In my last article I wrote about how tracking has evolved within the performance marketing industry and some of the mistakes that have been made, and I outlined the problems that are inherent with “traditional” cookie based tracking.
Now, I would like to look to the future. Over the years, the affiliate industry has matured into a heavyweight of the performance marketing arena, and this has brought with it some interesting technical challenges to all involved.
At PHG, we had the luxury of a clean slate in 2010 – designing a completely new platform from the ground up, combining years of market experience with the advantage of new technologies at our disposal.
The term “big data” is one which I am beginning to hate more by the day. It is becoming far too all encompassing, much like “the cloud”, but the reality is that today there are some clients in the affiliate space who generate billions of human actions through their campaigns. Thankfully, some of these clients are using PHG technology to do so and we’re processing billions of clicks and hundreds of millions of conversions each month.
Performance marketing demands high availability of tracking; that is, a service that should be functional 24/7/365. Nobody, not even the likes of Google with their seemingly unlimited resources, has ever achieved 100% uptime (as an aside, there’s an interesting article regarding uptime issues) but you need to give yourself the best possible chance.
At scale, by definition, the numbers are bigger, therefore the chances for failures are bigger and even the smallest amount of downtime equates to a significant amount of business lost for all involved.
In terms of the tracking methodologies used at scale, it is prudent to attempt to eliminate as many points of potential failure as possible. In the ‘cookie’ tracking scenario, there are several – but one important step which can also affect the user experience is the requirement for an initial redirect from the publisher site through to the tracking provider’s domain.
If the tracking provider’s domain is not accessible then the user does not get redirected to the advertiser’s site, so the brand experience is tarnished and the potential sale is lost. Additionally, if the tracking provider’s servers are geographically located far away from the user then the time to redirect may become so cumbersome that the user loses interest.
This situation becomes further amplified if the user is accessing the web through a mobile connection such as 3G. This increased time taken to perform the redirect, even if located close to the tracking provider’s servers, can seriously hamper conversion rates; it’s another potential pitfall of the traditional tracking methods.
So what impact does this have on the methods of tracking that will be deployed in the future? My view is that many of the traditional methods mentioned in my previous article need to be thrown out of the window. But what alternatives are there? In my opinion, these are the four leading contenders.
1. First-party tracking
Something already offered by PHG, this form of tracking eliminates the need for any interaction between the consumer and the tracking provider through the sales journey.
All of the necessary tracking information is asynchronously passed through to our secure tracking API, and this method of data transfer is transparent to the user’s Web browser, thus removing many security concerns. There is nothing that a tracking provider gains from a redirect that cannot be easily replicated from the Advertiser’s own domain.
At PHG, we offer this as one of our recommended tracking alternatives, and we have seen how well this works in large-scale environments. The beauty of asynchronous transfer is that it does not hamper the user experience, and it can be performed in real-time or retrospectively – whatever works best for the client and its publishers.
Also, as the tracking methodology still employs cookies or sessions (albeit first party this time), privacy concerns are still adhered to. Arguably they are better regarded as the Advertiser has the direct relationship with its users and can inform them if -and how- they are tracked.
2. Device specific identifiers
We have already seen this occur on mobile, for example Apple removed the access to the device UDID (Unique Device Identifier) and instead offered their IDFA (Identifier for Advertisers). These are examples of a consistent identifier for the device, but they give the owner complete control over how it can be accessed.
It does not solve the whole problem though: tracking mobile Web to in-App is still an unresolved ‘grey’ area, with methods such as fingerprinting techniques being used to connect the dots.
This does not just stop at mobile devices. For most advertisers the desktop is still the majority stakeholder in their conversion traffic, and as browsers continue to evolve it’s only a matter of time until we see the release of identifiers specific to the user’s browser being released. Plans for Chrome have already begun and others are sure to follow.
3. Network injection
For a user to interact with a publisher’s site, there are a myriad of servers that the user will bounce between before their request is responded to. Increasingly I’m seeing more reports of DNS hijacking being implemented by the ISPs themselves, in an unscrupulous attempt to monetise their client base .
However, there could be legitimate uses of this technique to include a consistent identifier for all user requests around the Web, which would be ideal for tracking.
4. Centralised service
Again, we already see this today: web giants such as Amazon are able to achieve the “holy grail” of tracking their customers’ activity across multiple devices, because those customers are usually logged into their single Amazon account across all of those devices.
Also, Facebook takes this one step further: not only is their user login incredibly persistent, but they also have lots of “Like” buttons scattered around the Web that track your activity on every site that implements them. In the future, there could be specific cloud mechanisms that allow third parties to access user defined information across devices to determine who they are.
This is a fascinating area and whilst there are plenty more alternatives, I realise from my word count that I’m already rambling on. I predict that the debate and the speed at which new techniques will be devised will continue to increase and I’ll be sure to keep track of it all.
Happy to answer any questions in the comments section or please drop me a line if you want to geek out about any of this stuff.