Multiple devices - An exciting & unpredictable futurePublished: Monday, 5 September 2016 by Paul Siddall, Senior Pre-Sales Consultant
As more and more devices offer internet access in different ways, so our interactions with brands change. Gone are the days of one PC in a household which was the only way to access the internet. We are seeing more mobile device usage, not only for browsing a website and using native apps on a smartphone or tablet, but increasingly with wearable apps too. All of this makes for a very exciting and unpredictable future, with every improvement making our ability to keep in touch with our loved ones, and hopefully our lives as a whole, easier. I, like many, am waiting to see what the next big innovation will be and who will bring it to market.
As with all opportunities, this also raises challenges for a business. Customers are engaging with the business across multiple devices, both because they use multiple devices throughout the day and also because the lifecycle of most devices is shortening, especially those which are increasingly cheap to switch to which is driving ever-faster, and lower risk, upgrading. Even with customers engaging across multiple devices however, individuals still expect the business to know and understand them as a coherent, unambiguous whole - so how does a business approach this?
Tracking multiple sessions on a single browser or app
Across a single browser, first party permanent cookies have always been considered the standard for linking data across multiple sessions. I won't go into the differences between first and third party cookies, but broadly first party cookies have a higher trust factor and so are less likely to be blocked or deleted. Cookies are excellent for tracking multiple sessions on a single browser because they are standard in the web world, but also critically the user has complete control over them allowing the user to clear their cookies so the website 'forgets' them. A similar approach can be taken on native apps on mobile and wearable devices where user preferences (again something the user controls and can delete) can be used to provide persistence across sessions.
There are alternative approaches to linking multiple sessions on the same browser and even device, such as device fingerprints. Device fingerprints rely on scanning all system settings that are visible to the browser as well as other data such as IP address, then applying advanced algorithms to uniquely identify the device. However the concern with device fingerprinting lies in the fact that all control of the identification is outside of the user's control, which can cause issues around privacy.
Personally Identifiable Information (PII)
First party browser cookies are usually set with a computer generated unique identifier that is suitably random, meaning no two browsers will have the same identifier. This allows us to link multiple sessions to the same random identifier so we can understand multiple sessions in the same browser, but very importantly no PII is contained within the identifier or being captured at all. This is particularly important when using a third party to collect and maintain the data, as most vendors do not allow PII to be captured. Companies operating in the EU should also be considering the upcoming General Data Protection Regulation (GDPR), which applies much stricter controls to the way a company should manage PII.
True understanding across devices
In order to truly understand an individual as they use multiple devices you need to maintain the data at the individual level, and capture a value the user enters into the website which uniquely identifies them - typically an e-mail address or customer number used when logging in. Once you have that value you can directly relate the different first party cookies and understand that to be a single user.
What if multiple people use the same browser or app?!
It would feel like avoiding the most complex part of this topic if I did not talk about multiple users logging into the same browser or app - for example at an internet cafe or a home laptop. We’ve all heard the stories of people being shown adverts based on other people’s browsing with unfortunate consequences, such as birthday surprises being ruined or secrets uncovered. So how do we avoid that?
At the extreme a business could choose to do no personalization, therefore missing out on chances to better engage with their customers. This is not what customers want, but they do want a sensible balance. I don’t believe there is a perfect answer that fits all businesses, but there are sensible approaches and I’ll outline one of those. One of the best systems I’ve seen is based on confidence. This involves applying a confidence score about how confident we are that a specific user has done a particular action.
We’ll start off by maintaining all interactions at the most granular level and timestamping them to the millisecond as people browse the website. From there we can then also split the data by actions that occur when a user is logged in and those that occur when the user is not. Actions occurring when the user is logged in can be trusted to be that user with the top level of confidence, whereas those that occur when the user is not logged in have a lower degree of confidence. That lower degree of confidence can be based upon a number of items, such as:
- Have we seen multiple users logging in on the same browser or app?
- Does the action occur directly before logging in?
- Is this a typical interaction for this user?
At the extreme we could treat all actions outside the authenticated space as untrusted, but a better approach is to apply a confidence factor to the actions and then make decisions based on the confidence that the specific user has performed that action. For example, based on untrusted actions you may choose to send an e-mail with a standard generic offer, but the generic offer is the one that best fits what you believe the user would be interested in. By contrast, when dealing with trusted actions and identifications you may wish to drive e-mail campaigns with offers tailored closely to the individual.
Offline campaigns are a great way to connect with users, but are not always engaging the user at the best time. The best time to engage with the user is while they are actually on the website and their focus is not elsewhere. By utilizing real-time personalization we can take the same confidence approach, applying it to both our confidence that it is the expected user on the website and the offer we wish to select. For example we can choose to change which promotional banner is displayed first on the home page, but keep it generic. But we can also provide truly personalized offers direct to the individual once we are confident it is them, for example personalizing the first page after the user logs in to the website with a user specific offer.
By truly understanding the user across devices we can make sure we are not causing offer 'fatigue' by repeatedly showing the user the same offer they have ignored on other devices. At the bigger picture level, we can also deliver offers and communication relevant to each individual user, delivering on the long-touted promise of true one-to-one personalization, all whilst maintaining user privacy.