:
Where this view becomes more interesting is when all traffic is included, i.e. not just the authentic devices: Google jumps to the top of the table, having not even been in the top ten for authentic mobile phone visits:
Highlighting the red exclamation mark illustrates what is happening here: this is essentially all bot traffic. It is returned as bot here because it is self-declared as bot traffic:
More surprising, perhaps, is the level of traffic representing itself as Samsung devices; almost a quarter of all Samsung mobile phone traffic is non-authentic, i.e. it is making a claim to be a device which conflicts with the hardware profile and/or behavior:
To dig into this further, examining the non-authentic traffic highlights the claimed devices: for some reason, the iPhone 12 and the Galaxy S9 Plus are cover for a material proportion of falsified traffic:
So how is bot traffic broken down?
A quick glance at the breakdown reveals that, while bots make up 21% of traffic to deviceatlas.com, most of these are good wholesome all-American bots from Google, Apple and Microsoft. Huawei comes in at #4 (Petalbot), while Yandex commonly puts in an appearance:
The 82% made up by Googlebot mobile indicates how Google managed to jump to the top of the mobile phone brand list: the Nexus 5X traffic to the DeviceAtlas website was made up almost entirely of (self-declared) bots:
So, where does this take us?
What DeviceAtlas Discover helped DeviceAtlas discover was not just that web traffic is a rich and varied seam for analysis; it also points to ways we can protect and improve business results. Some practical learnings, from the perspective of a data business:
For these real-time interventions, a commercial license for DeviceAtlas is required; but there is no cost to setting up DeviceAtlas Discover, to understand the traffic profile of any web property and identify whether interventions would make sense to protect the business from clandestine and non-authentic visitors.
For more information and to set up DeviceAtlas Discover, visit https://deviceatlas.com/products/deviceatlas-discover.
